Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 Cost Efficiency in Emerging Economies: Evidence from the MENA Banking Sectors Idries Mohammad Al-Jarrah and Khalid Shams Al-Abdulqader This study utilizes the Translog stochastic frontier model to estimate cost efficiency levels for banking sectors in the Middle East and North Africa (MENA) countries over the 2007-2013 period. The stake of the banks under study in terms of their assets holdings in the banking sectors of MENA countries is about 80 percent, comprising a sample of 1308 observations over the study period. The estimated cost efficiency averaged around 77 percent for the banks in MENA region with slight changes to this score over the period 2007-2013. This suggests that the same level of output could be produced with approximately 77% of current inputs if banks under were operating on the most efficient frontier. The results show that the banks operating in the Gulf Cooperation Council (GCC) countries are the most efficient while those operating in countries that have witnessed political instability are the least. The results also show that the efficiency scores for commercial and Islamic banks are similar while those for Investment are less. Besides, the efficiency scores are positively correlated with the bank size, the relative concentration in the market, the measure of asset quality, and with the bank capital adequacy ratio. In addition, the recent financial crisis seems to have slight impact on the observed efficiency scores of the banks under study. Finally, though many of non-oil producing countries under study have implemented programs of financial reforms to enhance the performance of their banking sectors, these may still need further reforms to catch-up with the level of efficiency observed in oil-producing countries. Keywords: Cost efficiency, Translog Form, Stochastic Frontier, Intermediation Approach JEL Classifications: G21, G28, G32, C52, F23 1. Introduction The aim of this study is to investigate the efficiency levels in the banking sectors of MENA countries over the 2007-2013 period. The empirical evidence on bank efficiency aims to highlight the features associated with the major events that faced the banking sectors in these countries including the global financial crisis in 2009 and onward, high instability in oil prices and observing the role of economic development and financial reforms that have taken place in many of these countries over the last decade. At the beginning, it is worth saying that the financial sectors in developing countries, including those in the MENA region have traditionally been characterised by relatively high levels of government controls where regulatory authorities maintained a protected banking environment that inhibited competition. However, market conditions in banking have undergone extensive changes over the last two decades or so. On the demand side, customer preferences have changed substantially, becoming more sophisticated and price conscious. On the supply side, the globalisation of financial markets has been ______________________________________________________________________ Prof. Idries Mohammad Al-Jarrah, Qatar University, Doha 2713 Qatar, email: Idries@qu.edu.qa, Work Tel.: +974 44036491 Dr. Khalid Shams Al-Abdulqader, Qatar University, Doha 2713 Qatar, email: Khld-shams@qu.edu.qa, Work Tel.: +974-44035088 1 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 accompanied by governmental deregulation, financial innovation and automation. As such, banks are now faced with competition from banks and non-bank institutions, and this also accentuates competition within the banking and financial services sector overall. To assist banks in confronting the new challenge, various MENA countries have passed a substantial body of legislation (over the last few years) aimed at liberalising their financial systems through the reduction of direct government control, at the same time, it is associated with upgrades of prudential regulations (Al-Jarrah, 2003). In this context, among the recent studies that addressed the impact of financial deregulation on bank efficiency is that of Kumar (2013) who analyses the trends of cost efficiency across Indian public sector banks (PSBs) during the post-deregulation period from 1992/93 to 2007/2008. The results reveal substantial cost inefficiencies of about 25.6% for Indian PSBs and the analysis shows the presence of convergence in cost efficiency levels during the post-deregulation years. As for the more broad studies that addressed the efficiency in emerging countries, Oluitan (2014) used the translog functional form to examine the cost efficiency of banking sectors in forty seven African countries over ten years period. The estimated levels of inefficiency range from about 10 to 26 percent for the financial sectors under study. The author argues that much of the inefficiency for countries under study are attributed to poor intermediation and low skilled staff in the banking sectors under study. Given the observed inextricable link between financial liberalisation and efficiencies, it is therefore interesting to highlight the impact of economic and financial reforms in various MENA countries on the efficiency levels of the financial institutions operating in these countries. Furthermore, our study period embraces the global financial crisis in 2007 and on and high instability in oil prices; and we expect these events to have impact on the performance of the banking sectors in various MENA countries. On the other hand, despite the extensive literature that has examined productive 2 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 efficiency, especially in the US banking system and other European markets, empirical research on financial sectors in developing countries, including MENA region, is much less. Besides, there are other motives for examining efficiency levels in MENA banking systems. First, such a study should help assessing the impact of the economic and financial reforms that have taken place in the countries under study. Furthermore, this study aims to provide empirical evidence about efficiency differences across various MENA banking industries (and across various types of financial institutions operating in these countries such as commercial, investment and Islamic banks). In addition to this introduction, this study is structured into five sections. Section 2 describes features of the banking sectors in MENA countries. Section 3 presents the study sample and methodology. Section 4 presents the empirical results. Finally, Section 5 summarizes the findings and gives the conclusion. 2. Features of the Banking Sectors in MENA countries The total assets of the banking sectors in MENA region has grown from about $1,510 billion in 2007 to about $2,327 billion in 2013; with an annual growth of almost 8 percent over this period. The largest banking sector in the region are those of Saudi Arabia, United Arab Emirates (UAE), Iran and Kuwait whose total shares of the region banking sectors is about 58 percent over the study period. The banking sectors in MENA region are generally characterized by high concentration, in terms of the share of the largest three banks of total assets in their respective banking sectors. The 3-firm concentration ratio (3-CR) averaged 56 percent for MENA banking sectors, with insignificant changes for this share over the period of 2007-2013 (Table 1). On the other hand, the individual bank market share (MS) in its respective market averaged about 8 percent over the study, with insignificant changes to this share over time (Table 1). Both the 3-CR ratio and the market share indicate that the MENA banking 3 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 sectors are rather concentrated markets and this might affect negatively the competitiveness and the performance of the banking sectors in these countries. As for the quality of loan portfolios, as measured by the ratio loan-loss reserves as a percent of total loans, in the MENA banking sectors, this ratio averages around 6 percent with high dissimilarities among the banks over the study (Table 1). On the other hand, the capital adequacy ratio, measured by the ratio of equity capital as percent of total assets, this ratio average 13 percent for the countries in the region (Table 1). Both the asset quality and capital adequacy ratios show high dissimilarities for banks operating in MENA region countries. Thus, the ability of the MENA banking sectors to stand potential negative shocks that might strike these sectors will not be symmetrical. As for the efficiency in utilization of assets of banking sectors in MENA region, in term of the return on assets (ROA), the banking sectors of GCC countries and the banking sector of Algeria seem to be the most efficient given the return on asset in these countries average more than 1.5 percent over the study period (Table 1). As for the profitability of the MENA region banking sectors, as measured by return on equity (ROE), the ROE has witnessed some decline from around 16 percent in 2007 to about 12 percent in 2013 (Table 1). Both the ROA and ROE ratios show clearly that the profitability of banking sectors vary substantially based on the geographical location; both indicates that the most profitable sectors are those of GCC countries over the study. As for the distribution of banks according to specialization in MENA countries, it is clear that the commercial banks dominate the banking sectors, as measured by the ratio of their asset holding as percent of the total banking sector assets in their respective sectors. The share of the commercial banks exceeds 90 percent in many countries over the period 2007 to 2013. On the other hand, the Islamic banks dominate the banking 4 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 sector in Iran as it seems to be not allowed for other non-Islamic based banks to conduct business there. Finally, the investment seem to have relatively substantial role in some MENA banking sectors including those of Malta, Kuwait, Palestine, Oman and Jordan where the share of these banks, in terms of the assets holdings in their respective banking sectors, exceeded 5 percent. To conclude, this section shows that the size of the banking sectors in various MENA countries vary considerably and the banking sectors of oil-producing GCC countries, Iran and Egypt are the largest. The banking sectors in MENA countries region seem to be highly concentrated as measured by the individual bank market share and by the 3-CR in these markets which signal worries about the competitiveness of these sectors . The asset quality of the banking sectors in MENA region countries vary substantially and this applies to the observed capital adequacy ratios in these countries over the period 20072013. The profitability measures display different trends at country level, as measured by ROA, ROE and NIM and again the banks operating in GCC countries seem to be the most profitable. 3. The Study Sample and Methodology This study employs the Translog functional form to derive measures of cost efficiency for banking sectors operating in MENA region countries over the 2007-2013 period. The stochastic frontier methodology was proposed by Gallant (1981, 1982), discussed later by Elbadawi et al. (1983), Chalfant and Gallant (1985), Eastwood and Gallant (1991) and applied to the analysis of bank cost efficiency by Spong et al. (1995), Mitchell and Onvural (1996) and Berger et al. (1997), among many others. The stochastic cost function for a sample of N firms can be written as: lnTCi = lnTC(yi ,wi ,di ;B)+ui +vi, , i = 1, ..., N, 5 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 where TCi is the observed cost of bank i, y i is a vector of output, wi is a vector of input prices, di a vector of control variables that include firm-specific (and country-specific variables). B is a vector of parameters to be estimated and vi is a two-sided error term representing the statistical noise. u it represents non-negative variables that account for inefficiency and are assumed to be independently and identically distributed (iid) as truncations at zero of the N( it d, 2uit ) distribution; where mi = di where di are the control variables and is a vector of their parameters. In the light of these concerns, the Translog functional form for the cost function has been specified as: 2 2 1 2 2 B ln( w / w ) ln(w j / w3 ) ln(C / w ) B ln( w / w ) ln y 3 i i 3 k k 2 ij i 3 i 1 k 1 i=1 j=1 9 3 3 2 3 1 ln y ln y ln( w / w ) ln( y ) di uit vit k m ik i 3 k 2 k 1 m 1 km i 1 i 1k 1 where lnTC is the natural logarithm of bank’s total costs (operating and financial); ln yi is the natural logarithm of bank outputs (i.e. loans and other earnings assets); ln wi is the natural logarithm of ith input prices (i.e. interest rate, wage rate and cost of financial capital); di are set of control variables and , , , , and are coefficients to be estimated. Before running the maximum-likelihood to get estimate for parameters in the Translog stochastic frontier, we employed panel Ordinary Least Squares (OLS) regression with various combination of control variables to check the need to add these variables in final model that employed to derive efficiency measures. All the specified models, based on Ftest that enables comparing various nested models’ specifications, Wald test that check for restricting the coefficients to zero and that likelihood ratio that allow examining the redundancy in the specified model indicate that the model should include all the control 6 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 variables (only the best specified model is reported, Table 2). The sample of this study comprises data for banks operating in the financial systems of MENA countries over the 2007-2013. The share of the banks included in our study in terms of their assets holdings in their respective banking sectors is about 80 percent over the period 2007-2013 (Table 3). This study employs the intermediation approach for defining bank inputs and outputs. The bank inputs include deposits, labour and financial capital (k) while the output are loans (y1) and other earning assets (y2), measured in millions of US dollars. In line with Aly et al. (1990), the cost of deposit (w1) is derived by taking the total interest expense on various types of deposits and other banks borrowing as percent of total deposits. As staff numbers were not available for the banks in the sample, the cost of Labour (w2) is measured by personnel expenses as a percent of total earning assets minus loans (given loans are used in calculation of w1). The cost of financial capital (w3) is measured by the return on equity (ROE). The description, means, standards of deviation of the input and output variables are reported in table 4. In addition to the input and output variables, we include nine control variables to the cost function to rule out the effect of other factors that might explain the observed dissimilarities in efficiency scores for the banks under study. These variables include measures for asset quality (S), scale economies (SE), financial capital (k), liquidity risk (L), 3-firm concentration ratio (3-CR), market share (MS), bank specialization (Sp), time trend (T) and a country indicator variable (Cn) (Table 5 shows descriptive statistics of the control variables). 4. Study Results The cost efficiency scores for banks in the MENA countries based on our sample averaged around 77 percent over the 2007-2013 period, with slight changes to this score over the study period (Table 6). This efficiency level suggests that the same level of 7 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 output could be produced with approximately 77% of current inputs if banks under study were operating on the most efficient frontier. This level of inefficiency is somewhat higher than the range of 10-15% for the 130 studies surveyed by Berger and Humphrey (1997) and Berger and DeYoung (1997). These inefficiency scores are somehow also comparable to the level of inefficiency observed in European studies including Carbo et al.’s (2000) whose findings for a sample of banks, from twelve countries, show mean cost inefficiency of around 22 % for the period 1989 to 1996. Finally, these levels of efficiency are comparable to more recent studies of Kumar (2013) on Indian banking sectors and Oluitan (2014) on African banking sectors. As for the efficiency scores based on the geographical location, the banks operating in Gulf Cooperation Council (GCC) countries seem to be the most cost efficient as the efficiency scores for these countries ranges from about 83 percent in UAE to 99 percent in Qatar, on average, over the study period. On the other hand, the least efficient banks are those operating in countries that have witnessed or witnessing political instability including those of Lebanon, Yemen, Palestine and Syria. The trend of cost efficiency scores over time for the countries included seem to be highly correlated, which most probably attributed to adverse impact of latest financial crisis. For instance, the efficiency scores for the banks operating in GCC countries have witnessed identical drift over time in a form of slight deterioration for their efficiency scores over the study period. As for the cost efficiency scores based on specialization, the efficiency scores for both commercial and Islamic banks are somehow analogous while the efficiency scores for Investment banks are slightly less. Thus, though it has been argued that the Islamic banks have cost advantage over other banks given part of the sources of funds for these banks carry no cost, our results do not find rich evidence that support this argument. This result may be attributed to intense competition between commercial and Islamic banks to 8 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 attain greater market share in the market. As for the impact of capital adequacy ratio, as measured by the bank owners’ equity as percent of its total assets, on the estimated efficiency scores, we observed that the efficiency scores are highly correlated with capital adequacy ratios. Specifically, the efficiency scores ranged from 68 percent for relatively undercapitalized banks with capital adequacy ratio of less than 6 percent to reach 80 percent for banks with capital adequacy ratio that ranges between 10 to 25 percent. The efficiency scores based on bank size, as measured by the total asset holdings, Table 10 shows that efficiency scores are positively correlated with bank size. The efficiency scores ranged from 68 percent for small banks whose assets holdings are less than $1 billion to 82 percent for banks whose asset holdings are $25 billion or more. Thus, large banks seem to be enjoying higher levels of efficiency, most probably due to the concurrent scale economies, market power and other favourable features including the ability of these banks to utilize more efficient technology with less cost, the ability of these banks to set up more specialized staff for the most profitable activities and the ability of these banks to provide better quality output and therefore charge higher prices. As for the impact of bank’s market share, as measured by the share of individual bank of total assets in the respective market, on the observed efficiency scores. Our results show that the banks with larger market shares are relatively more cost efficient than their counterparties banks. This finding supports our argument that large banks enjoy economies of scale and market power over other banks in their respective market. The movements in trends of cost efficiency for the banks under study seem to be highly correlated over time; as the negative or positive shocks appear to affect banks of various sizes in a similar way. 9 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 On the other hand, the efficiency scores based on the share of the largest three banks of total assets (3-CR) in their respective banking market, the results show that the efficiency scores is positively associated with the increase in the concentration but this relationship does not hold for the most concentrated sectors in the study where the share of the largest three banks exceeds 70 percent. Thus, the increase in the concentration seems to be advantageous for most banks in the MENA region, probably due to the market power enjoyed by the largest banks operating in these markets. As for the impact of asset quality (measured by the ratio of loan reserve as percent of total loans - LLR ratio) on the efficiency scores for the banks under study, the results show that the efficiency scores are positively correlated with better asset quality. Specifically, the efficiency scores range from 66 percent for banks with the lowest asset quality (highest LLR ratio) to about 79 percent for the banks with the highest asset quality. Finally, as for the impact of liquidity ratio, as measured by the ratio of cash and cash due with other banks as percent of total assets, on the efficiency scores, the results show that more liquid banks are relatively less cost efficient. To sum up, the average cost efficiency is 77 percent for the MENA countries over the 2007-2013 period, with slight decline for this score over time. The efficiency scores for both commercial and Islamic banks are comparable and better than that of the Investment. The efficiency scores for oil-producing countries mainly the GCC countries are better than the efficiency scores observed in the other countries especially the countries that have witnessed or witnessing political and economic instability including Palestine, Yemen and Syria. The observed efficiency scores for the banks in the MENA are highly correlated with the market share, the concentration in the market and the bank size. Furthermore, the efficiency scores for overcapitalized banks with low ratio of loan10 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 loss reserves to total assets are much better than the scores observed for undercapitalized banks with inferior asset quality. Finally, the liquidity is found to be negatively correlated with efficiency scores for the banks under study. Thus, while many of non-oil producing countries under study have implemented programs of financial and economic reforms with the aim of upgrading the performance of their banking sector in the economic development, these may still need further geared reforms to catch-up with the level of efficiency observed in relatively rich oil-producing countries including those of GCC countries. We argue that the observed cost efficiency scores are negatively correlated with political instability, where the latter has heavily (and negatively) affected the performance of the banks in these countries. In this regard, the political conflicts in various countries including Libya, Syria, Yemen and Iraq might contribute to the observed level of efficiency for the banks operating in these countries. Furthermore, the instability in the oil prices has negative impact on the observed efficiency levels in some oil-importing countries including those of Jordan, Morocco, Egypt, Tunis and Lebanon. On the other side of MENA countries, the move to create single GCC market may positively help the banks operating in these countries, in a similar style to that of countries in European Monetary Union (EMU) which intensified the competition and result in a better observed efficiency for the banks in these countries (see European Commission (1997)). We expect that the operation of a single banking market in the GCC to force greater merger and acquisition activities in the financial sector that could result in the realization of larger economies and greater efficiency for their banking and financial systems. Finally, we observed that the efficiency scores for banks under study are slightly affected by the latest financial crisis, in general. 5. Conclusion This study employs the cost function with stochastic frontier to estimate the cost efficiency 11 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 levels for banking sectors operating in MENA region countries over the period 2007-2013. The share of the banks included in the study in terms of their assets holdings in their respective banking sectors is about 80 percent, comprising an unbalanced sample of 1308 observations over the same period. The intermediation approach is used for defining bank inputs and outputs. The bank inputs include deposits, labour and financial capital (k) while the output are loans (y1) and other earning assets (y2), measured in millions of US dollars. In addition to the input and output variables, we include a nine control variables to the cost function to rule out the effect of other factors that might explain the observed differences among efficiency scores for the banks under study. The maximum likelihood estimates for the parameters of the Translog function form that include control variables is estimated using Battese and Coelli’s (1995) model. Before deriving the maximum-likelihood estimate for parameters in the Translog stochastic frontier, we employed panel OLS regression with various combinations of control variables to check the necessity to add these variables in our well-specified model that used to derive efficiency measures. The various specified models we estimated, based mainly on Wald test and the likelihood ratio indicate that the best specified model is the one that include all the control variables. The cost efficiency scores for banks in the MENA countries based on our sample averaged around 77 percent over the 2007-2013 period, with slight changes to this score over the study period. This suggests that the same level of output could be produced with approximately 77% of current inputs if banks under study were operating on the most efficient frontier. This level of inefficiency is somewhat higher than the range of 10-15% for the 130 studies surveyed by Berger and Humphrey (1997) and Berger and DeYoung (1997). These inefficiency scores are somehow also comparable to the level of inefficiency observed in European studies including Carbo et al.’s (2000) whose findings 12 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 for a sample of banks, from twelve countries, show mean cost inefficiency of around 22 % for the period 1989 to 1996. Furthermore, our inefficiency results for banking sectors in MENA region countries are comparable with more recent studies including those of Kumar (2013) on Indian banks and Oluitan (2014) on African banking sectors. As for the efficiency scores based on the geographical location, the banks operating in Gulf Cooperation Council (GCC) countries seem to be the most cost efficient while the least efficient banks are those operating in countries that have witnessed or witnessing political instability. As for the cost efficiency scores based on specialization, the efficiency scores for both commercial and Islamic banks are analogous while the efficiency scores for Investment are less. This result may be attributed to intense competition between commercial and Islamic banks. The derived efficiency scores are found to be positively correlated with the bank’s size; the latter reflect the ability of relatively large banks to benefit from the concurrent scale economies, market power and other favourable features including the ability to utilize more efficient technology with less cost. The results also show that the efficiency scores is positively associated with the increase in the bank’s market share and the degree of concentration in the market; probably due to the market power enjoyed by the largest banks operating in these markets. The observed efficiency scores are found to be positively correlated with superior asset quality but it were negatively correlated with the increase in the bank’s liquidity. Furthermore, the efficiency scores for overcapitalized banks with low ratio of loan-loss reserves to total assets are much better than the scores observed for undercapitalized banks with inferior asset quality. To sum up, the efficiency scores for oil-producing countries mainly the GCC countries are better than the efficiency scores observed in the other countries especially the MENA countries that have witnessed or witnessing political and economic instability. 13 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 Furthermore, while many of non-oil producing countries under study have implemented programs of financial and economic reforms with the aim of upgrading the performance of their banking sector and their roles in the economic development, these may still need further geared reforms to catch-up with the level of efficiency observed in relatively rich oil-producing countries. The move to create single GCC market may positively help the banks operating in these countries, in a similar style to that of countries in European Monetary Union (EMU) which intensified the competition and result in a better observed efficiency for the banks in these countries (European Commission (1997)). Finally, we argue that the observed efficiency scores are slightly affected by the latest financial crisis but are heavily affected by the political instability in various MENA region countries under study. References 1. Al-Jarrah, I (2002). ‘Efficiency in Arabian Banking’, Ph.D. thesis, University of Wales (Bangor). 2. Aly, H., R. Grabowski, C. Pasurka, and N. Rangan (1990). ‘Technical, Scale, and Allocative Efficiencies in U.S. Banking: An Empirical Investigation’, ‘The Review of Economics and Statistics’, 72, pp. 211-218. 3. Battese, G. and T. Coelli (1995). ‘A Model for Technical Inefficiency Effects in a Stochastic Frontier Production Function for Panel Data’, ‘Empirical Economies’, 20, pp. 325-332. 4. Berger, A. and D. Humphrey (1997). ‘Efficiency of Financial Institutions: International Survey and Directions for Future Research’, ‘European Journal of Operational Research’, 98, pp. 175-212. 5. Berger, A. and R. DeYoung (1997), ‘Problem Loans and Cost Efficiency in 14 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 Commercial Banks’, ‘Federal Reserve Board, Washington, D.C.’, (WP 8/1997). 6. Carbo, S., E. Gardener, and J. William (2000). ‘Efficiency in Banking: Empirical Evidence from the Savings Banks Sector’, ‘Institute of European Finance, University of Wales, Bangor, Un-published’. 7. Chalfant, J. and A. Gallant (1985). ‘Estimating substitution elasticities with the Fourier cost function’, ‘Journal of Econometrics’, 28, pp. 205-222. 8. Eastwood, B. and A. Gallant (1991). ‘Adaptive rules for semi-nonparametric estimators that achieve asymptotic normality’ ‘Economic Theory’, 7, pp. 307-340. 9. Elbadawi, I., A. Gallant and G. Souza (1983). ‘An elasticity can be estimated consistently without a priori knowledge of functional form’, ‘Econometrica’, 51, pp. 1731-1753. 10. European Commission (1997). ‘The Single Market Review’ ‘Credit Institutions and Banking, Subseries II: Impact on Services’, Vol. 3, (London: Kogan Page). 11. Gallant, A. (1981). ‘On the bias in Flexible Functional forms and essentially unbiased form: The Fourier Flexible form’, ‘Journal of Econometrics’, 15, 211-245. 12. Gallant, A. (1982). ‘Unbiased determination of production technologies’, ‘Journal of Econometrics’, 20, pp. 285-324. 13. Mitchell, K. and N. Onvural (1996). ‘Economies of Scale and Scope at Large Commercial Banks: Evidence from the Fourier-Flexible Functional Form’, ‘Journal of Money, Credit and Banking’, 28(2), pp. 178-199. 14. Oluitan, R. (2014). ‘Bank Efficiency in Africa’, The Journal of Developing Areas, Vol. 48, No. 2, Spring. 15. Spong, K., R. Sullivan and R. DeYoung (1995). ‘What makes a bank efficient? A look at financial characteristics and bank management and ownership structure’, ‘Financial Industry Perspective, Federal Reserve Bank of Kansas City’. 15 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 Appendix Table 1: The Financial Indicators of the MENA Banking Sectors over 2007-2013, on Average Country/ Year 3-CR** MS** LLR/ Equity/A ROA** ROE** NIM** ** ** Loans ssets Algeria 0.70 0.07 0.01 0.19 0.020 0.13 0.046 Bahrain 0.51 0.07 0.03 0.14 0.014 0.12 0.024 Egypt 0.52 0.04 0.13 0.10 0.013 0.15 0.032 Iran 0.56 0.11 0.02 0.12 0.012 0.13 0.015 Iraq* 0.78 0.04 0.07 0.28 0.038 0.15 0.026 Jordan 0.68 0.07 0.05 0.14 0.013 0.1 0.034 Kuwait 0.51 0.10 0.05 0.11 0.015 0.12 0.027 Lebanon 0.44 0.04 0.08 0.08 0.009 0.13 0.024 Libya* 0.63 0.19 0.07 0.11 0.008 0.10 0.038 Malta 0.70 0.16 0.02 0.09 0.011 0.13 0.024 Morocco 0.57 0.09 0.03 0.09 0.014 0.16 0.029 Oman 0.64 0.14 0.05 0.12 0.019 0.17 0.035 Palestine 0.87 0.23 0.02 0.16 0.010 0.08 0.047 Qatar 0.68 0.11 0.02 0.18 0.025 0.15 0.03 Saudi 0.44 0.08 0.03 0.16 0.019 0.14 0.031 Syria 0.55 0.15 0.07 0.09 0.007 0.09 0.032 Tunis 0.37 0.06 0.10 0.13 0.013 0.12 0.025 UAE 0.49 0.05 0.05 0.16 0.020 0.13 0.035 Yemen 0.71 0.22 0.28 0.09 0.012 0.13 0.042 Average 0.56 0.08 0.06 0.13 0.015 0.13 0.03 Source: Adapted from Bankscope – Bureau van Dijk, World Banking Information Source. *The size of the sample drawn from Libya and Iraq is less than 70 percent, in their respective market and thus these will be excluded from our discussion as the samples might not be representative. The 3-CR is the 3-firm concentration ratio as measured by the share of the largest three banks in the total assets of the banking sector in the respective market, MS is the share of individual bank, on average, in the total assets of the banking sector in the respective market, LLR/Loans is the loan-loss reserve as percent of total loans in the respective market, Equity/Assets is the ratio of total owners’ equity as percent of total assets in the respective market, ROA is the average ratio of net income as percent of total owners’ equity in the respective market, ROE is the ratio of net income as percent of total owners’ equity in the respective market, and NIM is the net interest margin as measured by the ratio of net interest income as percent of total earning assets in the respective market. 16 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 Table 2:Maximum Likelihood parameter estimates and Ordinary Least Squares for the Study Model for MENA banks using Translog functional form Maximum Likelihood Parameter Estimates Panel OLS Estimates Symbol (in Variables (all are coefficient t-ratio Coefficient P-value the model) logged) (0.00) (0.01) 10.83 0.24 lny1 0.52 6.65 0.63 0.00 *** lny2 0.62 7.32 1.61 0.00 *** Lnw1/w3 (0.18) (1.64) 4.46 0.00 *** lnw2/w3 0.67 5.79 (1.11) 0.32 lny1lny1 0.06 12.48 (5.97) 0.00 *** lny1lny2 (0.12) (11.57) 5.04 0.00 *** lny1lnw1/w3 0.02 1.64 (0.28) 0.01 *** lny1lnw2/w3 0.00 0.17 0.01 0.96 lny2lny2 0.06 10.57 0.17 0.16 lny2lnw1/w3 0.04 3.17 (0.43) 0.00 *** lny2lnw2/w3 0.00 0.06 0.10 0.34 lnw1/3lnw1/w3 (0.08) (15.47) 0.42 0.01 *** lnw1/w3lnw2/w3 0.00 5.87 (0.22) 0.12 lnw2/w3lnw2/w3 0.00 1.98 (0.49) 0.00 *** (LLR/ Loans) (S) 0.93 6.21 0.00 0.00 *** (Liquid asset/ 1.88 16.99 (0.00) 0.01 *** Asset) (L) (Ln(TA)) (SE) 0.01 3.25 (3.15) 0.06 * (OE/TA)(K) (0.60) (4.76) 1.27 0.46 Specialization (Sp) (0.00) (0.03) (3.85) 0.00 *** 3-FCR (0.30) (3.11) 11.07 0.00 *** MS (0.24) (1.73) (10.26) 0.00 *** time trend (T) (0.03) (5.40) 0.63 0.00 *** Country (Cn) 0.02 5.94 11.56 0.00 *** sigma-square 0.13 24.50 gamma 0.00 0.01 Sigma-squared 0.00 R-squared 0.99 Sigma-squared (v) 0.13 Adjusted R-squared 0.99 Lamda 0.00 S.E. of regression 5.18 Contribution of the inefficiency 0.00 Sum squared resid 34,481 effect/ T. variance Log likelihood function (535.71) Log likelihood (3,996) [this statistic has a mixed chi-squared distribution] F-statistic 7,475 Prob(F-statistic) 0.00 Source: Authors’ own estimation. *, **, and *** means significant at 10%, 5% and 10% respectively. Table 3: Share of the Banks in their respective markets according to specialization Country/ Size of Sample (% of No. of Obs. Over Commercial Islamic 17 Investment Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 Year T. Assets) 2007-13 banks banks banks Algeria 0.83 81 1.00 0.00 0.00 Bahrain 0.59 61 0.74 0.26 0.00 Egypt 0.86 134 0.96 0.03 0.01 Iran 0.97 63 0.00 1.00 0.00 Iraq* 0.21 40 0.39 0.17 0.44 Jordan 0.99 102 0.86 0.07 0.07 Kuwait 0.81 55 0.64 0.25 0.11 Lebanon 0.9 159 0.99 0.00 0.01 Libya* 0.38 14 1.00 0.00 0.00 Malta 0.75 33 0.42 0.00 0.58 Morocco 0.84 67 0.99 0.00 0.01 Oman 0.86 42 0.92 0.00 0.08 Palestine 0.98 30 0.69 0.23 0.08 Qatar 0.99 65 0.82 0.18 0.00 Saudi 0.96 81 0.83 0.17 0.00 Syria 0.77 36 0.84 0.16 0.00 Tunis 0.78 99 0.94 0.02 0.04 UAE 0.93 123 0.84 0.16 0.00 Yemen 0.72 23 0.45 0.55 0.00 Average 0.80 1308 Source: Adapted from Bankscope – Bureau van Dijk, World Banking Information Source *The size of the sample drawn from Libya and Iraq is less than 70 percent, in their respective market and thus these will be excluded from our discussion as the samples might not be representative. Table 4: Descriptive statistics of the total costs, inputs and outputs for banks in MENA region over 2007-2013 Variables Description Mean St. Dev. Min. Max. TC ($Millions) Total cost of banks includes 314 547 0 6,285 both operating and financial costs (US$ millions). Bank Output y1 Loans ($Millions) The US $ value of total 5,876 9,745 0 85,360 aggregate loans (all types of loans) (US$ millions). y2 Other Earning Assets ($Millions) The US $ value of other earning 3,380 5,222 1 37,780 assets (short-term investment, equity and other investment and public sector securities (US$ millions)). Cost of bank Inputs w1 cost of deposits Cost of customer deposits (%) 0.07 0.55 0.00 18.39 (total interest expense/ total customer deposits (demand, saving and time deposits)). w2 cost of labor Price of labour (%) ((total 0.01 0.02 0.00 0.36 personnel expense/(Earning Assets - Loans). w3 Cost of Financial capital Price of financial capital (%) 0.16 0.67 0.00 22.19 (Net Income/ Owners’ Equity). Source: Adapted from Bankscope – Bureau van Dijk, World Banking Information Source 18 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 Table 5: Descriptive statistics of the control variables for banks in MENA countries over 2007-2013 Control Variables Description Mean Standard Minimum Maximum Deviation Asset Quality (S) The ratio of loan loss reserves as 0.06 0.10 0.00 0.87 percent of total loans (LLR ratio) Scale Economies The value of total assets 10,824 16,086 30 121,837 (TA, US$ Millions) (SE) Capital Adequacy The ratio of owners' equity as 0.13 0.08 -0.01 0.90 (%) (K) percent of total assets Liquidity Risk (L) The ratio of cash and due with 0.12 0.14 0.00 0.95 other financial institutions as percent of total assets The 3-Firm the largest 3 banks total assets of 0.56 0.13 0.21 1.00 Concentration Ratio /Total assets of all banks in the (3-CR) bank country for the respective years Market share (MS) Bank assets market share (%) for 0.08 0.10 0.00 0.74 each year Specialization (Sp) Bank specialization; Commercial, 1.3 0.6 1.0 3.0 Islamic and Investment Time trend (T) Time trend for the years of the 4.0 2.0 1.0 7.0 study period Country code (Cn) Indicator variable for the 9.3 4.9 1.0 19.0 countries included in the study Source: Adapted from Bankscope – Bureau van Dijk, World Banking Information Source Table 6: Efficiency Scores for MENA Countries over 2007-2013 19 Proceedings of 10th Annual London Business Research Conference 10 - 11 August 2015, Imperial College, London, UK, ISBN: 978-1-922069-81-8 Country/ Year Qatar 2007 1.00 2008 1.00 2009 0.97 2010 1.00 2011 0.99 2012 0.99 2013 0.97 Average 0.99 No. of Obs. 65 Bahrain 0.92 0.92 0.94 0.94 0.92 0.94 0.94 0.93 61 Oman 0.92 0.96 0.98 0.87 0.94 0.86 0.85 0.91 42 Kuwait 0.89 0.86 0.84 0.82 0.85 0.88 0.86 0.86 55 Saudi 0.85 0.81 0.84 0.87 0.85 0.86 0.83 0.85 81 Morocco 0.91 0.83 0.85 0.87 0.83 0.77 0.73 0.83 67 UAE 0.83 0.83 0.81 0.83 0.85 0.83 0.81 0.83 123 Jordan 0.86 0.84 0.79 0.85 0.82 0.80 0.77 0.82 102 Malta 0.85 0.77 0.81 0.89 0.81 0.81 0.81 0.82 33 Algeria 0.79 0.81 0.77 0.82 0.79 0.79 0.78 0.79 81 Egypt 0.79 0.79 0.72 0.76 0.75 0.76 0.73 0.76 134 Tunis 0.74 0.78 0.74 0.75 0.69 0.71 0.67 0.73 99 Iran 0.74 0.73 0.72 0.70 0.69 0.72 0.66 0.71 63 Palestine 0.75 0.77 0.70 0.54 0.73 0.65 0.73 0.69 30 Lebanon 0.65 0.64 0.67 0.68 0.66 0.60 0.58 0.64 159 Yemen 0.72 0.67 0.58 0.61 0.56 0.53 0.63 0.61 23 Syria 0.54 0.53 0.55 0.58 0.66 0.58 0.49 0.56 36 Iraq* 0.29 0.41 0.44 0.41 0.38 0.40 0.45 0.40 40 Libya* 0.22 0.24 0.23 0.49 0.52 0.50 0.25 0.35 14 Average 0.78 0.77 0.76 0.78 0.77 0.76 0.74 0.77 1308 Source: Authors’ own estimation *The size of the sample drawn from Libya and Iraq is less than 70 percent, in their respective market and thus these will be excluded from our discussion as the samples might not be representative. 20