Measuring the Efficiency of Islamic Banks in Indonesia and Malaysia ...



Ascarya, Diana Yumanita, Noer A. Achsani1, and Guruh S. Rokhimah Center for Central Banking Education and Studies, Bank Indonesia Jl. M.H. Thamrin 2, Sjafruddin Prawiranegara Tower, 20th fl., Jakarta 10110, Indonesia, Email: [email protected], [email protected], [email protected], [email protected]; Phone: 6221-3817345; Fax: 6221-3501912

This study will measure and compare the efficiency of Conventional and Islamic banks in Indonesia using parametric approach stochastic frontier approach (SFA) and distribution free approach (DFA), as well as nonparametric approach data envelopment analysis (DEA). These measurements will provide comprehensive and robust results of efficiency of individual bank compare to its peer group in every aspect considered. The results using parametric SFA show that Conventional and Islamic banks have been improving and converge to the highest level of efficiency. The DFA results show that conventional banks are only slightly more efficient than Islamic banks; conventional public bank is the most efficient, while Islamic regional bank is the least efficient. Meanwhile, the results using nonparametric DEA show that Islamic banks are slightly more efficient than conventional banks. However, they are improving and converging to a high level of efficiency. Moreover, income has become the most efficient, while labor is always inefficient in both banks. Deposit is improving in conventional banks, while it is worsening in Islamic banks. Financing has been a problem in conventional banks, while it is always high in Islamic banks. Therefore, Islamic banks should redirect their marketing and communication strategies to focus more on targeting floating customers, while the shortage in human resource should be given serious attention with short term and long term strategies. JEL Classification: C14, C33, G21, G28 Keywords: Islamic Banking, Efficiency, Stochastic Frontier Approach, Distribution Free Approach, Data Envelopment Analysis



Paper presented at University of Melbourne International Symposium and Conference on Islamic Banking and Finance: Ethics and Financial Practice in Global Perspective, University of Melbourne, Melbourne, Australia, November 19-20, 2008. 1 Visiting researcher from International Center for Applied Finance & Economics (InterCAFÉ), Department of Economics, Faculty of Economics and Management, Bogor AgricultureUniversity.

1.1

Background

Islamic banks have been in existence since early 1960s. The first Islamic bank established in 1963 as a pilot project in the form of rural savings bank in a small town of Egypt, Mit Ghamr. After that, Islamic banking movement came back to life in mid 1970s. The establishment of Islamic Development Bank in 1975 triggered the development of Islamic banks in many countries, such as Dubai Islamic Bank in Dubai (1975), Faisal Islamic Bank in Egypt and Sudan (1977), and Kuwait Finance House in Kuwait (1977). Joharris (2007) predicted that there are over 276 Islamic financial institutions (IFI) in the world, spread over 70 countries - sprawling from London, New York and Zurich to the Middle East, Africa and Asia with capitalization in excess of US$13 billion. These include banks, mutual funds, mortgage companies and takaful providers. The pool of money held by Muslim is predicted more than US$3.0 trillion. At present, there is an estimated US$1 trillion Islamic fund in the market. Moreover, global Islamic capital market is growing at 15% - 20% per annum, including deposits in Islamic banks which are estimated to be over US$560 billion. A large part of the banking and Takaful concentration is in Bahrain, Malaysia, and Sudan. A significant part of mutual funds concentrate in the Saudi Arabian and Malaysian markets in addition to the more advanced international capital markets. In Indonesia, Islamic financial institutions started to emerge in early 1980s with the establishment of Baitut Tamwil-Salman in Bandung dan Koperasi Ridho Gusti in Jakarta. The first Islamic Bank in Indonesia, Bank Muamalat Indonesia, established in 1992. The development of Islamic bank has been accelerated since Bank Indonesia (the central bank of Indonesia) allowed conventional banks to open Islamic branch. This Islamic branch can offer Islamic banking products and services separated from its conventional parent with its own infrastructure, including staff and branches. By June 2008, the Islamic banking system in Indonesia is represented by 3 Islamic banks, 28 Islamic branches, and 124 Islamic Rural Banks, with 743 offices and more than 1250 office channeling spread throughout the country. They offer comprehensive and wide range of Islamic financial products and services and cater 2.08% of the banking market share. It is expected that the Islamic banking industry in Indonesia would reached 5% of the banking market share in 2011. Despite these impressive achievements, Islamic banking in Indonesia has experiencing a slower growth in the past two years. There are many factors that could be attributed to this slower growth. One of these factors is the competitiveness of Islamic Banks within the banking system, since, under dual banking system, they have to compete head to head with conventional banks. One important aspect of competitiveness is efficiency. Inefficiency would become a great disadvantage to face a fierce competition in the banking industry. To win the competition, Islamic banks should know the strengths and weaknesses of themselves as well as of their competitor. Know yourself and know your competitor is a halfway to success. Therefore, analysis of the efficiency of Islamic banks in comparison with conventional banks is very important to provide a big picture of the strengths and weaknesses of Islamic banks and their competitors. 1

Despite of the importance, there are very limited studies comparing the efficiency of Islamic and conventional banks within a country using parametric and nonparametric approach, especially in Indonesia. Therefore, there should be a study that measure efficiency of Islamic and conventional banks using parametric and nonparametric approaches in Indonesia to provide comparison and to improve the robustness of previous measurements. These measures could also be used as a guide for Islamic banks to improve their weaknesses to be able to compete head to head with conventional banks and to achieve the intended goals to improve the market share. Moreover, the goal to strengthen Islamic banking structure could be achieved. 1.2

Objective

The objective of this study is to measure and compare the efficiency of conventional and Islamic banks in Indonesia using parametric approach stochastic frontier approach (SFA) and distribution free approach (DFA), as well as nonparametric approach data envelopment analysis (DEA). These measurements will provide comprehensive and robust results of efficiency of individual bank compare to its peer group in every aspect considered. 1.3

Scope of Study

Islamic banks included in this study are all full fledged Islamic banks and business unit Islamic banks in Indonesia, while conventional banks included in this study, to be comparable, are those with asset less one million US$ in real term. The measurement will compare the efficiency of conventional and Islamic banks in Indonesia using parametric approach (SFA and DFA) and nonparametric approach (DEA). 1.4

Data and Methodology

The time frame of this study is 2002 – 2006. The data used in this study are the data of published annual financial statements (balance sheets and income statements) of conventional and Islamic banks in Indonesia, with total asset less than one million US$ in real term. This study will apply stochastic frontier approach (SFA) and distribution free approach (DFA). SFA and DFA are two well known parametric approaches to measure efficiency using cross section or panel data of multiple inputs and outputs of business units. This study will also apply Data Envelopment Analysis (DEA). DEA is a non-parametric and non-deterministic method to measure relative efficiency of production frontier, based on empirical data of multiple inputs and multiple outputs of decision making units. The non parametric nature of DEA makes it does not need assumption of the production function. DEA will generate production function based on data observed. Therefore, misspecification can be minimized. DEA can be applied to analyze different kind of inputs and outputs without initially assigning weight. Moreover, the efficiency produced is a relative efficiency based on observed data. Preference of decision maker can also be accommodate in the model. 1.5 Benefit of the Study The results of this study will be very useful for many stakeholders of conventional and Islamic banks in Indonesia, especially the regulator (Bank Indonesia), to formulate appropriate policy recommendations to improve the synergy between conventional and Islamic banks in facilitating intermediation to the real sector.

2

Conventional and Islamic banks in Indonesia will also benefit from this study to see where they are in the competitiveness of the banking system. They will also be able to determine the potential improvements of weak aspects.

Banking efficiency has been a very important issue in a transition economy. All countries in transition have been encounter at least with one banking crisis, and many with more than one crisis (Jemrić and Vujčić, 2002). Banking efficiency is also an important issue in a developing open economy, since most of them have also been faced a banking crisis in the past. Malaysia and Indonesia are no exception. There are many studies about banking efficiency using parametric and non-parametric methods. Moreover, those studies are applied to conventional as well as Islamic banks. 2.1 Efficiency Measurement using Parametric Approach SFA and DFA have been used for some studies to measure the X-efficiency of commercial bank or other financial institutions such as studies that were conducted by Allen and Rai (1996), Yildirim and Philippatos (2003), Hadad et al. (2003), Saaid et al. (2003), Hussein (2004), Hassan (2003 and 2006). The firs tree studies measure the efficiency of conventional banks, while the last four studies measure the efficiency of Islamic banks. Allen and Rai (1996) measured operational efficiency in banking internationally during 1988-1992 using SFA and DFA, Yildirim and Philippatos (2003) measured the efficiency of banks in Europe during 1993-2000 using SFA to evaluate impact of transition economies to bank’s efficiency, while, Hadad et al. (2003) used SFA and DFA methods to measure the efficiency of banks in Indonesia during 1995-2003. Meanwhile, Saaid et al. (2003) measured Islamic banking xefficiency (technical and allocative efficiencies) in Sudan using SFA, Hussein (2004) compared the efficiency of Islamic and conventional banks in Bahrain during 19852001 using SFA (profit efficiency), Hassan (2003) measured the efficiency of Islamic banks in Pakistan, Iran, and Sudan during 1994-2001 using SFA (cost and profit efficiencies) and DEA (cost, allocative, technical, pure technical, and scale efficiencies), while Hassan (2006) measured the efficiency of Islamic banking industry in the world during 1995-2001 using SFA (cost and profit efficiencies) and DEA (cost, allocative, technical, pure technical, and scale efficiencies). [Insert Table 2.1] From those studies (read table 2.1 in the appendix), it can be summarized that most studies apply cost and profit efficiencies of SFA using translog or its derivatives functional form. The dependent variable is cost and profit, respectively, while the independent variables are input prices and output amounts. Off-balance sheet items are added as output to improve the efficiency estimates, especially for Islamic banking, since restricted investment accounts are not recorded in the balance sheet and considered as off-balance sheet items. 2.2 Efficiency Measurement using Non-parametric Approach DEA have been used extensively to measure banking efficiency. Some of those studies that measure efficiency of Islamic banks using DEA application are conducted 3

by Yudistira (2004), Sufian (2006), Ascarya and Yumanita (2006, 2007a, and 2007b), Mochtar et al. (2007), Zamil and Rahman (2007), and Bader et al. (2008). Yudistira measured the efficiency of 18 Islamic banks from various countries during 1997 – 2000 using intermediation approach, since intermediation is a fundamental principle of Islamic banking. Sufian measured the efficiency of Islamic window banks in Malaysia during 2001-2004 using intermediation approach with the same reason as that of Yudistira. Ascarya and Yumanita (2006) measured the efficiency of Islamic banks in Indonesia during 2002-2004 using intermediation and production approaches, since Islamic banking not only can be viewed as intermediary institution, but can also be viewed as a production entity. Their studies in 2007 compared the efficiency between Islamic bank in Malaysia and Indonesia, as well as between conventional and Islamic banks in Indonesia during 2002-2005 using intermediation approach. Mochtar et al. (2007) measured the efficiency of 22 Islamic banks (20 windows and 2 full-fledged) and 20 conventional banks in Malaysia during 19972003. Zamil and Rahman (2007) measured the efficiency of Islamic banks and conventional banks in Malaysia during 2000-2004 using intermediation approach. Meanwhile, Bader et al. (2008) compared 43 Islamic banks and 37 conventional banks, internationally, in 21 OIC member countries, during 1990-2005, using intermediation approach. Other studies of banking efficiency using DEA are done by Jemrić and Vujčić (2002) and Hadad et al. (2003). Jemrić and Vujčić measured efficiency of banks in Croatia during 1995-2000 using intermediation and production approach, since banking is not just functioned as intermediary, but also as a producer of loans and investments. Meanwhile, Hadad et al. measured efficiency of banks in Indonesia during 1995-2003 using asset approach to see the impact of merger and acquisition. The efficiency measurement, parametric or non-parametric, of financial institution like banks can be approached from their activities. There are three main approaches to explain the relationship between input and output of banks. Two approaches, namely, production (or operational) approach and intermediation approach, apply the classical microeconomic theory of the firm, while one approach, namely modern (or assets) approach applies modified classical theory of the firm by incorporating some specificities of banks’ activities, namely risk management and information processing, as well as some form of agency problems, which are crucial for explaining the role of financial intermediaries (Freixas and Rochet, 1998). The production approach describes banking activities as the production of services to depositors and borrowers using all available factors of production, such as labor and physical capital. The intermediation approach describes banking activities as intermediary in charge of transforming the money borrowed from depositors (surplus spending units) into the money lent to borrowers (deficit spending units). Meanwhile, the asset approach or the modern approach tries to improve the first two approaches by incorporating risk management, information processing, and agency problems into the classical theory of the firm. The summary of approaches applied by previous authors can be read in table 2.2 in the appendix. [Insert Table 2.2] From those studies it can be concluded that asset approach is an advanced approach that views bank not only has a classical function of intermediary, but also has other various new functions. Therefore, asset approach is not suitable to be applied to 4

Islamic banking which focuses on extending financing to the real sector. Production approach can be applied for Islamic banking, since this approach views Islamic bank as a general business unit. However, it becomes too general, so that the very essence of Islamic banking is not represented. Meanwhile, intermediation approach can be applied for Islamic banking since this approach views Islamic banking as an intermediary institution. However, the input and output variables should be selected carefully to really reflect the true essence of Islamic banking. Input and output variables selected by Sufian (2006) are the closest to the characteristics of Islamic banking. Some refined modifications might be needed to make it more representative.

The methodology of parametric Stochastic Frontier Analysis (SFA) and Distribution Free Analysis (DFA), as well as nonparametric Data Envelopment Analysis (DEA) will be used in this study. SFA, DFA and DEA applications are derived from the theory of efficiency. Therefore, this chapter will first discuss the theory of efficiency, the measurement of efficiency, the connection of SFA, DFA and DEA to efficiency theory, and then discuss their details. Moreover, bank’s efficiency can be measured from its functions. Three approaches to measure the efficiency of bank’s functions are intermediation approach, production approach, and modern or asset approach. The theory of efficiency in general, its relation to SFA, DFA and DEA, and the measurement of bank’s efficiency can be described in figure 3.1.

Producer Theory (Production Frontier Line)

Allocative Efficiency

Input – Output Concept

Nonparametric DEA

Efficiency Concept

Technical Efficiency

Tools Parametric SFA, TFA, DFA

Consumer Theory

Measurement Economic Efficiency

Theory

Minimum Input Maximum Output Concept

BANK EFFICIENCY

Production Approach

Constant Return to Scale Variable Return to Scale

Intermediation Approach Modern Approach

Figure 3.1 Theory of Efficiency 3.1 The Theory of Efficiency The concept of efficiency rooted from the microeconomic concept, namely, consumer theory and producer theory. Consumer theory tries to maximize utility or satisfaction 5

from individual point of views, while producer theory tries to maximize profit or minimize costs from producer point of views. In the producer theory, there is a production frontier line that describes the relationship between inputs and outputs of production process. This production frontier line represents the maximum output from the use of each input. It also represents the technology used by a business unit or industry. A business unit that operates on the production frontiers is technically efficient. Figure 3.2 shows the production frontier line.

Figure 3.2 Production Frontier Line Considered from economic theory, there are two different types of efficiency, namely technical efficiency and economic efficiency. Economic efficiency has macro economic point of view, while technical efficiency has micro economic point of view. The measurement of technical efficiency limited to technical and operational relationship in a conversion process of input to output. Whereas, in economic efficiency price can not be considered as given, since price can be influenced by macro policy (Sarjana, 1999). According to Farell (1957), efficiency comprises of two components, namely: a. Technical efficiency describes the ability of a business unit to maximize output given certain amount of input. b. Allocative efficiency describes the ability of a business unit to utilize inputs in optimal proportion based on their price. When the two types of efficiency combined, it will produce economic efficiency. A company is considered to be economically efficient if it can minimize the production costs to produce certain output within common technology level and market price level. Kumbhaker and Lovell (2000) argue that technical efficiency is only one of many components economic efficiency as a whole. Nevertheless, in order to achieve economic efficiency a company should produce maximum output with certain amount of input (technical efficiency) and produce output with the right combination within certain price level (allocative efficiency). 3.2 The Measurement of Efficiency In the past few years, performance measurement of financial institution has increasingly focused on frontier efficiency or X-efficiency (rather than scale efficiency), which measures deviation in performance of a financial institution from the best practices or costs-efficient frontier that depicts the lowest production costs for a given level of output. X-efficiency stems from technical efficiency, which gauges 6

the degree of friction and waste in the production processes, and allocative efficiency, which measures the levels of various inputs. Frontier efficiency is superior for most regulatory and other purposes to the standard financial ratios from accounting statements, such as, return on asset (ROA) or cost/revenue ratio, that are commonly employed by regulators, managers of financial institutions, or industrial consultants to assess financial performance. This is because frontier efficiency measures use programming or statistical techniques that removes the effects of differences in input prices and other exogenous market factors affecting the standard performance ratios in order to obtain better estimates of the underlying performance of the managers (Bauer, et al., 1998). Frontier efficiency has been used extensively in regulatory analysis to measure the effects of merger and acquisition, capital regulations, deregulation of deposit rates, removal of geographic restrictions on branching and holding company acquisitions, etc., on financial institution performance. Furthermore, Bauer et al. (1998) argue that the main advantage of frontier efficiency over other indicators of performance is that it is an objectively determined quantitative measure that removes the effects of market prices and other exogenous factors that influence observed performance. Tools to measure efficiency could be parametric and non-parametric. Parametric approach to measuring efficiency uses stochastic econometric and tries to eliminate the impact of disturbance to inefficiency. There are three parametric econometric approaches, namely: 1) Stochastic frontier approach (SFA); 2) Thick frontier approach (TFA); and 3) Distribution-free approach (DFA). These approaches differ in the assumptions they make regarding the shape of the efficient frontier, the treatment of random error, and the distributions assumed for inefficiencies and random error. The parametric methods have disadvantages relative to the non-parametric methods of having to impose more structure on the shape of the frontier by specifying a functional form for it. However, an advantage of the parametric methods is that they allow for random error, so these methods are less likely to misidentify measurement error, transitory differences in cost, or specification error for inefficiency (Bauer, et al., 1998). Meanwhile, non-parametric linear programming approach to measuring efficiency uses non-stochastic approach and tends to combine disturbance into inefficiency. This is built based on discovery and observation from the population and evaluates efficiency relative to other units observed. One of the non-parametric approaches, known as data envelopment analysis (DEA), is a mathematical programming technique that measures the efficiency of a decision making unit (DMU) relative to other similar DMUs with the simple restrictions that all DMUs lie on or below the efficiency frontier (Seiford and Thrall, 1990). The performance of a DMU is very relative to other DMUs, especially those that cause inefficiency. This approach can also determine how a DMU can improve its performance to become efficient. DEA was first introduced by Charnes, Cooper, and Rhodes in 1978. Since then its utilization and development have grown rapidly including many banking-related applications. The main advantage of DEA is that, unlike regression analysis, it does not require an a priori assumption about the analytical form of the production function so imposes very little structure on the shape of the efficient frontier. Instead, it constructs the best practice production function solely on the basis of observed data, and therefore the possibility of misspecification of the production technology is zero. On the other hand, the main disadvantage of DEA is that the frontier is sensitive to 7

extreme observations and measurement error (the basic assumption is that random errors do not exist and that all deviations from the frontier indicate inefficiency). Moreover, there exists a potential problem of “self identifier” and “near-selfidentifier”. 3.3 Parametric Approaches: SFA and DFA The parametric methods of SFA and DFA have been widely used to analyze efficiency of banking industry, especially in the US and other well-developed countries (see, among others, Berger, Hunter and Timme (1993), Berger and Humphrey (1997), Berger and Mester (1997) for an extensive review of literature on efficiency of financial institution). The two methods have also been used in the previous researches on the efficiency of banking industry in the transition countries. For more details see for example Yildirim and Philippatos (2003) for central and east european countries, Bhattacharya et al (1997) and Srivastava (1999) for India, Hasan and Marton (2000) for Hungary and Isik and Hassan (2002) for Turkey. 3.3.1 The Cost and Profit Frontiers Before going into the details about measuring efficiency, it is important to discuss the concept of cost and profit. For further readings, please refer to Berger and Mester (1997) and Yildirim and Philippatos (2003). Cost efficiency measure the performance of banking firm relative to the best-practice bank that produces the same output bundle under the same exogenous condition. The cost frontier is determined by estimating the following cost function: C  C  y , w, z , u , e 

where C measures total costs for bank, y is a vector of outputs, w is vector of input prices, z represents the quantities of fixed bank parameters, u is the inefficiency term that captures the difference between the efficient level of cost for given output levels and input prices and actual cost, and e is the random error term. Assuming the inefficiency and random error term are multiplicatively separable from the rest parameters, the above cost function can be expressed in logarithmic form as follows: ln C  f  y , w, z   ln u  ln e

After estimating a particular cost function, the cost efficiency for bank i is measured as the ratio between cost (Cmin) necessary to produce that bank’s output and the actual cost (Ci) : COSTEFFi 

C min exp f  y, w, z x exp ln u min  u min   Ci exp f  y, w, z x exp ln u i  ui

where umin is the minimum ui across all the samples. Profit efficiency measures how close a bank is attaining the possible profit as a bestpractice firm on the frontier for given levels of input and output prices and other exogenous variables. Previous studies offered two different approaches for calculating profit maximization objective, namely standard and alternative profit function. However, as the situation in Indonesia and Malaysia, Berger and Mester (1997) suggested that the alternative profit specification is preferred over the standard specification when (a) there are differences in the quality of banking services, (b) 8

markets are not perfectly competitive, (c) outputs are not completely variable and (d) output prices are not available. The alternative profit frontier is formulated as follows: P  P  y, w, z , u , e 

where P is the variable profits of the firm. Furthermore, in line with the formulation of cost function, the profit function can be expressed in the log terms as follows:

ln  P     f  y, w, z   ln e  ln u Where  is a constant added to every bank’s profit to make it positive, so that the natural log can be obtained. Profit efficiency is measured by the ratio between the actual profit of a bank and the maximum possible profit that is achieved by the most efficient bank. PROFEFF i 

Pi exp f  y, w, z x exp ln u i     Pmax exp f  y, w, z x exp ln u max   

Where umax is the maximum ui of all banks in the sample. 3.3.2 The SFA and DFA The Stochastic Frontier Approach (SFA) SFA asserts that managerial or controllable inefficiencies can only increase costs (reduce profits) above (below) best-practice frontier and that random fluctuations or uncontrollable factors can increase or reduce cost (profits). Therefore, the model assumes that inefficiency measures, (ln u), which represent the departure from efficient frontier follow an asymmetric half-normal distribution, while random fluctuations are distributed as two-sided normal with a zero mean and variance 2. The Distribution Free Approach (DFA) tries to avoid the arbitrary assumptions of the stochastic frontier approach, where panel data are available. This approach also separates the composite error term into inefficiency and statistical noise component. However, it assumes that there exists a core inefficiency for banks, which persists over time while the random error part vanishes out over time (Berger, 1993 in Yildirim and Philippatos, 2003). According to DFA, inefficiency estimate of a bank is determined by the difference between average residual of the bank i, (ln u), and the average residual of the bank on the frontier (ln umin), assuming that the random errors will cancel out over time. The estimated average residual is than transformed into a measure of efficiency : EFFi  expln u min  ln u i 

where ln umin is the average resiual for the bank with the lowest average cost residual. The most efficient bank will be given score 1, and then the others will get score between 0 – 1. 3.3.3 Functional Form Following Yildirim and Philippatos (2003), we employ the multiproduct translog functional form to estimate cost and alternative profit frontiers. The cost and profit frontier functions can be read in table 2.1 (no.2) in the appendix. We impose the regular restrictions of symmetry and linear homogeneity for input prices in estimating the parameters as follows : 9

 kj   jk , lh   hl ;

3

  l  1, l 1

3

 h  1 , h 1

3

 l 1

lk

1

Cost and input prices are normalized by the price of capital before taking logarithms to impose linear input price homogeneity. Cost, profit and output variables are normalized by equity capital (z) to control the heteroscedasticity, scale and other estimation biases in addition to providing a more economic meaning. 3.4 Non-parametric Approach: DEA Data envelopment analysis or DEA is a methodology for analyzing the relative efficiency and managerial performance of productive or decision making units (DMUs), having the same multiple inputs and multiple outputs. DEA allows us to compare the relative efficiency of (Islamic or conventional) banks by determining the efficient banks as benchmarks and by measuring the inefficiencies in input combinations (slack variables) in other banks relative to the benchmark (Jemrić and Vujčić, 2002). DEA provides an alternative approach to regression analysis. While regression analysis relies on central tendencies, DEA is based on extremal observations. While the regression approach assumes that a single estimated regression equation applies to each observation vector, DEA analysis each vector (DMU) separately, producing individual efficiency measures relative to the entire set under evaluation (Jemrić and Vujčić, 2002). DEA is a non-parametric, deterministic methodology for determining the relative efficient production frontier, based on the empirical data on chosen inputs and outputs of a number of DMUs. 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 inefficiencies of other, interior points (relatively inefficient DMUs) that are below the efficient frontier (Jemrić and Vujčić, 2002). Besides producing efficiency value for each DMU, DEA also determines DMUs that are used as reference for other inefficient DMUs. p

Efficiency  of  DMU 0 

 k 1 m

yk 0

v x i 1

DMU = decision making unit m : different inputs p : different outputs

k

i

i0

n : number of DMU evaluated xij : number of input i consumed by DMUj ykj : number of output k produced by DMUj

There are two DEA models that are most frequently used, namely, the CCR model (Charnes, Cooper, and Rhodes, 1978) and the BCC model (Banker, Charnes, and Cooper, 1984). The main difference between these two models is the treatment of return to scale. The CCR assumes that each DMU operates with constant return to scale, while the BCC assumes that each DMU can operate with variable return to scale. CCR model assumes that the ratio of additional input and output is equal (constant return to scale). It means that an additional input of x times will produce additional output of x times. Another assumption is that every DMU operates on an optimal scale. Therefore the efficiency of DMU can be measured as a maximum of a ratio weighted outputs to weighted inputs. Meanwhile, BCC model assumes that every DMU has not (or not yet) operated on optimal scale. This model assumes that the ratio of additional input and output is not equal (variable return to scale). It means 10

that an additional input of x times will not produce additional output of exactly x times, but it can be less or greater than x times. Generally, the efficiency score of CCR model for each DMU will not exceed the efficiency score of BCC model. This is because BCC model analysis each DMU “locally” (i.e. compared to the subset of DMUs that operate in the same region of return to scale) rather than “globally (Jemrić and Vujčić, 2002). Furthermore, a business or DMU, like bank, has similar characteristics one to another. However, each bank usually varies in size and production level. This indicates that size matters in relative efficiency measurement. CCR model represents (the multiplication of) pure technical and scale efficiencies, while BCC model represents pure technical efficiency only. Therefore, the relative scale efficiency is a ratio of CCR model and BCC model. S k  q k ,CCR / q k , BCC

If the value of S = 1 means that the DMU operates in the best relative scale efficiency, or in optimal size. If the value of S is less than 1 means that there still exists scale inefficiency of the DMU. Therefore, the value of (1-S) represents the level of inefficiency of the DMU. Consequently, when a DMU is efficient under BCC model, but inefficient under CCR model, this means that the DMU has scale inefficiency. This is because the DMU is (pure) technically efficient, so that the inefficiency that exists comes from the scale. OE  TE  SE

-->

SE  OE / TE

OE: overall technical efficiency of CCR Model; TE: pure technical efficiency of BCC Model; SE: scale efficiency.

4.1 Data Description The data needed for this empirical analysis comes from financial statements of conventional and Islamic banks in Indonesia in the period of 2002 – 2006. There are six types of conventional banks, namely public bank listed on capital market, conventional domestic foreign exchange bank, conventional domestic bank, conventional regional bank, conventional mixed bank owned by domestic and foreign investors, and conventional foreign bank owned by foreigner. While Islamic banks in Indonesia are of three types, namely, full-fledged Islamic bank, conventional bank that have separate Islamic branch or Islamic business unit, and Islamic Regional Development Branches. The data of type and number of banks included in this study can be read in table 4.1. [Insert Table 4.1] This study will adopt a modification of intermediation approach to better reflect Islamic bank activities, as also adopted by Sufian (2006), Ascarya and Yumanita (2007a and 2007b). Accordingly, we assume that conventional and Islamic banks produce Total Loan/financing (y1) and Income (y2) by employing Total Deposit (x1), Labor (x2), and Fixed Asset (x3). Liquid assets are not included in this study as output variable, since banks are naturally not in the business of financial instruments in the 11

financial markets, but in the business of providing financing to the real sector. As data on the number of employees are not readily made available, we use personnel expenses as a proxy measure. The aggregate series of inputs and outputs of conventional and Islamic banks included in this study can be read in table 4.2. [Insert Table 4.2] 4.2 Results and Analysis The efficiency of conventional and Islamic banks in Indonesia is measured in several ways by applying parametric SFA and DFA, as well as non-parametric DEA methods. To make a comparable measurement, conventional and Islamic banks are pooled together annually to form a common frontier. All banks for each year (2002-2006) are pooled to measure efficiency. 4.2.1 Parametric SFA and DFA The overall results of SFA and DFA measurements can be read in table 4.3 in the appendix. SFA annual results show that since 2004 conventional and Islamic banks have reached the highest efficiency of 1.0. Meanwhile, DFA aggregate results show that the average efficiency of conventional bank (0.89) is slightly better than that of Islamic bank (0.87). [Insert Table 4.3] Parametric SFA method requires time series data, so that the results can be grouped for each year of observation. The summary can be seen in figure 4.1.

Figure 4.1 Efficiency of Conventional and Islamic Banks in Indonesia using SFA The results in figure 4.1 show that in 2002, the average efficiency of conventional banks (0.79) is slightly better than that of Islamic banks (0.77). But in 2003, the efficiency of Islamic banks has improved to 0.84, while the efficiency of conventional banks has worsened to 0.76. In 2004, average efficiencies of both banks have improved to the top level of 1.00, and have not been changed until 2006. Figure 4.2 shows bank efficiency of each group using SFA method. In 2002, regional conventional bank was the most efficient (0.83) among conventional banks in Indonesia, while regional Islamic bank was the most efficient (0.86) among Islamic banks. In 2003, public conventional bank was the most efficient (0.78) among 12

worsened conventional banks, while regional Islamic bank was still the most efficient (0.90) among improved Islamic banks.

Note: Public: conventional public bank listed on capital market; Domestic fx: conventional domestic foreign exchange bank; Domestic: conventional domestic bank; Regional C: conventional regional bank; Mixed: conventional bank owned by domestic and foreign investors; Foreign: conventional foreign owned bank; Islamic: average Islamic bank; Full Fledged: Islamic full fledged bank; Full Branch: Islamic full branch bank; Regional I: Regional Islamic bank.

Figure 4.2 Group Efficiency of Conventional and Islamic Banks in Indonesia using SFA Meanwhile, parametric DFA method requires panel data with minimum complete 5 period series. Therefore, only 52 conventional banks and 7 Islamic banks can be included in the sample. The results show that conventional banks have exhibited slightly better efficiency (0.89) than that of Islamic banks (0.87) in the period of observation (2002-2006).

Note: Conventional: average conventional bank; Public: conventional public bank listed on capital market; Domestic fx: conventional domestic foreign exchange bank; Domestic: conventional domestic bank; Regional C: conventional regional bank; Mixed: conventional bank owned by domestic and foreign investors; Foreign: conventional foreign owned bank; Islamic: average Islamic bank; Full Fledged: Islamic full fledged bank; Full Branch: Islamic full branch bank; Regional I: Regional Islamic bank.

Figure 4.3 Group Efficiency of Conventional and Islamic Banks in Indonesia using DFA Figure 4.3 shows bank efficiency of each group using DFA method. The figure shows that conventional public bank is the most efficient (0.93), while regional Islamic full branch is the least efficient (0.84). Overall, conventional banks are slightly more efficient than Islamic banks. Comparing parametric measures of banks efficiencies and their respective traditional OCOI (operating costs divided by operating expenses) in figure 4.4 shows that efficient conventional banks do not always have low OCOI ratio. Moreover, banks that have better (lower) OCOI usually have better profitability (ROA), as can be read in table 4.4 in the appendix. Moreover, Islamic full branch is most DEA efficient which also have low OCOI and high ROA. 13


Figure 4.4 Efficiency of Conventional and Islamic Banks in Indonesia using DFA and SFA 2006 vs. OCOI 2006 Ratio [Insert Table 4.4] 4.2.2 Non-parametric DEA Summary of DEA results can be read in table 4.5 in the appendix and figure 4.5 below. The results suggest that overall efficiencies of conventional bank have exhibited continuous improvement, except in 2004, and have reached the highest mean of 0.85 in 2006. The decomposition of overall efficiency into its technical and scale efficiency components suggest that technical efficiency has been improving significantly from 0.67 in 2004 to 0.88 in 2006, while scale efficiency has always been high and stable to reach 0.96 in 2006. These imply that during the period of study, conventional banks in Indonesia have been operating at efficient scale; while technically have been operating at higher and higher efficiency. [Insert Table 4.5] Meanwhile, the overall efficiencies of Islamic bank have also been improving, except in 2004, and have reached the highest mean of 0.88 in 2006. The efficiency decomposition suggests that scale efficiency has been improving significantly in the last two years from 0.85 in 2004 to 0.99 in 2006, while technical efficiency, after 3 years trend of improvement, has been stagnant since 2005 at its highest level of 0.89. These show that during the period of study, Islamic banks in Indonesia have been operating at high and improving level of efficiency, except between 2003-2004 where Islamic banks were experiencing aggressive expansion (read figure 4.5, right).

14

Figure 4.5 Efficiency of Conventional and Islamic Banks in Indonesia using DEA Overall, it can be concluded that during the period of observation from 2002 to 2006 conventional and Indonesian Islamic banks in Indonesia have showed continues improvement, except in 2004, and have converged to a comparable high level of efficiency. Conventional banks mostly improved in technical efficiency, while Islamic banks improved in scale and technical efficiencies. In sum, overall efficiency of Islamic bank is only slightly better than conventional bank.


Figure 4.6 Group Efficiency of Conventional and Islamic Banks in Indonesia using DEA Figure 4.6 and table 4.6 in the appendix show bank efficiency of each group using DEA method. During 2002-2006, regional conventional bank has become the most improved conventional bank, while regional Islamic bank has become the most improved Islamic bank. Foreign conventional bank has become the most efficient in 2006 (1.00). Moreover, Islamic full fledged bank has almost always become the most efficient Islamic bank. [Insert Table 4.6] Moreover, the scale efficiency of Islamic banks can also be viewed from the trend of the return to scale (RTS)2 measured by DEA. Scale efficient banks exhibit constant return to scale (CRS). Banks experiencing economies of scale exhibit increasing return to scale (IRS), which means that the bank operates at a wrong scale of operation. Banks experiencing diseconomies of scale exhibit decreasing return to scale (DRS). Figure 4.7 and table 4.7 show the results of return to scale. [Insert Table 4.7] The number of conventional banks operating at efficient scale (CRS) has been increasing since 2004 from 8% to 36%, after a decrease during 2003-2004. Conventional banks experiencing economies of scale (IRS) have been increasing 2

RTS are the increase in output that results from increasing all inputs. There are three possible cases. (1) Constant Returns to Scale or CRS (RTS=0), which arise when percentage change in outputs = percentage change in inputs; (2) Decreasing Returns to Scale or DRS (RTS=-1), which occur when percentage change in outputs < percentage change in inputs; (3) Increasing Returns to Scale or IRS (RTS=1), which occurs when percentage change in outputs > percentage change in inputs.

15

during the period of study, except in 2004. Moreover, conventional banks experiencing diseconomies of scale (DRS) have been decreasing sharply during the period of study, except in 2004 (read figure 4.7, left).

Figure 4.7 Return to Scale of Conventional and Islamic Banks in Indonesia Meanwhile, Islamic banks operating at efficient scale (CRS) have been decreasing in 2002-2004 from 43% to 8%, and have been sharply increasing in 2004-2006 from 8% to 63%. Islamic banks experiencing diseconomies of scale (DRS) have been increasing in 2002-2004 from 43% to 77%, and have been sharply decreasing in 2004-2006 from 77% to 26%. Moreover, Islamic banks experiencing economies of scale (IRS) have been slightly decreasing during the period of study from 14% to 11% (read figure 4.7, right). Aggressive expansion of Islamic banking industry in Indonesia during 2003-2004 has been reflected in the increase of DRS and the decrease of CRS Islamic banks. Overall, it can be concluded that conventional and Islamic banks in Indonesia have been experiencing improvement in scale efficiency, especially Islamic banks. At the end of 2006, there are more Islamic banks operating at scale efficient (CRS), while there are more conventional banks operating at economies of scale (IRS). Other than generating efficient frontier, one salient feature of DEA is that it can generate set of references for inefficient DMUs (banks) to benchmark to. Table 4.8 shows conventional and Islamic banks that are referenced by other inefficient banks in 2002-2006. In the first three years of observation, there are always more conventional banks on efficient frontiers that set as benchmarks for other inefficient banks to make improvements. Conversely, in the last two years of study, there are always more Islamic banks benchmarked by other inefficient banks. [Insert Table 4.8] Another useful feature of DEA is that it can identify the source of inefficiency for each DMUs or Islamic banks. The results in table 4.9 and figure 4.8 are generated by running each year data of each country, separately. [Insert Table 4.9] In 2002-2006, conventional banks (read figure 4.8, left) have always been efficient in generating income, since they have been able to provide sophisticated diverse banking services that are demanded by general customers, such as e-banking, internet-banking, phone-banking, sms-banking, and so on. Conversely, labor has always been inefficient part of conventional banks. This could be attributed to the nature of service 16

industry where the most important capital is skilled and experienced human capital. Moreover, deposit has been improving, while loan extension or financing has been worsening. These can be attributed to high policy rate (the rate of Bank Indonesia Certificate or SBI) which is higher than deposit rate that makes conventional banks race to accumulate deposit and put the excess liquidity into SBI which serves like investment instrument with higher rate than deposit rate. No wonder that LDR has always been low and SBI outstanding has always been increasing. Conventional banks do not have to extend loan to make easy profit.

Figure 4.8 Potential Improvements for Conventional and Islamic Banks in Indonesia Meanwhile, financing has always been the most efficient elements in Indonesian Islamic banks, except in 2004 and 2006 (read figure 4.8, right), as also showed in the figure of FDR that have always been high around 100%. In early 2004, there was a temporary shift of deposits from conventional banks to Islamic banks due to the release of fatwa of the haram-ness of interest by DSN-MUI (National Shariah Advisory Council of Indonesia) in late 2003. Income has been improving considerably from the most inefficient element in 2002 to the most efficient element in 2006, due to the improvements of Islamic banks in providing financial services comparable to those provided by conventional banks. Deposits have been worsening, and have become the most inefficient element in 2006. The problem of deposits in Indonesian Islamic banking intensified due to its high growth, but limited customer base. Islamic banks have been targeting faithful customers which counted for only 110 percent, while they have not been focusing on floating customers which counted for 80 percent. Islamic banks should shift their marketing and communication strategies to target more on floating customers. Moreover, the problem of deposit is also due to the high interest rate mentioned before, so that floating customers opt to put their money into conventional bank. Meanwhile, labor has always been inefficient part of Islamic bank. This is the case in Indonesia, and most countries adopting dual banking system, where the supply of human resource is always lagging behind the demand of this still fast growing industry. Even though there are more and more universities and higher educational institutions offering Islamic Economic and Finance, the number of graduates are still could not catch up with the demand. The consequence of this is either the wage goes up or/and the human resource quality goes down. Therefore, Indonesian Islamic banks should give more attention to human resource to improve their efficiency. Moreover, other elements of input can also be improved further in less priority than human resource. 17

Note: Domestic fx: conventional domestic foreign exchange bank; Domestic: conventional domestic bank; Regional C: conventional regional bank; Mixed: conventional bank owned by domestic and foreign investors; Foreign: conventional foreign owned bank; Islamic: average Islamic bank; Full Fledged: Islamic full fledged bank; Full Branch: Islamic full branch bank; Regional I: Regional Islamic bank.

Figure 4.9 Efficiency of Conventional and Islamic Banks in Indonesia using DEA vs. OCOI Ratio Figure 4.9 and table 4.10 show DEA results compare to their respective traditional OCOI (operating costs divided by operating expenses) and ROA (return on assets). It can be shown that efficient conventional banks, like regional and foreign banks, always have lower OCOI ratios, while efficient Islamic banks do not always have lower OCOI ratios. Moreover, conventional banks that have better (lower) OCOI usually have better profitability (ROA), while Islamic banks that have better OCOI do not always more profitable (better ROA). [Insert Table 4.10]


Figure 4.10 Efficiency of Conventional and Islamic Banks in Indonesia using DFA, SFA 2006, and DEA 2006 Figure 4.10 and table 4.11 show the comparison of parametric and non-parametric measures of efficiencies. It can be concluded that different approaches have given somewhat different results. DFA gives average efficiency for the observation period, where conventional banks are slightly more efficient than Islamic banks. SFA gives annual efficiency for each year of observation, where conventional and Islamic banks converge to highest efficiency since 2004. Meanwhile, DEA gives annual efficiency for each year of observation with the breakdown of overall efficiency into technical 18

and scale efficiencies. Islamic banks are slightly more efficient than conventional banks. In general, SFA and DEA measures of efficiency provide similar trend of improving efficiency, especially since 2004. However, DEA measure gives more detailed results for deeper analysis. [Insert Table 4.11]

5.1 Conclusions 

Overall, conventional banks exhibit improving and convergence measure of high efficiency to those of Islamic banks using parametric SFA and DFA methods. However, using non-parametric DEA method, Islamic banks are relatively slightly more efficient than conventional banks in all three measures (technical, scale, and overall efficiencies) during the period of study. This can be attributed, among others, to efficient financing activities. Financing to deposit ratios has always been high around 100 percent, reflecting high contribution of Indonesian Islamic banking to the real sector.



Conventional and Islamic banks show a convergence in the characteristics of inputs and outputs, where income has become the most efficient element, while labor or human resource should be given top priority for improvements. Income from banking services come from sophisticated diverse banking services provided by conventional and Islamic banks, such as e-banking, internetbanking, phone-banking, sms-banking, and so on. Conversely, labor has always been inefficient part of conventional and Islamic banks. This could be attributed to the nature of service industry where the most important capital is skilled and experienced human capital. Moreover in Islamic banking, the supply of human resource is always lagging behind the demand of this still fast growing industry.



The differences between conventional banks and Islamic banks are shown in deposit and financing. Deposit is improving in conventional bank, while it is worsening in Islamic bank, due to high interest rate and competition. High and increasing SBI rate which is higher than deposit rate makes deposits flow to conventional banks. Meanwhile, conventional banks always have low loan to deposit ratio (LDR), while Islamic banks always have high financing to deposit ratio (FDR). Moreover, conventional banks can put their excess liquidity into SBI and do not have to extend loan to make profit, while Islamic banks have to extend financing to make profit.

5.2 Recommendations 

There is an anomaly in interest rate structure, where policy rate (BI rate) is higher than deposit rate, which makes banks aggressively collect deposits without worry about extending loan, since they can make easy money just by placing it in SBI (certificate of Bank Indonesia). Intermediation to the real sector will be affected. One alternative way to solve this problem is by selling SBI retail, so that it will compete with deposits and will force deposits rate to increase above SBI rate, and the structure of interest rate will back to normal.

19



Islamic banks in Indonesia are still young and small, so that socialization and expansion should be the number one priority to make Islamic bank familiar to public and to reach economies of scale in the shortest time possible. Other than organic expansion that naturally slow, to accelerate expansion in Indonesia the government should also have the political will, commitment, and courage to expand inorganically by converting one state owned conventional bank into Islamic bank, preferably the one that have large networks.



Funding (deposits) has become a new problem in Indonesian Islamic banking. Head to head competition with conventional banks should force Islamic banks to redirect their marketing and communication strategies to focus more on floating customers and strengthen their bases of loyal customers.



Human resource has always been a problem in Islamic banks in Indonesia, specifically due to high demand and short supply. The improvement of the human resources could be done with two strategies, namely, short term and long term. In the short term, education and training should be conducted for every level of management. In the long term, special fields of study in Islamic economic and finance should be opened in graduate and undergraduate levels, as well as inserting Islamic economic and finance curriculum in high school.



The improvement of the human resources from the regulator side could be done by requiring banks to spend minimum budget for human resources development. Moreover, the government or regulator could give incentives by financing participation in human resources development. The regulator could also provide free training for Islamic bank officers.

Allen, L. and A. Rai. “Operational Efficiency in Banking: An International Comparison,” Journal of Banking and Finance,” 20, 655-72 (1996). Ascarya and D. Yumanita. “The Competitiveness of Islamic Banks within Indonesian Dual Banking System”, Paper, USIM Islamic Economics Conference (IECONS 2007): “Comprehensive and Balanced Development among OIC Countries: Cooperation, Opportunities, and Challanges”, Kuala Lumpur, 1719 July (2007b). Ascarya and D. Yumanita. “Comparing the Efficiency of Islamic Banks in Malaysia and Indonesia”, Paper, IIUM International Conference on Islamic Banking and Finance (IICiBF); “Research and Development: The Bridges between Ideals and Realities”, IIUM, Kuala Lumpur, 23-25 April (2007a). Ascarya and D. Yumanita. “Analisis Efisiensi Perbankan Syariah di Indonesia dengan Data Envelopment Analysis”, TAZKIA Islamic Finance and Business Review, Vol.1, No.2 (2006). Bader, M.K.I., S. Mohamad, M. Ariff, and T. Hassan. “Cost, Revenue, and Profit Efficiency of Islamic versus Conventional Banks: International Evidence using Data Envelopment Analysis,” Islamic Economic Studies, Vol.15, No.2, January (2008).

20

Banker, R.D., A. Charnes, and W.W. Cooper. “Some Models for Estimating Technical and Scale Inefficiencies in Data Envelopment Analysis,” Management Science 30 (9), 1078-92 (1984). Bauer, P.W., A.N. Berger, G.D. Ferrier, and D.B. Humphrey. “Consistency Conditions for Regulatory Analysis of Financial Institutions: A Comparison of Frontier Efficiency Methods”, Federal Reserve, Financial Services Working Paper, 02/97 (1997). Berger, A.N., W.C. Hunter, and S.G. Timme. “The Efficiency of Financial Institutions: a Review and Preview of Research Past, Present, and Future,” Journal of Banking and Finance, 17(2-3),221-50 (1993). Berger, A.N., D. Hancock, and D.B. Humphrey. “Bank efficiency derived from the profit function,” Journal of Banking and Finance, 17, 317-47 (1993). Berger, A.N. and R. DeYoung. “Problem Loans and Cost Efficiency in Commercial Banks,” Journal of Banking and Finance, 21, 1-29 (1997). Berger, A.N. and D.B. Humphrey. “Efficiency of financial institutions: International survey and directions for future research,” European Journal of Operational Research, 98, 175-212 (1997). Berger, A.N. and L.J. Mester. “Inside the black box: What explains differences in the efficiencies of financial institutions?,” Journal of Banking and Finance, 21, 895-947 (1997). Bhattacharya, A., C.A.K. Lovell, and P. Sahay. “The impact of liberalization on the productive efficiency of Indian commercial banks,” European Journal of Operations Research, 98, 332-45 (1997). Charnes, A., W.W. Cooper, and E. Rhodes. “Measuring the efficiency of decision making units,” European Journal of Operational Research, 2, 429-44 (1978). Coelli, T., D.S.P. Rao, and G.E. Battese. An Introduction to Efficiency and Productivity Analysis, Kluwer Academic Publishers, 1998. Denizer, C.A., et al. Measuring Banking Efficiency in the Pre and Post Liberalization Environment: Evidence from the Turkish Banking System, World Bank, 2000. Farrell, M.J. “The Measurement of Productive Efficiency,” Journal of The Royal Statistical Society, 120, 253-81 (1957). Freixas, X. and J.C. Rochet. Microeconomics of Banking, The MIT Press, Cambridge, Massachusetts, London, England, 1998. Hadad, M.D., et al. “Analisis Efisiensi Industri Perbankan Indonesia: Penggunaan Metode Nonparametrik Data Envelopment Analysis (DEA)”, Biro Stabilitas Sistem Keuangan Bank Indonesia, Research Paper, no. 7/5, (2003). Hadad, M. D., et al. “Pendekatan Parametrik Untuk Efisiensi Perbankan Indonesia”, Research Paper, no. 4/5, Biro Stabilitas Sistem Keuangan Bank Indonesia, (2003).

21

Hasan, I. and K. Marton. “Development and Efficiency of the Banking Sector in a Transitional Economy: A Hungarian Experience,” Journal of Banking and Finance 27 (12), 2249-2271 (2003). Hasan, Z. “Evaluation of Islamic Banking Performance: On the Current Use of Econometric Models,” in Iqbal, Munawar, Salman Syed Ali and Dadang Muljawan (eds). Advances in Islamic Economics and Finance, Vol.1, IRTI, Islamic Development Bank, Jeddah, Saudi Arabia, 2007. Hassan, M.K. “Cost, Profit and X-Efficiency of Islamic Banks in Pakistan, Iran and Sudan,” Paper, International Conference on Islamic Banking: Risk Management, Regulation and Supervision, Bank Indonesia – Islamic Research and Training Institute, Jakarta, 30 September – 2 October (2003). Hassan, M.K. “The X-Efficiency in Islamic Banks,” Islamic Economic Studies, Vol.13, No.2, February (2006). Isik, I. and M.K. Hassan. “Cost and Profit Efficiency of the Turkish Banking Industry: An Empirical Investigation,” The Financial Review, 37(2), May,: 257-80 (2002). Jemrić, I. and B. Vujčić. “Efficiency of Banks in Croatia: A DEA Approach,” Croatian National Bank Working Paper, 7 February (2002). Kumbhaker, S.C. and C.A.K. Lovell. Stochastic Frontier Analysis, Cambridge University Press, United Kingdom, 2004. Mokhtar, H.S.A., S.M. Alhabshi, and N. Abdullah. “Determinants of Islamic Banking Efficiency in Malaysia,” Paper, IIUM International Conference on Islamic Banking and Finance (IICiBF); “Research and Development: The Bridges between Ideals and Realities”, IIUM, Kuala Lumpur, 23-25 April (2007). Saaid, A.E., S.A. Rosly, M.H. Ibrahim, and N. Abdullah. “The X-Efficiency of the Sudanese Islamic Banks,” IIUM Journal of Economics and Management, 11(2), 123-141 (2003). Sufian, F. “The Efficiency of Islamic Banking Industry: A Non-parametric Analysis with Non-discretionary Input Variable,” Islamic Economic Studies, Vol.14, No.1&2, August (2006) & January (2007). Sufian, F. “The Efficiency of Islamic Banking Industry in Malaysia: Foreign Versus Domestic Banks”, Paper, INCEIF Colloquium, Malaysia, April (2006). Yildirim, H.S. and G.C. Philippatos. “Efficiency of Banks: Recent Evidence from the Transition Economies of Europe 1993-2000,” Paper, School of Administrative Studies, York University, Toronto, ON Canada (2003). Yudistira, D. “Efficiency in Islamic Banking; An Empirical Analysis of Eighteen Banks”, Islamic Economic Studies, Vol.12, No.1, August (2004). Zamil, N.A.M. and A.R.A. Rahman. “Efficiency of Islamic and Conventional Commercial Banks in Malaysia: A Data Envelopment Analysis (DEA) Study,” Paper, IIUM International Conference on Islamic Banking and Finance (IICiBF); “Research and Development: The Bridges between Ideals and Realities”, IIUM, Kuala Lumpur, 23-25 April (2007). 22

Table 2.1 Summary of Parametric Approach Applied No

Author

Functional Form

Input

Output

1

Allen & Rai ‘96

Translog

Price of labor, price of capital, price of borrowed funds

2

Yildirim & Philippatos ’03

Multiproduct Translog

Price of borrowed funds, price of Loans, investment, deposit labor, price of physical capital

3

Hadad et al. ‘03

Translog

Price of labor, price of funds

Loans to the bank, loans to the stakeholder, securities

4

Saaid et al. ‘03

Translog

Price of labor, price of fixed capital, price of deposit

Investment asset

5

Hussein ’04

Fourier Flexible Price of labor, price of fund, price of physical capital

Financing; investment, off-BS items

6

Hassan ’03 & ‘06

Translog

Total loan, other earning assets, off-BS items

Price of labor, price of capital, price of fund

Loans, investment assets

Functional forms: No 1

Functional Form Cost Efficiency: 2 2 2 2 2 2 2 2 4  ln TC   o   i ( pi )    j ln( y j )  0.5    ik ( pi )( pk )  0.5    jh ln( y j ) ln( yh )     ij yi ln p j    l YRl  i 1 j 1 i 1 k 1 j 1 h 1 i 1 j 1 l 1 TC total operating and interest costs, pi price of inputs (p1 price of labor, p2 price of fixed capital, p3 price of borrowed funds), yi output amounts (y1 traditional banking assets [loan], y2 investment assets), YR dummy of time period, ε = v + u, v random error, u inefficiency

2

Cost Efficiency: 2

2

2

3

ln(C / w3 z )      ln( w1 / w3 )  0.5  lh ln( w1 / w3 ) ln( w1 / w3 )    k ln( y k / z ) l 1

l 1 h 1

3

k 1

3

3

2

0.5  kj ln( y k / z ) ln( y k / z )    lk ln( y k / z ) ln( w1 / w3 )  1 ln Z k 1 j 1

k 1 l 1

3

2

k 1

l 1

0.5 2 (ln Z ) 2   k ln( y k / z ) ln Z    l ln( wl / w3 ) ln Z  ln eti  ln u it

Profit Efficiency: ln(C/w3z) is replaced by ln(P/w3z) and the inefficiency term is –u. C total interest, personnel and other operational expenses, wi input prices, yi output amounts, z equity capital, e random error, u inefficiency

4

Cost Efficiency: 3

3

3

3

ln TC   0   ln w1  0.5  ij ln wi w j    ij ln wi   ln y  0.5 yy (ln y ) 2   i 1

i 1 j 1

i 1

TC total cost, wi price of inputs (w1 price of labor, w2 price of fixed capital, w3 price of deposit), y investment asset, ε = v + u, v random error, u inefficiency

5

Profit Efficiency: 23

3

3

3

3

3

3

ln    0    i ln Qi    j ln Pj  0.5  ik ln Qi ln Qk  0.5  jh ln Pj ln Ph l 1

j 1

3

3

i 1 k 1

j 1 h 1

3

   ij ln Qi ln Pj   [  i cos( Qi )   i sin( Qi )] i 1 j 1 3

i 1

3

  [ i cos( Qi  Q j )  ij sin( Qi  Q j )]  ln v a  ln u a i 1 j 1

π net profit, Qi output amounts (Q1 financing, Q2 investments, Q3 off-balance sheet items), Pi price of inputs (P1 price of labor, P2 price of fund, P3 price of physical capital), v random error, u inefficiency, ε = v + u

6

Cost Efficiency: 3 3 3 3 3 3 3 3  ln C b   o  i ln( yi )   k ln( pk )0.5   ij ln( yi ) ln( y j )0.5    lm ln( pl ) ln( pm )   ik ln yi ln pk  b i 1 k 1 i 1 j 1 l 1m1 i 1k 1 Profit Efficiency: 3 3 3 3 3 3 3 3 ln(  a )   o   i ln Yist   i ln Pist 0.5    ij ln Yist ln Y jst 0.5    kl ln Pkst ln Plst    ki ln Pkst ln Yist vst ust i 1 i 1 i 1 j 1 k 1l 1 k 1i 1 C variable cost, Yi output amounts (Y1 total loan Y2 other earning assets, Y3 off-balance sheet items), Pi price of input (P1 price of labor, P2 price of capital, P3 price of fund), π net profit, v random error, u inefficiency, ε = v + u

Table 2.2 Summary of Non-parametric Approach Applied Author

Input

Output

Intermediation Approach Jemrić & Vujčić’02

No. of Employees; Fixed Assets & Software; Total Deposits

Total Loans; Short-term Securities

Yudhistira’04

Staff Costs; Fixed Assets; Total Deposits

Total Loans; Other Income; Liquid Assets

Hassan ’03 & ‘06

Labor, Fixed Assets, Total Funds, and their respective prices

Total Loans, Other Earning Assets, Off-BS Items, and their respective prices

Ascarya & Yumanita’06

Staff Costs; Fixed Assets; Total Deposits

Total Loans; Other Income; Liquid Assets

Sufian’06

Labor Costs3; Fixed Assets; Total Deposits

Total Loans; Income

Zamil & Rahman‘07

Staff Cost; Capital (net book value of premises and fixed asset); Total Deposits & Loanable Funds

Loans and Advances; Income (total interest income, Non-interest Income and income from IBS)

Mochtar et.al‘07

Labor Costs; Total Deposits, Other Operating/Overhead Expenses

Total earning assets (financing/loans; dealing securities; investment securities; placements with other banks)

Ascarya & Labor Costs; Fixed Assets; Total Deposits Yumanita‘07a, ’07b

Total Loans; Income

Bader et al.’08

Total Loans, Other Earning Assets, Off-BS Items, and their respective prices

Labor, Fixed Assets, Total Funds, and their respective prices

Production Approach Jemrić & Vujčić’02

Interest & Related Costs; Commissions for Services & Related Costs; Labor Related

3

Interest & Related Revenues; Non-interest Revenues

As data on the number of employees are not readily made available, this study uses personnel expenses as a proxy measure.

24

Adm. Costs; Capital Related Adm. Costs Ascarya & Yumanita’06

Interest Costs; Staff Costs; Operational Costs

Interest Income; Other Operational Income

Asset Approach Hadad et.al’03.

Staff Costs to Total Assets; Interests Costs Financing to Connected Party; Financing to Total Assets; Other Costs to Total Assets to Other Party; Financial Papers

Table 4.1 Data of Conventional and Islamic Banks SFA and DFA Conventional

2002 52

2003 52

2004 52

2005 52

2006 52

Public Domestic+ Domestic Regional Mixed Foreign

3 14 10 17 2 6

3 14 10 17 2 6

3 14 10 17 2 6

3 14 10 17 2 6

3 14 10 17 2 6

Islamic

7

7

7

7

7

Domestic Full Fledged Domestic Full Branch Regional Full Branch

2 4 1

2 4 1

2 4 1

2 4 1

2 4 1

DEA Conventional Public Domestic FX Domestic Regional Mixed Foreign

Islamic Domestic Full Fledged Domestic Full Branch Regional Full Branch

2002

2003

2004

2005

2006

72 1 15 24 22 8 2 7 2 4 1

101

97 1 22 32 23 14 5 13 3 7 3

63 1 14 14 20 9 5 19 3 9 7

44 9 27 3 4 1 19 3 9 7

22 38 20 16 5 8 2 5 1

Table 4.2 DEA Inputs and Outputs Data (Real US$.000) 2002

2003

2004

2005

2006

15,697 241,318 12,813 22,534 9,481

8,330,315 1,949,255 7,174,381 253,859 19,945,804

8,950,342 1,782,812 7,419,603 262,873 20,392,106

6,086,346 982,839 4,831,932 129,032 14,204,088

9,829,870 2,097,861 10,941,882 386,983 17,972,278

347,468 51,847 110,371 8,580 433,713

580,334 83,175 541,520 12,427 830,556

1,041,176 140,256 940,023 19,084 1,400,265

1,093,134 141,101 885,359 20,173 1,395,608

1,485,325 255,105 1,284,758 32,909 2,024,293

Conventional Financing Income Deposit Labor Asset

Islamic Financing Income Deposit Labor Asset

25

Conventional : Islamic Financing Income Deposit Labor Asset

0.05 4.65 0.12 2.63 0.02

14.35 23.44 13.25 20.43 24.02

8.60 12.71 7.89 13.77 14.56

5.57 6.97 5.46 6.40 10.18

6.62 8.22 8.52 11.76 8.88

Table 4.3 Summary of Parametric SFA and DFA Efficiencies

Conventional

2002 0.790

2003 0.760

SFA 2004 1.000

Public Domestic fx Domestic Regional C Mixed Foreign

0.803 0.784 0.716 0.833 0.830 0.757

0.777 0.774 0.756 0.771 0.730 0.683

0.997 0.997 0.997 0.997 0.997 0.997

0.999 0.999 0.999 0.999 0.999 0.999

0.999 0.999 0.999 0.999 0.999 0.999

0.93 0.91 0.88 0.87 0.89 0.90

Islamic

0.770

0.840

1.000

1.000

1.000

0.87

Full Fledged Full Branch Regional I

0.625 0.825 0.860

0.805 0.838 0.900

0.997 0.997 0.997

0.999 0.999 0.999

0.999 0.999 0.999

0.86 0.88 0.84

BANK

2005 1.000

2006 1.000

DFA 0.89

Table 4.4 Summary of Parametric SFA and DFA vs. OCOI and ROA Ratios BANK Conventional

DFA 0.890

SFA 1.000

OCOI

ROA


0.929 0.914 0.881 0.867 0.890 0.896

0.999 0.999 0.999 0.999 0.999 0.999

0.875 0.875 0.971 0.729 0.733 0.719

1.560 1.648 0.832 3.760 2.685 3.355

Islamic

0.870

1.000

Full Fledged 0.859 0.999 0.871 1.655 Full Branch 0.882 0.999 0.698 1.085 Regional I 0.843 0.999 0.495 4.500 Note: OCOI = Operating Costs/Operating Income; ROA = Return on Assets

Table 4.5 Summary of Non-parametric DEA Efficiency Efficiency Measures Conventional Overall Efficiency Technical Efficiency Scale Efficiency

2002

2003

2004

2005

2006

0.45 0.58 0.82

0.72 0.76 0.95

0.62 0.67 0.94

0.69 0.73 0.95

0.85 0.88 0.96

0.70 0.74 0.94

0.79 0.86 0.92

0.69 0.80 0.85

0.81 0.89 0.91

0.88 0.89 0.99

Islamic Overall Efficiency Technical Efficiency Scale Efficiency

26

Table 4.6 Summary of Non-parametric DEA Efficiency by Group BANK Conventional

2002 0.45

2003 0.72

2004 0.62

2005 0.69

2006 0.85


0.23 0.44 0.46 0.32 0.79 0.75

0.70 0.81 0.51 0.80 0.75

1.00 0.53 0.64 0.56 0.77 0.71

1.00 0.62 0.71 0.64 0.82 0.72

0.80 0.85 0.97 0.80 1.00

Islamic

0.70

0.79

0.69

0.81

0.88

Full Fledged Full Branch Regional I

0.76 0.73 0.45

0.85 0.86 0.36

0.76 0.68 0.65

0.96 0.87 0.67

0.97 0.88 0.83

Table 4.7 Summary of DEA Return to Scale 2002 Bank

2003

2004

2005

2006

Share

Bank

Share

Bank

Share

Bank

Share

Bank

Share

14 10 49 73

19% 14% 67%

21 41 40 102

21% 40% 39%

8 26 64 98

8% 27% 65%

8 24 31 63

13% 38% 49%

16 25 3 44

36% 57% 7%

3 1 3 7

43% 14% 43%

3 1 4 8

38% 13% 50%

1 2 10 13

8% 15% 77%

9 1 9 19

47% 5% 47%

12 2 5 19

63% 11% 26%

Conventional CRS IRS DRS TOTAL

Islamic CRS IRS DRS TOTAL

Table 4.8 Summary of DEA Reference Set 2002 Bank Conv MIXED Conv DOM Conv DOM fx Conv DOM fx Conv MIXED Conv DOM Conv MIXED Islamic FF Islamic FB Conv MIXED

Freq 66 49 39 39 24 17 10 9 5 1

2003

2004

2005

2006

Bank Freq Conv MIXED 64 Conv FOREIGN 56 Conv DOM 46 Conv DOM+ 45 Conv DOM 21 Conv DOM 19 UFJ 16 Islamic FB 14 Conv DOM 12 Conv MIXED 3 Conv FOREIGN 2

Bank Freq Conv MIXED 101 Conv FOREIGN 95 Conv DOM fx 26 Conv MIXED 24 Conv PUBLIC 22

Bank Freq Conv MIXED 57 Islamic REG 23 Islamic FB 21 Islamic FB 21 Islamic FF 19 Islamic FF 16 Conv DOM 15 Conv MIXED 10 Islamic FB 7 Conv FOREIGN 5 Conv MIXED 1

Bank Freq Conv DOM 28 Islamic FF 21 Islamic FB 18 Islamic REG 13 Islamic REG 12 Islamic REG 12 Conv DOM fx 12 Conv DOM 10 Conv DOM 7 Conv REG 7 Islamic FB 6 Islamic FB 5 Conv DOM 5 Conv REG 2 Conv MIXED 1 Conv FOREIGN 0

27

Table 4.9 Summary of DEA Potential Improvements (in %) 2002

2003

2004

2005

2006

Conventional Financing Income Deposit Labor Asset

15% 0% 29% 28% 29%

13% 1% 34% 27% 25%

1% 2% 37% 35% 26%

17% 1% 26% 37% 20%

15% 3% 23% 33% 26%

4% 38% 13% 30% 16%

0% 9% 28% 35% 28%

25% 21% 21% 17% 17%

0% 20% 26% 30% 25%

14% 0% 35% 25% 25%

Islamic Financing Income Deposit Labor Asset

Table 4.10 Summary of Non-parametric DEA vs. OCOI and ROA Ratios BANK Conventional Public Domestic fx Domestic Regional C Mixed Foreign

OE

TE

SE

OCOI

ROA

0.80 0.85 0.97 0.80 1.00

0.86 0.88 1.00 0.87 1.00

0.93 0.97 0.97 0.92 1.00

1.05 0.99 0.64 0.79 0.64

0.20 1.01 4.52 2.87 2.50

Islamic Full Fledged 0.97 1.00 0.97 0.86 2.25 Full Branch 0.88 0.89 0.99 0.78 1.98 Regional I 0.83 0.85 0.98 0.95 2.95 Note: OE = Overall Technical Efficiency; TE = Pure Technical Efficiency; SE = Scale Efficiency; OE = TE x SE; OCOI = Operating Costs/Operating Income; ROA = Return on Assets

Table 4.11 Summary of Parametric SFA and DFA vs. Non-parametric DEA BANK

OE 0.850

DEA TE 0.884

SE 0.962

0.999 0.999 0.999 0.999 0.999 0.999

0.804 0.853 0.974 0.799 1.000

0.862 0.876 1.000 0.874 1.000

0.934 0.975 0.974 0.915 1.000

1.000

0.877

0.890

0.985

DFA

SFA

Conventional

0.890

1.000


0.929 0.914 0.881 0.867 0.890 0.896

Islamic

0.870

Full Fledged 0.859 0.999 0.973 1.000 0.973 Full Branch 0.882 0.999 0.882 0.889 0.993 Regional I 0.843 0.999 0.831 0.845 0.981 Note: OE = Overall Technical Efficiency; TE = Pure Technical Efficiency; SE = Scale Efficiency; OE = TE x SE.

28