Review of literature - The Ohio State University

4 downloads 0 Views 240KB Size Report
provides estimates of the various efficiency scores for the UAE banking sector, investigates the .... the long run socially efficient level of operation. Therefore ...

Measuring and Explaining the Efficiencies of the United Arab Emirates Banking System

By Mohamed Y. El-Bassiouni i

Fatima S. Al Shamsi

Hassan Y. Aly

UAE University

The Ohio State University

UAE University

Accepted and forthcoming in Applied Economics i

The authors would like to thank a reviewer for his/her helpful comments and would like to acknowledge with gratitude the financial support provided by the UAE University (interdisciplinary research grant # 01-4-12/02).

Measuring and Explaining the Efficiencies of the United Arab Emirates Banking System

Abstract Using a newly collected data from a survey distributed to all banks in the United Arab Emirates (UAE), this paper measures economic efficiency in the banking industry, namely allocative, technical, pure technical and scale efficiency. Employing a non-parametric -DEA- approach, the study estimates the efficiency for a cross section of the UAE banks in 2004. The results indicate that the dominant source of inefficiency in the UAE banking is stemming from allocative inefficiency rather than technical inefficiency. Furthermore, the main source of the relatively small size, technical inefficiency in the UAE banking industry is not the scale inefficiency but rather the pure technical inefficiency. The results further indicate that the UAE banks are able to use their input resources more efficiently when they have more branches, and that newer banks are performing better than older banks on average. Moreover, the results also show that short experiences of employees affect efficiencies negatively and government ownership tends to reduce efficiency (as the government shares increases in the bank, the efficiency scores get lower). Finally, the most interesting results have to do with finding higher average efficiencies in banks that employ more women, more mangers, and less national citizens of the UAE.


Introduction The banking industry worldwide is facing competitive pressure as the world

financial structure is changing rapidly due to the ramifications of the establishment of the World Trade Organization. Deregulating the financial sectors and financial services, the increasing use of information technology, and the huge speed of dispensing financial information are also among the factors leading to reevaluating and restructuring of financial institutions worldwide. In such conditions, bank regulators, managers, investors as well as governments are concerned about how efficiently banks transform their inputs into various financial products and services or in simple wording how efficiently banks performs their functions. Thus, in order for financial institutions to survive the upcoming battles created by the expected stiff worldwide competition, as by product of the enactment of WTO regulations, it is imperative that these institutions be subjected to measurement and evaluation of its own economic efficacies and performance. The UAE banking sector is no exception. Having served as a financial center for the Middle East region, the UAE banking institutions need to revaluate their performance. Moreover, the financial sector needs to compare its economic efficiencies with the competitors from around the world who soon might be able to enter the UAE market and seize part of the local share. It is a matter of vital importance for bank managers, bank regulators, and the Central Bank authority of the UAE to get full information about the bank’s economic efficiencies (technical, allocative and scale efficiencies). To this end, the current study provides estimates of the various efficiency scores for the UAE banking sector, investigates the types of inefficiencies (if any) and identifies their sources.

However, while the measurement and the comparisons of efficiencies are significant and vital objectives for this study, explaining the efficiencies and attempting to analyze them is more important than ever. Accordingly, additional demographic, social and technical data were collected and used to go one step further beyond the classical analysis and explain why the efficiency scores are what they are and what are the factors that could help enhance them and finally how governmental regulations may help their improvement. The main findings of the study indicate that UAE banks are generally technically efficient but allocatively less efficient. Also, the results show real small scale inefficiency, if any, exist in the UAE banking sector. In explaining these results, the study confirms that factors such as branching, having more women employees as well as more managers, and having longer experience for employees do positively impact average efficiencies. On the other hand, government ownership and having more national citizens as employees do negatively affect the average efficiencies In section II a background of the UAE banking industry is provided whereas a pertinent review of literature is presented in Section III with special attention to studies conducted on MENA (Middle East and North African) countries or comparable economies. Section IV outlines the methodology used in this study in addition to identifying the models, variables and data sources. Section V presents the empirical DEA results (efficiency scores), while Section VI explains these results in light of the explanatory variables (demographic, social, regulatory, and economic) that are used for the first time in such studies. Section VII concludes by policy implications, limitations, and a brief summary.


Background of the UAE Banking Industry The UAE banking sector consists of the UAE Central Bank, the commercial

banks (local and foreign) and other specialized banks and financial institutions. The Central Bank of the UAE was formed in 1980. Its duties include advising the government on monetary and financial issues, issuing currency, maintaining gold and foreign currency reserves, formulating credit policy and providing regulation and supervision. Total assets of the UAE Central Bank (CB) reached AED 67.6 billion in 2004 and its net earnings during 2004 increased by 34.4% from AED 559.2 million to AED 801.9 million (MENAFN.Com, 2005). Banks play a critical role in the UAE economy, and the banking system is strong and developed, technologically advanced and more integrated into the world economy. According to the IMF, the banking system is broadly profitable, well supervised, and capitalized (IMF public information notice, March 2003). Benefiting from relatively low operating costs and widening margins between deposit and lending interest rates, combined net profits of UAE banks rose by 39.55% in 2004 (MENAFN.Com, 2005). The ratio of gross nonperforming loans to total loans continued to decline, but remains high at 14% compared to other GCC countries (Kuwait is 6%), while capital adequacy ratios remained significantly above international standards. Bank assets constitute around 184% of non-oil GDP and 125% of total GDP in 2003 (Central Bank, 2004). The growth in the banking GDP is attributed to the high credit and deposit growth on the back of low interest environment. Consolidated assets of the banking sector grew by 11.3% to AED 413.7 billion due to the growth of total credit by 17.3% to AED 238.2 billion in 2004 (Central Bank, 2004). Some selected monetary and banking indicators are given in Table 1.


Table 1. UAE Selected Monetary and Banking Indicators 1990 1995 1999 2000 2001 2002 2003

CB assets (billion 21.1 27.2 39.2 45.8 52.5 55.3 54.5 AED*) CB assets to equity 20.5 20.2 20 18.9 18.2 ratio Banks assets 153.9 180.9 251.1 264.2 235.1 331.6 366.9 (billion AED*) Gross non 15.2 14.4 13.6 12.7 15.7 15.3 14.3 performing loan/total lending Return on assets 1.9 2.0 1.5 1.8 2.6 2.2 2.3 Return on equity 17.1 18.3 12.8 14.9 16.7 15.6 16.4 Average interest 4.4 4.5 5.8 3.7 4.7 5.6 4.7 rate spread (%) UAE Banks** (No.) 207 241 295 370 324 345 367 F. Banks** (No.) 119 119 110 109 109 112 112 Labor (No.) 9,677 10,916 14,274 14,385 15,191 16,080 17,229 GDP*** share (%) 4.2 4.9 6.4 5.8 6.6 6.6 6.3 Non-oil GDP*** 8.7 9.6 9.5 9.5 share (%) Notes: * 1 AED = US$ 3.668 ** includes head offices and their branches *** For all financial institutions and insurance Sources: Central Bank of the UAE, Statistical Bulletin volume 24, No.3 September 2004 and IMF (2004), Country report No. 04/ 175, June 2004.

2004 59.9


377 112 18,381

The UAE commercial banks market is quite large compared to other Gulf States. At the end of the 2004, there were 46 banks, with 489 branches, serving a population of around 3.6 million compared to 14 banks in Saudi Arabia serving some 23.5 million inhabitants (Central Bank & Global Investment House, 2004). Of the UAE commercial banks, 21 were locally incorporated, with a total of 377 branches and 70-100% UAE ownership. The remaining 25 banks, with 112 branches, are foreign owned. It is worth mentioning here that foreign banks are restricted to no more than five branches. The total assets of the commercial banks have expanded significantly during the past years reaching AED 413.7 billion in 2004 (of which more than 27% are foreign assets). The UAE banking sector is the second after Saudi Arabia within the GCC in term of asset size. Asset size of national banks increased by 11.5% to AED 314.3 billion in 2004, while that of foreign banks increased by a similar rate (11.2%) to AED 99.5 billion for the same year (Central Bank, 2004). The

financial institutions’ GDP share reached 6.3% in 2003 and their share of non-oil GDP reached almost 10% for the same year, while the total workforce in the banking sector reached 18,381 workers in 2004 (of whom less than 2% are nationals). The top five banks dominate the industry, accounting for roughly 50% of all assets and 75% of all deposits. They are partly owned by the governments of Abu Dhabi and Dubai. Among the local banks, there are four Islamic banks that follow the Islamic banking principle. The UAE commercial banks provide a full range of financial services; covering retail and investment banking, and their commercial lending are dominated by short- and medium-term loans. Total bank credit extended to residents and nonresidents reached AED 238.2 billion in 2004, a rise of 17% from 2003. An analysis of the components of bank credit according to economic activities reveals that the largest share of bank credit was extended to the trade sector which attains an average share of 29% for the last three years. The second sector in term of acquired credit is the construction sector with a share of 14.5% on average for the period from 20022004 (see Table 2). The distribution of credit by economic activities shows an increase in the credit extended to the government in the last three years by 22.5% in 2003 and 32.5% in 2004, which increased the government share in total bank credit from 9.2% in 2002 to 12.8% in 2004. Table 2. Bank Credit by Economic Activity 2002-2004 (Million AED) Activity Trade Construction personal loans for Business personal loans for consumption Government manufacturing Transport, communication Electricity, water and gas Mining and quarrying Agriculture

2002 48870 27063 20716 17704 15222 9901 5127 3219 2213 1154

% 29.5 16.3 12.5 10.7 9.2 5.9 3.1 1.9 1.3 0.7

2003 57053 26845 23965 21443 19650 11082 6325 11110 2077 830

% 28.9 13.6 12.2 10.9 9.9 5.6 3.2 5.6 1.1 0.4

2004 69083 31856 30392 23195 29134 13717 6573 9113 2497 880

% 29.0 13.5 12.8 9.7 12.2 5.8 2.8 3.8 1.0 0.4

Non-banking financial institution 1903 1.1 2272 1.2 Other 12512 7.8 14254 7.4 Total 165604 100 196906 100 Source: Central Bank of the UAE, Statistical Bulletin volume 24, No.3 September 2004.

3745 18023 238208

In compliance with the Central Bank regulation, all 46 banks operating in the UAE met the 10% capital asset ratio. Bank lending has grown substantially, increasing at an annual rate of 17% during 2001-2003. Securities markets are a new development in the UAE financial systems that were launched in 2000. There are two exchanges in the UAE, Abu Dhabi securities market (ADSM) and Dubai financial market (DFM). ADSM is the third stock market by capitalization in the Arab world after Saudi Arabia and Kuwait which was estimated to be at more than AED 111 billion at the end of 2003 and DFM has a market capitalization at about AED 60 billion. The transaction volume of the UAE securities markets increased by 40% in 2003; however, market liquidity remains low and speculation is restrained. The general market index grew at 32% in 2003, "among the lowest appreciation in the GCC market" (IMF, 2004, p47). Dubai International Financial Center (DIFC) is another new development that added to UAE financial system in 2004. DIFC was initiated by Dubai government as a free zone financial center and is expected to encompass a comprehensive set of international financial functions. DIFC activities will include institutional and investment banking, insurance and re-insurance, asset management, Islamic financial services, back office operation and an international exchange that will trade a full range of financial instrument (IMF, 2004, p49). DIFC is intended to be segregated from the UAE financial sector and its banking operation is confined to institutional wholesale banking. The launch of DIFC is expected to give the country an opportunity to be a "universally recognized hub for institutional financial services and

1.6 7.4 100

the regional gateway for capital and investment " (Global Investment House, 2004, p.3). III.

Review of Pertinent Literature There are two major approaches for measuring efficiency that have been used

in all studies; the parametric approach and the non-parametric approach. In general, the parametric approaches specify a functional form for the cost, revenue, profit, or production relationship among inputs, outputs and others (for instance, environmental factors), and allow for random error. The most famous technique used in the parametric approach is Stochastic Frontier Approach (SFA). On the other hand, the nonparametric approach most well-known technique is the Data Envelopment Analysis (DEA). The DEA- to be used in this study- computes the relative efficiency of each bank by using multiple inputs and multiple outputs. Linear programming techniques allow for the construction of best practice cost and production frontiers from these data and the performance of a particular bank is then judged relative to this frontier. The specific efficiency measures calculated can be given fairly simple interpretations. The technical efficiency measure gives the proportional reduction in input usage, which could have been achieved if the firm operated on the efficient frontier. The technical efficiency can be decomposed into the proportional reduction in input usage, if inputs were not wasted (pure technical efficiency) and that reduction if there existed constant returns to scale (scale efficiency). As such, pure technical inefficiency reflects excess input levels for a given level of output. This inefficiency is unique in that it is caused and could be corrected by proper management. From the economics’ standpoint, firms that operate at constant returns to scale represent the long run socially efficient level of operation. Therefore, choosing non-constant scale of operation also constitutes inefficiency.

In this section a review of the DEA literature limited to recent studies that applied the technique to MENA (Middle East and North African) countries or comparable emerging economies or made international comparisons between different countries is presented ii . First, a comprehensive study by Berger and Humphrey (1997) surveyed over 125 studies that applied frontier approaches to measure the performance and efficiency of financial institutions in more than 20 different countries. Most of the studies are conducted in the U.S. banking industry between 1990 and 1998. Very few studies were done outside the US and a call for the need to examine the efficiency of banks outside the United States was emphasized. In their article, Berger and Humphrey, illustrated differences and dispersion in efficiency estimates between nonparametric and parametric frontier techniques. They reviewed and critically analyzed empirical estimates of financial institution efficiency in order to address the implications of efficiency results in the areas of government policy, research, and managerial performance. Mohd Zaini Abd Karim (2001) investigates whether there are significant differences in bank efficiency across select ASEAN countries (Indonesia, Malaysia, Philippines, and Thailand). The study indicates that the substantial proportion of a total variability is associated with inefficiency of input used. It also highlights the fact ii

For more details on the issue of measuring bank efficiency, using the non-parametric DEA approach and its comparability across different countries, regions of the world and under various operational and environmental working conditions, the interested reader is referred to Drake 2001, Devaney and Weber 2002, Kong and Tongzon 2006, Steinmann and Zweifel 2003, Sathye 2001, Miller and Noulas, 1996, Hasan and Marton 2003, Drake and Hall 2003, Akhigbe and McNulty, 2003, Esho 2001, Santomero and Seater, 2000, De Young, Robert 1997, Barnum and Gleason 2006, Allen and Anoop 1996, Lozanzo-Vivas et al. 2002, Orea 2002, Staat, Matthias 2006, Cook 2000, Clark and Siems 2002, Vennet, 2002, Von Hirschhausen et al 2006, Shujie Yao 2006, Wheelock and Wilson 1999, Athanassopoulos 1998, Peristiani 1997, Clark 1996 and Allen N. Berger and Loretta J. Mester (1997). These studies may be considered examples and are not intended to be an exhaustive survey.

that inefficiency tends to decrease with bank size and increase with government ownership. David A. Grigorian and Vlad Manole (2002), using both cross-country and cross-regional settings, and applying the DEA approach, tried to calculate an appropriate measure of commercial bank efficiency in a multiple input/output framework for transition economies, and to identify the effects of policy framework on the performance of commercial banks. The results of the study illustrated that banks with a larger market share and a larger controlling foreign ownership are likely to be more efficient than those with a smaller market which was attributed to their better risk management and operational techniques. They also discovered that banks in higher per capita income countries are more efficient in terms of attracting more deposits and generating stronger cash flows than banks in low income countries. It is also indicated that while privatization of state-owned enterprises, enterprise competition and corporate governance related improvements are important in boosting commercial bank efficiency, the securities market and non-bank financial institutions development hinders the efficiency of banks. Ihsan Isik, M.Kabir Hasssan (2002) estimated allocative efficiency, scale efficiency and overall cost efficiency of Turkish commercial banks. The study correlated 4 measures of financial performance with the 5 measures of cost efficiency and investigated whether higher performance impacts bank cost efficiency. They discovered that the efficiency of the Turkish commercial banks had deteriorated in the 90’s and the dominant source of the cost inefficiency is technical, which they suggested might be attributable partly to the recent abnormal growth of small and medium banks, and partly to the recent heavy investment in expensive computerization and automation projects and consequently idle capacity. They

examined the relationship between the X-efficiency measures and the proxy measures of performance to reveal that there is a statistically significant relationship. The study also suggests that strong competition might have induced more market discipline on small banks, leading to greater cost efficiency. The study also indicates that foreign banks are significantly more efficient than their domestic peers and private banks are more efficient than public banks. Their study suggested that publicly traded banks are more technically efficient than privately held banks, and banks under a holding company structure are more efficient than independent banks. In their study, Ali F. Darrat (2002), investigated the efficiency of banks in the MENA region. The study uses DEA and Malmquist total factor of productivity index to evaluate the performance of 8 locally incorporated Kuwaiti banks, in terms of their efficiency, productivity growth and technological change over the period 19941997. The study indicates that small banks are more efficient and there is an upward trend in cost efficiency of Kuwaiti banks. The study also suggests that while technical efficiency of Kuwaiti banks is consistently higher than allocative efficiency, scaleefficiency is also persistently higher than pure-technical efficiency over the estimation period. Market power plays an important role in cost and allocative efficiencies and capitalization of Kuwaiti banks positively impacts their cost efficiency. It is also suggested that Kuwaiti banks experienced productivity growth over the sample period from becoming more technologically advanced rather than being more technically efficient. Claudia Girardone (2004) examined the determinants of Italian banks’ cost efficiency over the period 1993-1996 by employing a Fourier-flexible stochastic cost frontier in order to evaluate X-efficiency and scale economies and to identify the main characteristics of efficient banks. The important findings of the study include:

deregulation has a positive impact and X-efficiency levels seem to decrease over time, for all sizes of banks, the smallest banks appear to be less inefficient than their larger counterparts, the main differences between the most efficient and the least efficient bank seem to be mainly related to staff expenses, efficient banks always appear to have lower levels of equity/assets and higher levels of non-performing loans, inefficiencies appear to be inversely correlated with capital and positively related to the level of non-performing loans, inefficient banks tended to have (on average) a greater retail banking orientation, higher interest margins and more branches compared with their efficient counterparts, no clear relationship between assets size and bank efficiency. Finally Berger (2007) conducted a comprehensive survey article where he reviewed and critiqued over 100 studies that compare bank efficiencies across nations. His goal was explaining the consolidation pattern among financial institutions in developed versus developing countries. He grouped these studies into three distinct categories: (1) studies that make the comparisons of bank efficiencies in different nations based on the use of a common efficient frontier, (2) other that make the comparisons of bank efficiencies in different nations using nation-specific frontiers, and (3) finally the ones that compare the efficiencies of foreign-owned versus domestically owned banks within the same nation using the same nation-specific frontier. He used the results to explain the consolidation patterns in different part of the world and he concluded that the efficiency disadvantages of foreign-owned banks relative to domestically owned banks tend to outweigh the efficiency advantages in developed nations on average, and this situation is generally reversed in developing nations.


Methodology and Data From section III, we can clearly conclude that each approach of measuring

efficiency (parametric or non-parametric) has its own advantage and shortcomings. The main disadvantage of the parametric approach is the imposition of certain functional forms on the technology used in production. The approach does dictate the type of technology used by utilizing certain functional forms and not others. Some of these restrictions on the technology could be very severe depending on the flexibility of the functional form (or lack thereof). While the non-parametric approach does not impose any restrictions on the technology, it does not allow for any stochastic random disturbances. Variations in efficiency might be due to random shocks stemming from external factors like weather changes or the like. The approach selected for use in this study is the non-parametric approach. The random disturbance in the banking industry is insignificant in comparison to another industry (e.g., agriculture) where weather and external factors could have great impact on productivities. Also, the benefits from using no restricting technology could be greater in the banking industry compared to others, due to the varied methods of producing financial services among different financial institutions. In addition, it is clear from the review of relevant literature (Berger and Humphrey, 1997, cited earlier) that efficiency estimates from nonparametric studies are very similar to those from parametric frontier models; but nonparametric methods generally yield slightly lower mean efficiency estimates. Also, since most of the studies conducted on the banking sectors in other countries have used the nonparametric approach, for the sake of consistency in comparison, this study uses the same approach.

The specific technique to be used to measure the overall technical efficiency index of a bank is the solution of a group of linear programming problems that are very well known by now. They can be illustrated by: Min λ such that


L1Z1 + L2Z2 + . . . + LnZn ≤ λLn K1Z1 + N2Z2 + . . . + KnZn ≤ λKn Y1Z1 + Y2Z2 + . . . + YnZn ≥ Yn Zi ≥ 0 , i = 1, . . ., n ,

where n denotes the number of observations, λ represents the overall technical efficiency index and Z is a variable measuring the intensity of production, Y denotes output and L and K denote labor and capital as defined below. Note that the condition Zi ≥ 0 imposes constant returns to scale in the underlying technology. In order to relax the assumption of “constant returns to scale” (CRS) and separate λ into “pure technical efficiency” (PTE) and “scale efficiency” (SE), an additional linear programming problem suggested by Färe, Jansson and Lovell (1985) is constructed and solved as follows: (2) Min β such that

L1Z1 + L2Z2 + . . . + LnZn ≤ βLn K1Z1 + K2Z2 + . . . + NnZn ≤ βKn Y1Z1 + Y2Z2 + . . . + YnZn ≥ Yn ∑n Zi = 1, where β is a measure of pure technical efficiency. Having calculated β, a measure of scale efficiency (SE) can be computed as follows:


SE = λ/β. The efficiency scores of all Decision Making Units (banks in this case) are

bounded between 0 and 1, with the most efficient banks receiving an efficiency score of unity. The CRS assumption is appropriate if all banks are operating with optimal scales. Otherwise, technical efficiency scores could be confounded with scale efficiency. For more details on the specificities of the techniques see among others Aly et al. 1988 and 1990. Now, we turn to the data and definition of input and output variables. That part of the required data is collected from balance sheets and income statements from most of the commercial bank, operating in the UAE for the last five years. The three major inputs targeted by this study are labor, capital and deposits, and the two outputs are loans and investments. The choice of inputs and outputs for banks has received much attention in literature due to the unique nature of the bank’s production process. There are two approaches in the literature. The production approach and intermediation approach. In the production approach, banks are considered as units producing services for clients such as performing transactions and processing documents. Therefore, inputs are measured by physical units, and outputs are measured by the number and type of transactions or documents processed over a given time period. Under the other intermediation approach, banks are viewed as channeling funds between depositors and borrowers. Thus, banks sustain labor, capital and loanable funds expenditures to transfer funds from those with surplus of funds to those with shortage of funds. Thus, total costs should include interest expenses as well as operating costs (Topuz and Isik, 2004). The study follows the second approach.

Labor (L) is measured by the total number of employees, capital (K) by the book value of fixed assets and premises, and deposits (D) by the sum of demand and saving deposits. The analysis also includes input prices in order to measure cost efficiency. The unit price of labor (WL) is the total cost of all bank employees (i.e., salaries, employee benefits, etc) divided by the total number of employees. The unit price of capital (RK) is measured by the total expenditure on fixed assets and premises divided by the book value of fixed assets and premises. The unit price of deposits (PD) is computed by the total interest expenses of deposits divided by the sum of demand and saving deposits. As to the outputs, loans (L) include all types such as real estate loans, commercial and industrial loans, and consumer loans. Investments (I) reflect the value of all securities other than those held in the bank’s trading accounts. It is important to mention here that the appropriate number of inputs and outputs is determined based on the available data. As a general rule, the product of inputs times outputs in a DEA approach should optimally be less than the sample size in order to effectively differentiate among banks. In our case, the general rule is satisfied since the total number of valid and non missing observations is 22 (twenty two banks) and the restriction based on the general rule requires only 6 (six banks). In addition, data on the education, gender, ethnicity and race mix of the employees; number of branches; size of foreign transactions; proportion of different types of loans; and number of financial products offerings are needed for explaining the efficiency scores as mentioned earlier. This major part of the data (demographic, social, and technical variables) was collected through a survey instrument (see Appendix I) that was designed and tested on two banks before it was distributed to all commercial banks operating in the UAE

in the year 2004. The Central Bank assisted in collecting the data by providing an attached letter encouraging the banks to respond to the survey. However, the response rate is about 54% (25 banks out of 47) and the usable observation is 22 banks. While the collection was conducted in the Spring and Fall of 2004, the data were collected for the year ending in December of 2002, the last year that banks had full information for the required data. To explain the efficiency scores based on socioeconomic, regulatory, and demographic variables, we used the regular ordinary least square method as well as a logistic regression approach due to the nature of the dependent variable which lies in the interval [0,1]. The logistic regression analysis was performed through the implementation of Weighted Least Square (WLS), with a backward stepwise elimination option, to relate the dependent variable: efficiency score (y), and the independent variables: number of years in operation (X1), the percentage of government participation (X2), the number of branches (X3), the number of IT employees (X4), the percentage of males (X5), the percentage of managerial employees (X6), the percentage of national employees (X7), the percentage of employees with higher education; university plus (X8) and the percentage of employees with short experience; less than 5 years (X9). The logit transformation was used to obtain the logistic regression model


⎛ y ⎞ ⎟⎟ = β 0 + β 1 X 1 + β 2 X 2 + ... + β 9 X 9 . ln⎜⎜ ⎝1− y ⎠

The weights which are inversely proportional to the variances of the logits were approximated by (yi(1-yi)). The descriptive statistics on the variables used in this study are reported in Table 3.

Table 3. Descriptive Statistics for the UAE Banking Data Item


Total wages** No. of employees (Labor) Wages per capita** (Unit price of labor) Book value of assets and premises** Total spending on fixed assets and premises** Sum of current accounts and savings deposits** Sum of commercial loans** Sum of industrial loans** Sum of consumer loans** Sum of other loans (if any) Total interest on current accounts and savings deposits** Sum of real estate loans** Total loans** (L) Total investments** (I) Unit price of capital** Unit price of deposit**



48.35 486.05 0.11 138.11 36.82

Std. deviation 45.75 431.13 0.03 151.87 67.24

3.92 39 0.07 4.00 0.00

148.26 1386 0.18 547.00 274.60





1469.98 1032.23 1448.74 719.42 406.50

2593.60 1959.31 2275.17 1971.31 1080.52

0.37 0.00 7.98 0.00 0.00

12131.65 8917.29 8262.31 9088.00 5054.00

779.66 2957.70 740.83 0.62 1.93

1591.10 4148.52 1123.20 1.18 3.62

0.00 217.09 0.00 0.00 0.00

6678.80 17815.00 3579.36 5.163 16.58

** in million AED


DEA Results The DEA analysis produced the various efficiency scores in Table 4. The

average overall cost efficiency is a bit low (55%). This means that the average UAE bank could have produced the same level of output using only 55% of the resources actually employed had it been producing on the frontier rather than at its current level. Another way to explain this is to say that the UAE banks do have potentials to increase their level of output by another 45% using the same level of input they actually have. Also, such an overall average of cost efficiency (55%) is a bit lower than those typically reported for developed countries. Here, we make the comparison between the efficiencies of banks in different nations, with banks from each nation measured against their own nation-specific frontier not against a common frontier iii . iii


In this case, the standardization of the DEA methodology seems to tolerate such comparisons of the size of the efficiency coefficients. However, one has to bear in mind the national differences in regulations, legal system, economic and financial markets conditions in the different countries, as well as the differences in the time period the research covers - that will surly results in different frontiers (see Berger, 2007).

example, Aly et al. (1990) reported overall efficiency of 65% for US banks, Berger et al. (1993) estimated cost efficiency at 80% for U.S. banks, and Altunbas et al. (1994) estimated it at about 95-90% for British banks. In another Gulf country (Kuwait)where the legal, economic, financial, and social conditions are almost the sameDarrat et al. 2003 reported banks to have cost efficiency of 67%.

Table 4. Efficiency Scores for the UAE Banks in 2002 Bank* Bank 1 Bank 2 Bank 3 Bank 4 Bank 5 Bank 6 Bank 7 Bank 8 Bank 9 Bank 10 Bank 11 Bank 12 Bank 13 Bank 14 Bank 15 Bank 16 Bank 17 Bank 18 Bank 19 Bank 20 Bank 21 Overall Average CE: Cost Efficiency AE: Allocative Efficiency TE: Technical Efficiency PTE: Pure Technical Efficiency SE: Scale Efficiency

CE 0.45 0.30 0.03 0.45 0.22 0.89 1.00 0.79 0.87 0.37 0.22 0.57 1.00 0.81 0.31 0.68 0.30 0.53 0.07 1.00 0.75 0.55

AE 0.45 0.42 0.82 0.94 0.38 0.92 1.00 0.79 1.00 0.43 0.28 0.57 1.00 0.81 0.31 1.00 0.36 0.54 0.57 1.00 0.75 0.68

TE 1.00 0.71 0.04 0.48 0.59 0.97 1.00 1.00 0.87 0.85 0.77 1.00 1.00 1.00 1.00 0.68 0.83 0.98 0.12 1.00 1.00 0.80

PTE 1.00 0.71 0.04 0.48 0.59 0.97 1.00 1.00 0.87 0.88 0.78 1.00 1.00 1.00 1.00 0.68 0.83 0.98 0.12 1.00 1.00 0.80

SE 1.00 1.00 1.00 0.99 1.00 1.00 1.00 1.00 1.00 0.96 0.98 1.00 1.00 1.00 1.00 1.00 1.00 0.99 1.00 1.00 1.00 0.99


RTS: Returns to scale CRS: Constant returns to Scale DRS: Decreasing returns to scale IRS: Increasing returns to scale

* Due to a confidentiality clause in the survey we used to collect the data, the names of the banks are not revealed in this research project and will be provided to the participating banks upon request.

The results also reveal several other very important inferences. Firstly, the low overall cost efficiency stems from allocative inefficiency rather than technical inefficiency. Actually, the technical efficiency (.80) of UAE banks is consistently higher than the allocative efficiency (.68) for most of the banks included in the

sample. Whether small or large, local or foreign, banks in the UAE seem to be doing a better job technically than allocatively. This indicates that the leading source of cost inefficiencies in the UAE banking industry is likely to be regulatory (rather than managerial) in nature. The results imply that UAE banks do a better job utilizing available inputs than choosing the proper input combination given the input prices. Secondly, the main source of the relatively small size technical inefficiency in the UAE banking industry is not the scale inefficiency (operating on non-optimal scale) but rather the pure technical efficiency (resources being underutilized or merely wasted). Actually, the UAE banks in the sample have consistently higher scale efficiency than pure-technical efficiency. This happened without exception over every single bank in the sample. Surprisingly, the UAE banks have almost perfect scores when it comes to scale efficiency. Only 1% of the technical inefficiency in the UAE could be attributed to scale. This is probably the lowest scale inefficiency in comparison to all previous studies done all over the world. VI.

Regression Results In this section the more interesting part of the study, namely explaining the

efficiency scores based on socioeconomic, regulatory, and demographic variables will be attempted. We have used both the regular ordinary least squares and the logistic regression methods, but we put more emphasis on the logistic approach due to the nature of the dependent variable which lies in the interval [0, 1]. Due to the importance of the logistic approach, its results for cost efficiency are detailed in Appendix II. Some of the interesting results obtained from the attempts above include: Firstly, the fact that only technical and pure technical efficiency scores seemed to be partially explainable by the number of branches. The coefficient of this variable

(branching) is positive and statistically significant (under 95% percent level of confidence with respect to technical efficiency scores). This means that the more branches the bank has, the higher its technical and pure technical efficiency scores. This result was confirmed by both logistic and OLS regressions. Overall, it matches other studies where banks are able to use their input resources more efficiently when they have more branches. Secondly, it is also interesting and surprising to see that the variable reflecting the bank’s number of years in business is negatively impacting all the efficiency scores, but at a statistically lower level of confidence (90% on average). This implies that newer banks are performing better than older banks on average. While this is surprising since experienced banks are expected to perform better, it may have to do with newer and more recent banks adopting newer and more modern technology. This interpretation is supported by a negative correlation of -0.686, with an associated pvalue of 0.007, indicating a highly significant inverse relationship between the bank’s number of years in operation and the IT share (percentage) of its budget. Also, newer banks are always able to attract experienced employees from existing and older banks. Thirdly, while the variable representing private versus government ownership (percentage of government ownership) is not statistically significant for any efficiency score (only at 75% level of confidence on average), it is interesting to see a negative sign in all regressions and with all types of efficiency scores. This indicates that as the government shares increases in the bank, the efficiency scores get lower. This may lead us to trust that privatization might be an appealing option for government owned banks that are under performing in terms of efficiency scores. Fourthly, an easy variable to explain is the short experience variable (percentage of employees with less than five years of experience to total employees).

This variable seems to be negative and statistically significant at all levels. Thus, banks with higher percentage of employees with short experience are definitely less efficient than its counter parts. This will give credence to human resource mangers as they always opt to hire employees with more experience. Fifthly, a very interesting variable to explain is the percentage of male employees to total employees.

The logit analysis indicates that this variable is

negative and statistically significant. This is good news to the women labor force in the UAE, as it signifies that banks with higher percentage of women are more efficient than their counter parts. While this result is hard to explain, the connotation is to give more attention to hiring more women in the banking industry if the bank wants to be more efficient. Definitely, more studies need to be done here to explain why women might have a comparative advantage in this industry. Also, the variable of percentage of mangers to total employees, having statistically positive impact on efficiency, needs a more careful look on why it is conducive to enhancing efficiency. Finally, the most interesting and important findings here have to do with the variable representing the percentage of nationals (citizens) to the total labor force. This variable is found to be negative and highly statistically significant in the logit analysis. This means banks with higher percentages of UAE nationals are under performing in terms of overall efficiency in comparison to their counter parts. This result is very important due to the ongoing efforts to nationalize the banking industry. The UAE established a law requiring banks to hire at least 4% of its labor force from the local nationals and has given a lot of consideration to this issue. This last result coupled with the allocative (regulatory) inefficiency being the major source of inefficiency, as shown in Table 4, should call attention to the wisdom of enforcing the nationalization policies on the UAE banking sector.


Summary and Policy Implications The descriptive analysis of the UAE banking system shows that it is strong

and highly developed in structure and size, technologically advanced and more integrated into the world economy. This superior picture may be attributed to many factors. The most important is an overall improvement in the country’s economic conditions that have enhanced the provision of the financial services to all sector of the economy. Also, there is always government intervention and support for the country financial sector. "the UAE has used its wealth to cushion financial sector shocks and has the authority to recapitalize systematically important financial institutions" (IMF: 2003). In addition, there is, to some extent, an insufficient competition from foreign banks. However, this rosy picture is not supported by the empirical results of efficiency measures, as highlighted in Table 4. The results indicate that the UAE banks' overall average of cost efficiency is a bit lower than those reported for developed countries. It is also lower than that of another GCC country, Kuwait. The study also indicates that the low overall cost efficiency of UAE banking stems from allocative inefficiency rather than technical inefficiency. Furthermore, the main source of the relatively small size, technical inefficiency in the UAE banking industry is not the scale inefficiency but rather the pure technical efficiency. It is also suggested that the UAE banks are able to use their input resources more efficiently when they have more branches and that newer banks are performing better than older banks on average. The results also show that short experiences of employees affect efficiencies negatively and government ownership may tend to reduce efficiency (as the government shares increases in the bank, the efficiency scores get lower).

The most interesting results have to do with finding higher efficiencies in banks that employ more women, more mangers, and less national citizens of the UAE. These last results should be investigated further in order to determine what can be done about it. The results go against the current government policies of nationalization and should be looked upon more carefully. Other means of employing nationals without imposing a restriction on banks in terms of employment policies should be considered. It goes against efficiency optimization to ask a bank to employ certain resources without given much attention to the bank’s own profit maximization policy. Perhaps, a direct subsidy to the employees who are working in the banking sector could be an alternative approach. The results of the study are very much consistent with previous studies (especially in the Gulf). Banks are overall more technically efficient than allocatively efficient. It should be realized that all across the globe, a great force of change is sweeping the banking industry, forcing radical adjustments to new business conditions. As the UAE is entering a new era of structure change and development, it is essential to prepare its domestic banking system for global competition, and it has no choice but to integrate into the global economy. Such development will require careful handling of the existing structural weaknesses and adopting bank policies that keep pace with the new evolution and development in the world financial sector. Thus, attention to the results of this study might prove useful in this regard.

References Abd Karim, Mohd Zaini, Comparative bank efficiency across select ASEAN countries, ASEAN Economic Bulletin. December 2001. Akhigbe, Aigbe, McNulty, James E., The profit efficiency of small US commercial banks, Journal of Banking & Finance, February 2003. Vol. 27, No.2, pp. 307-325. Allen, Linda, Rai, Anoop, Operational efficiency in banking: An international comparison, Journal of Banking & Finance, May 1996. Vol.20, No.4, pp. 655672. Altunbas, Y., L. Evans, and P. Molyneux , Universal Banks, Ownership and Efficiency-A Stochastic Frontier Analysis of the German Banking Market, Institute of European Finance, 1994, unpublished paper (cited in Darrat et al., 2003) Aly, Hassan , Grabowski, Richard, Pasurka, Carl and Rangan ,Nanda, Technical, Scale, and Allocative Efficiencies in U.S. Banking: An Empirical Investigation, The Review of Economics and Statistics, 1990, vol. 72, issue 2, pp. 211-18. Aly, Hassan , Grabowski, Richard, Pasurka, Carl and Rangan ,Nanda, The Technical Efficiency of US Banks, Economics Letters, Vol. 28, No. 2, December 1988. Athanassopoulos, Antreas D., Nonparametric frontier models for assessing the market and cost efficiency of large-scale bank branch networks, Journal of Money, Credit, and Banking, May 1998. Vol. 30, No. 2. Berger, A. N., International Comparisons of Banking Efficiency, Financial Markets, Institutions & Instruments, August 2007, Vol. 16, No. 3. doi:10.1111/j.1468-0416.2007.00121. 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, 1993, 17, 221-49. Berger, Allen N., Humphrey, David B. Efficiency of Financial Institutions: International Survey and Directions for Future Research, European Journal of Operational Research, 1997. Special Issue on “New Approaches in Evaluating the Performance of Financial Institutions” Berger, Allen N., Leusner, John H., Mingo, John J., The efficiency of bank branches, Journal of Monetary Economics, September 1997. Vol. 40, No. 1, p. 141.

Berger, Allen N., Mester, Loretta J. , Inside the Black Box: What Explains Differences in the Efficiencies of Financial Institutions?, Journal of Banking & Finance, July 1997. Vol. 21, Iss. 7, pp. 895-947. Barnum, Darold T., and John M. Gleason, Biases in technical efficiency scores caused by intra-input aggregation: mathematical analysis and a DEA application using simulated data, Applied Economics, 2006, Vol. 38, No. 14. Casu, Barbara, Molyneux, Philip, A comparative study of efficiency in European banking, Applied Economics, November 2003. Vol. 35, No. 17, p.1865. Central Bank of the UAE, Economic Bulletin, June 2004. Central Bank of the UAE, Statistical Bulletin, Volume 24, No. 3, September 2004. Clark, Jeffrey A., Economic cost, scale efficiency, and competitive viability in banking, Journal of Money, Credit, and Banking, August 1996. Vol. 28, No. 3, p. 342. Clark, Jeffrey A., Siems, Thomas F., X-Efficiency in banking: Looking beyond the balance sheet, Journal of Money, Credit, and Banking, November 2002. Vol. 34, No. 4, p. 987. Cook, Wade D., Multicomponent efficiency measurement and shared inputs in data envelopment analysis: An application to sales and service performance in bank branches, Journal of Productivity Analysis, November 2000. Vol. 14, No. 3, p. 209. Darrat, Ali, Topuz, Can, and Yousef.Tarik Assessing Cost and Technical Efficiency of Banks in Kuwait, Studies in Economics and Finance, Vol. 21, No. 2, Autumn 2003, pp.1-21. Devaney, Michael, Weber, William L., Productivity Growth, market structure, and technological change: Evidence from the rural banking sector, Applied Financial Economics, December 2002. Vol. 10, No. 6, p. 587. DeYoung, Robert, Comment on ‘Operational efficiency in banking: An international perspective’, Journal of Banking & Finance. October 1997. Vol. 21, No. 10, pp. 1325-1329. Drake, Leigh, Hall, Maximillian J.B., Efficiency in Japanese banking: An empirical analysis, Journal of Banking & Finance, May 2003. Vol. 27, No. 5, pp. 891-917. Drake, Leigh. Efficiency and productivity change in UK banking, Applied Financial Economics, October 2001. Vol. 11, No. 5, p. 557.

Esho, Neil, The determinants of cost efficiency in cooperative financial institutions: Australian evidence, Journal of Banking & Finance, May 2001. Vol. 25, No. 5, pp. 941-964. Fare, Rolf, S. Grosskopf, and C. A. K. Lovell, Measurements of Efficiency Production, (Boston: Kluwer Nijhoff, 1985). Giokas, D. , Bank Branch Operating Efficiency: A Comparative Application of DEA and the Loglinear Model, Omega International Journal of Management Science, 1991, vol. 19, 549-557. Girardone, Claudia, Molyneux, Philip, Gardener, Edward P.M., Analyzing the determinants of bank efficiency: the case of Italian banks, Applied Economics, Feb 20, 2004. Vol. 36, No. 3; p. 215. Global Investment House, UAE Economic& Strategic Outlook-I, September 2004. Grigorian, David A., Manole, Vlad. Determinants of Commercial Bank Performance in Transition: An Application of Data Envelopment Analysis. Haag, S. E., and P. V. Jaska, Interpreting Inefficiency Ratings: An Application of Bank Branch Operating Efficiencies, Managerial and Decision Economics, 1995, vol. 16, 7-14. Hasan, Iftekhar, Marton, Katherin, Development and efficiency of the banking sector in a transitional economy: Hungarian experience, Journal of Banking & Finance, December 2003. Vol. 27, No.12, pp. 2249-2271. Hughes, Joseph P, Lang, William, Mester, Loretta J, Moon, Choon-Geol, Efficient banking under interstate branching, Journal of Money, Credit, and Banking, Nov 1996. Vol. 28, No. 4; p. 1045. International Monetary Fund (IMF), UAE Country Report, No.04/ 175 June 2004. Isik, Ihsan, Hassan, M. Kabir, Technical, Scale and Allocative Efficiencies of the Turkish Banking Industry, Journal of Banking and Finance, April 2002. Vol. 26, No. 4, pp. 719-766. Kong, Nancy Y.C., and Jose Tongzon, Estimating total factor productivity growth in Singapore at sectoral level using data envelopment analysis, Applied Economics, 2006, Vol. 38, No. 19. Lozanzo-Vivas, Ana, Pastor, Jesus T., Pastor, Jose M., An Efficiency Comparison of European Banking Systems Operating under Different Environmental Conditions, Journal of Productivity Analysis, July 2002. Vol.18, No.1, p. 59.

Middle East North Africa Financial Network Newsletter, 2005, newsletter can be viewed at Miller, Stephen M., Noulas, Athanasios G., The technical efficiency of large bank production, Journal of Banking & Finance, April 1996. Vol. 20, No.3, pp. 495-509. Oral, M. and R. Yolalan ,An Empirical Study on Measuring Operating Efficiency and Profitability of Bank Branches, European Journal of Operational Research, (1990), Vol. 46, 282-294. Orea, Luis, Parametric Decomposition of a Generalized Malquist Productivity Index, Journal of Productivity Analysis, July 2002. Vol. 18, No. 1, p. 5. Peristiani, Stavros, Do mergers improve the X-efficiency and scale efficiency of U.S. banks? Evidence from the 1980s, Journal of Money, Credit, and Banking, August 1997. Vol. 29, No. 3, p. 326. Santomero, Anthony M., Seater, John J., Is there an optimal size for the financial sector?, Journal of Banking & Finance, June 2000. Vol. 24, No. 6, pp. 945-965. Sathye, Milind, X-efficiency in Australian Banking: An empirical investigation, Journal of Banking & Finance, March 2001. Vol. 25, No. 3, pp. 613-630. Staat, Matthias, Efficiency of hospitals in Germany: a DEA-bootstrap approach, Applied Economics, 2006, Vol. 38, No. 19. Shujie Yao, Chunxia Jiang, Genfu Fung and Dirk Willenbockel, WTO challenges and efficiency of Chinese banks.(World Trade Organization), Applied Economics, Vol. 39, No. 15. Steinmann, L. and Zweifel, P., On the (in)efficiency of Swiss hospitals, Applied Economics, 2003, Volume 35, No 10. Topuz, John and Isik, Ihsan, "Efficiency Change, Technological Regress and Total Factor Productivity Change of Real Estate Investment Trusts (REITs), 19891999", unpublished working paper, Southeastern Oklahoma State University, 2005. Vassiloglu, M., D. Giokas, A Study of the Relative Efficiency of Bank Branches: An Application of Data Envelopment Analysis, Journal of the Operational Research Society, (1990), Vol. 41, 591-597. Vennet, Rudi Vander, Cost and profit efficiency of financial conglomerates and universal banks in Europe, Journal of Money, Credit, and Banking, February 2002. Vol. 34, No.1, p. 254.

Von Hirschhausen, Christian, A. Cullmann, and A. Kappeler. "Efficiency analysis of German electricity distribution utilities--non-parametric and parametric tests, " Applied Economics. 2006, Vol. 38, No 21. Wheelock, David C., Wilson, Paul W., Technical progress, inefficiency, and productivity change in U.S. banking, 1984-1993, Journal of Money, Credit, and Banking, May 1999. Vol. 31, No. 2, p. 212.

Appendix I UAE Banking Sector Survey 2004 Serial Number: ………

Interviewer code: ……………

Date: … / … / 2004

Bank Title: ………………………………………………………………………… Number of years in operation Years Ownership Structure (% government participation) % Number of branches Branches Number of Employees by Gender Gender / Employees Males Females







Number of Employees by Ethnicity Ethnicity / Employees National Arab Asian Western Others Number of Employees by Education Education / Employees No education Primary (grades 1-6) Preparatory (grades 7-9) Secondary (grades 10-12) Two-years college University Higher degree (e.g., diploma, Master, Ph. D.)

Number of Employees by Experience Experience / Employees Lass than 5 years 5 to less than 10 years 10 to less than 15 years 15 years and more



Total Wages Employees Managerial Rest

Total wages

Department of Information Technology (IT) Number of employees in IT department Percentage of IT department’s operating budget to total operating budget Financial Services Offered Services Branch banking ATM EFTPOS Credit cards Telephone banking PC banking Internet banking Others (specify)

Yes / No

Financial Products Offered (to attract new deposits) Products Special accounts Certificate of Deposits (CD) Others (specify)


Bank Main Activities Item Book value of fixed assets and premises Total spending on fixed assets and premises Sum of current accounts and savings deposits Total interests on current accounts and savings deposits Sum of real estate loans Sum of commercial loans Sum of industrial loans Sum of consumer loans Sum of other loans (if any) Total investments (all security other than those held in trading accounts) Percentage of foreign transactions to total transactions Percentage of foreign assets to total assets


Off Balance Sheet Activities Item Swaps Future contracts Others (specify) Thank you very much for your time


Appendix II Regression Model 1 A logistic regression analysis was performed, through a backward stepwise elimination option, to relate the dependent variable: cost efficiency (y) and the independent variables in equation (4). The following model was obtained: ⎛ y ⎞ ⎟⎟ = 51.111 − 0.337 X 1 + 0.286 X 3 − 59.692 X 5 + 19.286 X 6 − 14.814 X 9 . ln⎜⎜ ⎝1− y ⎠ For Model 1, R2 = 0.784 and the standard error of the estimate = 0.284. The regression coefficients, their standard errors and the corresponding t and p-values are given in the following table: Table II-1: Logistic Regression Coefficients for Model 1 Std. Coeff.


Std. Error 14.777




.286 59.692

















Constant Numbers of years in operation (X1) Number of total branches (X3) Percentage of males (X5) Percentage of managerial employees (X6) Percentage of employees with short experience (X9)

t 3.459 4.336 3.681 3.406

pvalue .013 .005 .010 .014

Regression Model 2 A second logistic regression analysis was performed through a forward stepwise option, to relate the dependent variable: cost efficiency (y) and the independent variables in equation (4). The following model was obtained: ⎛ y ⎞ ⎟⎟ = 5.305 − 0.91 X 1 − 13.295 X 7 . ln⎜⎜ ⎝1− y ⎠ For Model 2, R2 = 0.563 and the standard error of the estimate = 0.330. The regression coefficients, their standard errors and the corresponding t and p-values are given in the following table:

Table II-2: Logistic Regression Coefficients for Model 2 Variable Coeff. Std. Error Std. Coeff. t p-value Constant 5.305 1.541 3.444 .007 Percentage of nationals (X7) -13.295 4.258 -.723 -3.122 .012 -.091 .041 -.521 -2.247 .051 Numbers of years in operation (X1)