Estimating the technical and scale efficiency of

Research in International Business and Finance22(2008)301–318

Estimating the technical and scale ef?ciency of Greek commercial banks:The impact of

credit risk,off-balance sheet activities,

and international operations

Fotios Pasiouras?

University of Bath,School of Management,Bath BA27AY,United Kingdom

Received14April2007;received in revised form11September2007;accepted14September2007

Available online18September2007

Abstract

This paper uses data envelopment analysis(DEA)to investigate the ef?ciency of the Greek commercial banking industry over the period2000–2004.Our results indicate that the inclusion of loan loss provisions as an input increases the ef?ciency scores,but off-balance sheet items do not have a signi?cant impact. The differences between the ef?ciency scores obtained through the pro?t-oriented and the intermediation approaches are in general small.Banks that have expanded their operations abroad appear to be more technical ef?cient than those operating only at a national level.Higher capitalization,loan activity,and market power increase the ef?ciency of banks.The number of branches has a positive and signi?cant impact on ef?ciency, but the number of ATMs does not.The results are mixed with respect to variables indicating whether the banks are operating abroad through subsidiaries or branches.

?2007Elsevier B.V.All rights reserved.

JEL classi?cation:G21;C24;C67;D61

Keywords:Banks;DEA;Ef?ciency;Greece

1.Introduction

The Greek banking sector has undergone major restructuring in recent years.Important struc-tural,policy and environmental changes that are frequently highlighted by both academics and

?Tel.:+441225384297.

E-mail address:f.pasiouras@https://www.360docs.net/doc/4d3293899.html,.

0275-5319/$–see front matter?2007Elsevier B.V.All rights reserved.

doi:10.1016/j.ribaf.2007.09.002

302F.Pasiouras/Research in International Business and Finance22(2008)301–318 practitioners are the establishment of the single EU market,the introduction of the euro,the inter-nationalization of competition,interest rate liberalization,deregulation,and the recent wave of mergers and acquisitions.

The Greek banking sector has also experienced considerable improvements in terms of com-munication and computing technology,as banks have expanded and modernized their distribution networks,which apart from the traditional branches and ATMs,now include alternative distri-bution channels such as internet banking.As the Annual Report of the Bank of Greece(2004) highlights,Greek banks have also taken major steps in recent years towards upgrading their credit risk measurement and management systems,by introducing credit scoring and probability default models.Furthermore,they have expanded their product/service portfolio to include activities such as insurance,brokerage and asset management,and at the same time increased their off-balance sheet operations and non-interest income.

Finally,the increased trend towards globalization that focused on the wider market of the Balkans(e.g.Albania,Bulgaria,FYROM,1Romania,Serbia)has added to the previously limited international activities of Greek banks in Cyprus and USA.The performance of the subsidiaries operating abroad is expected to have an impact on the performance of parent banks and conse-quently on future decisions for further internationalization attempts.

The purpose of the present study is to employ data envelopment analysis(DEA)and re-investigate the ef?ciency of the Greek banking sector,while considering several of the issues discussed above.We therefore differentiate our paper from previous ones that focus on the Greek banking industry2and add insights in several respects,discussed below.

First of all,we examine for the?rst time the impact of credit risk on the ef?ciency of Greek banks by including loan loss provisions as an additional input as in Charnes et al.(1990),Drake (2001),Drake and Hall(2003),and Drake et al.(2006)among others.As Mester(1996)points out“Unless quality and risk are controlled for,one might easily miscalculate a bank’s level of inef?ciency;e.g.banks scrimping on credit evaluations or producing excessively risky loans might be labelled as ef?cient when compared to banks spending resources to ensure their loans are of higher quality”(p.1026).We estimate the ef?ciency of banks with and without this input to adjust for different credit risk levels and examine its impact on ef?ciency.

Second,unlike previous studies in the Greek banking sector,we consider off-balance sheet activities during the estimation of ef?ciency scores.Several recent studies that examine the ef?ciency of banks,with DEA or stochastic frontier techniques,acknowledge the increased involvement of banks in non-traditional activities and include either non-interest(i.e.fee)income (https://www.360docs.net/doc/4d3293899.html,ng and Welzel,1998;Drake,2001;Tortosa-Ausina,2003)or off-balance sheet items(e.g. Altunbas et al.,2001;Altunbas and Chakravarty,2001;Isik and Hassan,2003a,b;Bos and Colari, 2005;Rao,2005)as an additional output.However,despite their increased importance for Greek banks,such activities have not been considered in the past.Again,we estimate the ef?ciency of the banks in our sample with and without off-balance sheet activities to observe whether it will have an impact on ef?ciency.

Third,we compare the results obtained from the intermediation approach that has been followed in most recent studies of banks’ef?ciency with a pro?t-oriented approach that was recently proposed by Drake et al.(2006)in the context of DEA,and is in line with the approach of Berger 1Former Yugoslav Republic Of Macedonia.

2Previous studies that focus on the ef?ciency of the Greek banking sector are:Karafolas and Mantakas(1996),Noulas (1997),Christopoulos and Tsionas(2001),Christopoulos et al.(2002),Tsionas et al.(2003),Halkos and Salamouris (2004),Apergis and Rezitis(2004)and Rezitis(2006).These studies are discussed in more detail in the next section.

F.Pasiouras/Research in International Business and Finance22(2008)301–318303 and Mester(2003)in the context of their stochastic frontier approach.This allows us to observe if different input/output de?nitions affect ef?ciency scores.

Fourth,we compare the ef?ciency scores of Greek banks that have expanded their operations abroad(i.e.international Greek banks,hereafter IGBs),with those of Greek banks whose oper-ations are limited in the domestic market3(i.e.purely domestic banks,hereafter PDBs).To the best of our knowledge,no study has undertaken such an analysis for Greece.However,in a study of the Turkish banking sector,Isik and Hassan(2002)found evidence that multinational domestic banks are superior to purely domestic banks in terms of all ef?ciency measures(i.e.cost ef?ciency, allocative ef?ciency,technical ef?ciency,pure technical ef?ciency)except for scale ef?ciency. The conclusions drawn from our study could be useful to the managers of Greek banks or other medium-sized banking sectors that are considering the internationalization of their operations.

Fifth,we run regressions to explain the ef?ciency of banks,an approach that has been followed in only two of the past studies in Greece(Christopoulos et al.,2002;Rezitis,2006).However,in our case we examine a most recent period that follows the numerous changes outlined above.

The rest of the paper is as follows:Section2reviews the literature that focuses on the ef?ciency of the Greek banking sector.Section3provides a brief discussion of DEA.Section4presents the data and variables.Section5discusses the empirical results,and Section6concludes the study.

2.Literature reviews

Karafolas and Mantakas(1996)use a second-order translog cost function to estimate(for the?rst time)an econometric form of the costs in the Greek banking sector and investigate economies of https://www.360docs.net/doc/4d3293899.html,ing data for eleven banks from the period1980to1989,they?nd that although operating-cost scale economies do exist,total cost scale economies are not present. Participation of the dataset in sub-samples by banks’size(https://www.360docs.net/doc/4d3293899.html,rge and small banks)and time periods(i.e.1980–1984,1985–1989)has not altered the results.Finally,the results indicate that technical change has not played a statistically signi?cant role in the reduction of average cost.

Noulas(1997)examines the productivity growth of ten private and ten state banks operating in Greece during1991and1992,using the Malmquist productivity index and DEA to measure ef?ciency.The author follows the intermediation approach and?nds that productivity growth aver-aged about8%,with state banks showing higher growth than private ones.The results also indicate that the sources of the growth differ across the two types of banks.State banks’productivity growth is a result of technological progress,while private banks’growth is a result of increased ef?ciency.

Christopoulos and Tsionas(2001)estimate the ef?ciency in the Greek commercial bank-ing sector over the period1993–1998using homoscedastic and heteroscedastic frontiers.They ?nd an average technical ef?ciency about80%for the heteroscedastic model and83%for the homoscedastic one.They also?nd that both technical and allocative inef?ciencies decrease over

3One could argue that the IGBs are the large Greek banks,and we therefore actually comparing large versus small banks.While this argument would have a basis,this is obviously the case in numerous studies that compare various groups of banks either in terms of ownership such as state/private(e.g.Noulas,1997),and foreign/domestic(e.g.Sturm and Williams,2004;Kasman and Yildirim,2006)or in terms of specialization such as commercial,savings,cooperative (e.g.Altunbas et al.,2001;Girardone et al.,2004).For example,domestic banks are in most cases quite larger than foreign banks operating in a country(i.e.subsidiaries),as commercial banks are usually larger than cooperative and savings banks. Noulas(1997)also mentions that the private banks in his sample are of much smaller size than the state ones.Hence, while one could keep in mind this note while interpreting the results,we do not believe that it reduces they usefulness of the study.

304F.Pasiouras/Research in International Business and Finance22(2008)301–318

time for smaller as well as larger banks.The regression of inef?ciency measures against a trend indicates that the improvement in technical and allocative inef?ciencies for small banks equal 19.7%and39.1%,accordingly.The corresponding?gures for large banks are10.4%and21.1%.

Christopoulos et al.(2002)examine the same sample with a multi-input,multi-output?exible cost function to represent the technology of the sector and a heteroscedastic frontier approach to measure technical ef?ciency.Regression of the ef?ciency measures over various bank characteris-tics indicates that larger banks are less ef?cient than smaller ones,and that economic performance, bank loans and investments are positively related to cost ef?ciency.

In a latter study,Tsionas et al.(2003)use the same sample as in Christopoulos and Tsionas (2001)and Christopoulos et al.(2002)but employ DEA to measure technical and allocative ef?ciency,and the Malmquist total factor productivity approach to measure productivity change. The results indicate that most of the banks operate close to the best market practices with overall ef?ciency levels over95%.Larger banks appear to be more ef?cient than smaller ones,while allocative inef?ciency costs seem to be more important than technical inef?ciency costs.They also document a positive but not substantial technical ef?ciency change which is mainly attributed to ef?ciency improvement for medium-sized banks and to technical change improvement for large banks.

Halkos and Salamouris(2004)also use DEA but follow a different approach,in contrast to previous studies,by using?nancial ratios as output measures and no input measures.The sample ranges between15and18banks depending on the year under consideration.The results indicate a wide variation in average ef?ciency over the period1997–1999,and a positive relationship between size and ef?ciency.Furthermore,there is non-systematic relationship between transfer of ownership through privatization of public banks and last period’s performance.

Apergis and Rezitis(2004)specify a translog cost function to analyze the cost structure of the Greek banking sector,the rate of technical change and the rate of growth in total factor productivity. They use both the intermediation and the production approach and a sample of six banks over the period1982–1997.Both models indicate signi?cant economies of scale and negative annual rates of growth in technical change and in total factor productivity.

Rezitis(2006)uses the same dataset but employs the Malmquist productivity index and DEA to measure and decompose productivity growth and technical ef?ciency,respectively.He also compares the1982–1992and1993–1997sub-periods,and employs Tobit regression to explain the differences in ef?ciency among banks.The results indicate that the average level of overall technical ef?ciency is91.3%,while productivity growth increased on average by2.4%over the entire period.The growth in productivity is higher in the second sub-period and is attributed to technical progress,in contrast to improvements in ef?ciency that was the main driver until1992. Furthermore,during the second sub-period pure ef?ciency is higher,and scale ef?ciency is lower, indicating that although banks achieved higher pure technical ef?ciency,they moved away from optimal scale.The regression results indicate that size and specialization have a positive impact on both pure and scale ef?ciency.

3.Methodology

From a methodological perspective,there are several approaches that can be used to examine the ef?ciency of banks,such as stochastic frontier analysis(SFA),thick frontier approach(TFA), distribution free approach(DFA),and DEA.Berger et al.(1993),Berger and Humphrey(1997) and Goddard et al.(2001)provide key discussions and comparisons of these methods in the context of banking.

F.Pasiouras/Research in International Business and Finance22(2008)301–318305

In the present study,following several recent studies we use DEA to estimate the ef?ciency of banks.4One of the well-known advantages of DEA,which is relevant to our study,is that it works particularly well with small samples.As Maudos et al.(2002a)point out,“Of all the techniques for measuring ef?ciency,the one that requires the smallest number of observations is the non-parametric and deterministic DEA,as parametric techniques specify a large number of parameters,making it necessary to have available a large number of observations.”(p.511).Other advantages of DEA are that it does not require any assumption to be made about the distribution of inef?ciency and that it does not require a particular functional form on the data in determining the most ef?cient decision making units(DMUs).On the other hand,the shortcomings of DEA are that it assumes data to be free of measurement error and it is sensitive to outliers.

We only brie?y outline DEA here,while more detailed and technical discussions can be found in Coelli et al.(1999),Cooper et al.(2000)and Thanassoulis(2001).The notations adopted below are those used in Coelli(1996)and Coelli et al.(1999),since we use their computer program DEAP 2.1to estimate the ef?ciency scores.

DEA uses linear programming for the development of production frontiers and the measure-ment of ef?ciency relative to the developed frontiers(Charnes et al.,1978).The best-practice production frontier for a sample of decision making units(DMUs),in our case banks,is con-structed through a piecewise linear combination of actual input–output correspondence set that envelops the input–output correspondence of all DMUs in the sample(Thanassoulis,2001).Each DMU is assigned an ef?ciency score that ranges between0and1,with a score equal to1indicating an ef?cient DMU with respect to the rest DMUs in the sample.

DEA can be implemented by assuming either constant returns to scale(CRS)or variable returns to scale(VRS).In their seminal study,Charnes et al.(1978)proposed a model that had an input orientation and assumed CRS.Hence,the output of this model is a score indicating the overall technical ef?ciency(OTE)of each DMU under CRS.

To discuss DEA in more technical terms,let us assume that there is data on K inputs and M outputs on each of N DMUs(i.e.banks).For the i th DMU these are represented by the vectors x i and y i,respectively.The K×N input matrix,X,and the M×N output matrix,Y,represent the data for all N DMUs.The input oriented measure of a particular DMU,under CRS,is calculated as:

Minθ,λθ,s.t.?y i+Yλ≥0,θx i?Xλ≥0,λ≥0

whereθ≤1is the scalar ef?cient score andλis N×1vector of constants.Ifθ=1the bank is ef?cient as it lies on the frontier,whereas ifθ?1the bank is inef?cient and needs a1?θreduction in the inputs levels to reach the frontier.The linear programming is solved N times,once for each DMU in sample,and a value ofθis obtained for each DMU representing its ef?ciency score.

Banker et al.(1984)suggested the use of variable returns to scale(VRS)that decomposes OTE into a product of two components.The?rst is technical ef?ciency under VRS or pure technical ef?ciency(PTE)and relates to the ability of managers to utilize?rms’given resources.The second is scale ef?ciency(SE)and refers to exploiting scale economies by operating at a point where the production frontier exhibits CRS.The CRS linear programming is modi?ed to consider VRS by adding the convexity N1 λ=1,where N1is a N×1vector of ones.The technical ef?ciency 4Examples of recent studies that use DEA are among others Haslem et al.(1999),Maudos et al.(2002a),Casu and Molyneux(2003),Drake and Hall(2003),Luo(2003),Ataullah et al.(2004),Hauner(2005),Ataullah and Le(2006), Casu and Girardone(2006)and Drake et al.(2006).

306F.Pasiouras/Research in International Business and Finance22(2008)301–318

scores obtained under VRS are higher than or equal to those obtained under CRS and SE can be obtained by dividing OTE with PTE(i.e.SE=OTE/PTE).

4.Data and variables

Our sample consists of the universe of commercial banks5with?nancial statements available in Bankscope database of Bureau van Dijk’s company,operating in Greece between2000and 2004.6Supplementary data for the banks(e.g.staff number,number of ATMs)were collected from the Hellenic Bank Association.The sample ranges between12and18banks per year and consists of78observations in total.

As mentioned in several studies,there is an on-going debate in the banking literature relative to the proper de?nition of inputs and outputs.The two main approaches are the“production approach”and the“intermediation approach”(Berger and Humphrey,1997).The production approach assumes that banks produce loans and deposit account services,using labour and capital as inputs,and the number and type of accounts measure outputs.The intermediation approach views banks as?nancial intermediates that collect purchased funds and use labour and capital to transform these funds into loans and other assets.Berger and Humphrey(1997)point out that neither of these two approaches is perfect because they cannot fully capture the dual role of?nancial institutions as providers of transactions/document processing services and being ?nancial intermediaries.They point out that the production approach may be somewhat better for evaluating the ef?ciency of branches of?nancial institutions and the intermediation approach may be more appropriate for evaluating entire?nancial institutions.Most recently,Drake et al. (2006)proposed the use of a pro?t-oriented approach in a DEA context that is in line with the approach of Berger and Mester(2003)in the context of their stochastic frontier approach.They point out that their results support the argument of Berger and Mester(2003)that a pro?t-based approach is better able to capture the diversity of strategic responses by?nancial?rms in the face of dynamic changes in competitive and environmental conditions.

In the present study,following most of the recent studies we adopt the intermediation approach. However,we also compare the obtained results with the ones of the pro?t-oriented approach suggested by Drake et al.(2006).We estimate?ve DEA models in total(Table1).

Models1–4are based on the intermediation approach but different inputs/outputs combinations are examined so as to explore the impact of credit risk and off-balance sheet activities on bank ef?ciency.In Model1,we select the following three inputs:?xed assets,customer deposits and short term funding,and number of employees.The two outputs of Model1are loans and other earning assets.Hence,this is a classical model under the intermediation approach found in most studies.In Model2,we introduce off-balance sheet items as an additional output,to account for the fact that in recent years banks are heavily involved in off-balance sheet activities.Model3 is a re-estimation of Model1but following Charnes et al.(1990),Drake(2001),Drake and Hall (2003),and Drake et al.(2006)among others,we include loan loss provisions as an additional 5On the basis of the classi?cation available in Bankscope.

6The study begins in2000for various reasons.First of all,this is the earliest year for which data were available in the online version of Bankscope to which we had access.Second,prior to2000the Greek banking industry witnessed a number of M&As that could complicate our analysis.Third,existing studies already provide evidence for various periods up to1999.Data for2005,that was the most recent year with available data,were not considered as the EU imposed the use of International Accounting Standards,and the data would not be comparable across the period of our analysis.

F.Pasiouras/Research in International Business and Finance22(2008)301–318307 Table1

Combination of inputs/outputs

Intermediation approach Pro?t-oriented approach

Model1Model2Model3Model4Model5 Inputs

Fixed assets Fixed assets Fixed assets Fixed assets Employee

expenses

Customer deposits and short term funding Customer deposits

and short term

funding

Customer deposits

and short term

funding

Customer deposits

and short term

funding

Other

non-interest

expenses

Number of employees Number of

employees Number of

employees

Number of

employees

Loan loss

provisions Loan loss provisions Loan loss provisions

Outputs

Loans Loans Loans Loans Net interest

income

Other earning assets Other earning assets Other earning assets Other earning assets Net

commission

income

Off-balance items Off-balance items Other operating

income

input in the DEA model to account for credit risk.7Finally,Model4is a re-estimation of Model1 that includes both off-balance sheet items and loan loss provisions,to simultaneously account for off-balance sheet activities and credit risk.Model5is the pro?t-oriented one,in which following Drake et al.(2006)revenue components are de?ned as outputs and cost components as inputs. The three inputs are:employee expenses,non-interest expenses,and loan loss provisions.The three outputs are:net interest income,net commission income and other income.As Drake et al.(2006)point out“from the perspective of an input-oriented DEA relative ef?ciency analysis, the more ef?cient units will be better at minimizing the various costs incurred in generating the various revenue streams and,consequently,better at maximizing pro?ts”(p.1451).

5.Empirical results

The discussion of the empirical results on the ef?ciency of banks in Greece is structured in three parts.First,we discuss the ef?ciency of the full sample of banks obtained through an input-oriented approach with VRS and the various inputs/outputs combination discussed above.8

7Mester(1996),Altunbas et al.(2000)and Drake and Hall(2003)among others point out that failure to adequately account for risk can have a signi?cant impact on relative ef?ciency scores.Berg et al.(1992)made the original observation and included nonperforming loans in a nonparmetric study of bank production,whereas Hughes and Mester(1993)applied the concept to parametric estimations(Berger and DeYoung,1997).Some other studies use equity capital as a control for risk(e.g.Altunbas et al.,2001;Maudos et al.,2002b;Akhigbe and McNulty,2003;Kasman and Yildirim,2006). However,Laeven and Majnoni(2003)mention that risk should be incorporated into ef?ciency studies via the inclusion of loan loss provisions that is actually a cost required to build up loan loss reserves.Altunbas et al.(2000)and Pastor and Serrano(2005)have used loan loss provisions in a stochastic frontier context as have the few recent studies in a DEA context mentioned in the text.

8Ef?ciency scores were estimated with DEAP2.1discussed in Coelli(1996).

308F.Pasiouras/Research in International Business and Finance22(2008)301–318

Table2

DEA results with intermediation approach(Models1–4)

Year TE(VRS)SE TE(VRS)SE

Mean Mean Mean Mean

Model1Model2

2004(N=18)0.8780.9830.8830.985 2003(N=17)0.9340.9780.9340.978 2002(N=17)0.9800.9510.9800.953 2001(N=14)0.9920.9950.9920.995 2000(N=12)0.9810.9350.9820.965 Overall(2001–2004;N=78)0.9490.9700.9500.975 Year TE(VRS)SE TE(VRS)SE

Mean Mean Mean Mean

Model3Model4

2004(N=18)0.9250.9940.9280.997 2003(N=17)0.9530.9810.9530.981 2002(N=17)0.9800.9670.9840.967 2001(N=14)0.9920.9960.9920.996 2000(N=12)0.9810.9550.9820.980 Overall(2001–2004;N=78)0.9640.9800.9660.984 Notes.TE,technical ef?ciency;SE,scale ef?ciency;VRS,variable returns on scale;Model1is estimated with?xed assets,customer deposits and short term funding,and number of employees as inputs,and loans and other earning assets as outputs;Model2is estimated as Model1but with off-balance sheet items as an additional output;Model3is estimated as Model1but with loan loss provisions as an additional input;Model4is estimated as Model1but with off-balance sheet items as an additional output and loan loss provisions as an additional input.

Next,we focus on the speci?c issue of the relative ef?ciency of IGBs versus PDBs.Finally,we investigate the determinants of ef?ciency using Tobit regression.9

5.1.Ef?ciency estimates—full sample

Table2presents the results from the four models that correspond to input/outputs selected on the basis of the intermediation approach.Table3reports the results of Model5that corresponds to the pro?t-oriented approach.

The average TE obtained by Model1ranges between0.878(2004)and0.992(2001),with an overall mean10over the entire period equal to0.949,while the corresponding?gures for SE are0.935(2000),0.995(2001)and0.970(overall mean),respectively.Hence,during2000–2004 banks could improve technical ef?ciency by5%and scale ef?ciency by3%on average.These ?gures increase very slightly when we include off-balance sheet items as an additional output,to 0.950(TE)and0.975(SE).However,when we consider loan loss provisions the overall mean tech-nical ef?ciency increases by1.5%.Thus,controlling for credit risk appears to have some impact on the ef?ciency scores.This is supported further by the only marginal increase by0.002in Model4 9Tobit analysis was performed with E-views5.1.

10This overall mean corresponds to the average calculated by pooling the ef?ciency scores calculated by year,and not to a model estimated with panel data.

F.Pasiouras/Research in International Business and Finance22(2008)301–318309 Table3

DEA results with pro?t-oriented approach(Model5)

TE(VRS)SE

Mean Mean

2004(N=18)0.9250.975 2003(N=17)0.9750.979 2002(N=17)0.9470.957 2001(N=14)0.9450.924 2000(N=12)0.9670.976 Overall(2001–2004;N=78)0.9510.963

Notes.TE,technical ef?ciency;SE,scale ef?ciency;VRS,variable returns on scale;Model5is estimated with employee expenses,other non-interest expenses and loan loss provisions as inputs,and net interest income,net commission income and other operating income as outputs.

where off-balance sheet items and loan loss provisions are simultaneously included,indicating that the increase from the base Model(i.e.Model1)is due mainly to loan loss provisions.Our results are similar to the ones obtained in previous studies for Greece that employ DEA and follow the intermediation approach.For example,Rezitis(2006)reports pure technical ef?ciencies between 0.977and0.994,and scale ef?ciencies between0.918and0.934depending on the period under consideration,while Tsionas et al.(2003)also report an overall technical ef?ciency equal to0.984.

Turning to the results obtained from the pro?t-oriented model(i.e.Model5)we observe that TE is between0.925(2004)and0.975(2003)with an overall mean equal to0.951.The corresponding ?gures for SE are0.924(2001),0.979(2003)and0.963(overall mean).The contrast between these results and the ones obtained from the intermediation approach are mixed.We only partially support the results of Drake et al.(2006)indicating that technical ef?ciency is generally higher under the intermediation approach than under the pro?t approach.In our study,this is not always the case and depends upon the models that are compared and the year of observation.More detailed,compared to Models1and2,technical ef?ciency under the pro?t-oriented approach (Model5)is higher during2003and2004and lower over the period2000–https://www.360docs.net/doc/4d3293899.html,pared to Models3and4,Model5’s technical ef?ciency is higher only during2003.However,it should be mentioned that the intermediation oriented model estimated in Drake et al.(2006)is most closely related to Model4of the present study.11

Looking at the overall mean now,we observe that the pro?t-oriented approach provides lower ef?ciency scores than Models3and4and almost identical scores with Models1and2.Never-theless,in our case the differences between the pro?t-oriented approach and the intermediation approach are much smaller than the ones reported in Drake et al.(2006).Another interesting point to emerge from the contrast of the results obtained by the two approaches is that the range in the ef?ciency scores is smaller when the pro?t-oriented approach is used.That is,the average technical ef?ciency scores for Model1range between0.878and0.992,and those of Model2 range between0.883and0.992.The corresponding?gures for Model3are0.925and0.992and those of Model4are0.928and0.992.In contrast,the range of the ef?ciency scores of Model5is only between0.925and0.975.This could in part be attributed to the following argument of Drake et al.(2006):“...the pro?t approach will capture the full impact of any adverse environmental factors on revenues as well as costs,while the intermediation approach tends to focus on the 11Drake et al.(2006)use personnel expenses as input whereas we use the number of staff members.They also use non-interest income rather than off-balance sheet items as an output for off-balance sheet activities.

310F.Pasiouras/Research in International Business and Finance22(2008)301–318

technical ef?ciency of the?nancial intermediation approach”(p.1462).However,in any case the differences among the models are generally quite small and therefore do not allow us to conclude whether the pro?t-oriented approach provides more reliable ef?ciency scores or not.

5.2.International versus purely domestic banks

We now turn to the ef?ciency of IGBs as opposed to the ef?ciency of PDBs.Morck and Yeung (1991)provide some empirical evidence of the multinational advantage based on the transfer of intangible assets such as technology and reputation from the home country to subsidiaries. Furthermore,operating abroad gives banks the opportunity to follow their customers and con-sequently retain them12(i.e.defensive expansion theory,see Williams,2002).This is obviously one type of transfer in the?rm,while the opposite or the transfer from one subsidiary to another is possible as well.Hence,banks that operate abroad might be able to transfer resources such as technology or employees with increased skills and experience in terms of risk management, regulatory and reporting practices,gained from working in more sophisticated and advanced environments(https://www.360docs.net/doc/4d3293899.html,,USA).In that case,the ef?ciency of IGBs will be higher than that of PDBs.On the other hand,IGBs will have to transfer efforts and resources to the subsidiaries that would otherwise be available to compete in the domestic market,and this might have a negative effect on their ef?ciency relative to PDBs.

Six of the banks in the sample(i.e.Alpha Bank,EFG Eurobank Ergasias,Egnatia Bank, Emporiki Bank of Greece,National Bank of Greece,Piraeus Bank)had subsidiaries abroad over the entire period of our analysis.Most of these banks as well as Agricultural Bank of Greece also had branches abroad,while Novabank had branches in2002and switched to an international presence through subsidiaries in2003and2004.We therefore adopt two de?nitions of IGBs.First,we consider as IGBs only those banks that are operating abroad through subsidiaries (SIGBs).Then,we consider as IGBs those banks that have an international presence either through subsidiaries or branches(SBIGBs).Table4reports the average ef?ciency scores for the two types of banks estimated by Models413and5,while distinguishing between PDBs and SIGBs(Panel A)and PDBs and SBIGBs(Panel B).

The contrast in the overall mean ef?ciency scores calculated from the pooled sample indicates that IGBs are more ef?cient than PDBs in almost all cases.The largest difference is observed in the case of TE while comparing IGBs and PDBs under the pro?t-oriented approach(0.043).That is,both SIGBs and SBIGBs can improve technical ef?ciency by2.4%whereas PDBs can improve it by6.7%.In contrast,the smallest difference is obtained in the case of SE and equals0.002.

The comparison of the technical ef?ciency scores of PDBs and IGBs by year indicates that under the intermediation approach PDBs have lower TE than both SIGBs and SBIGBs in all years.In general,this?nding holds under the pro?t-oriented approach as well,except in2003. With respect to SE estimated under the intermediation approach,the results are mixed,whereas under the pro?t-oriented approach PDBs have higher SE only during2000.

To examine whether the differences between the groups of interest are statistically signi?cant, we perform a Kruskal–Wallis(K–W)non-parametric test.Due to the small number of observa-

12It is possible that otherwise these customers would switch to another bank that provides services both abroad as well as in the home market.

13From this point and for the rest of our analysis we select Model4as representative of the intermediation approach, assuming that according to numerous recent studies it represents a more appropriate combination of inputs and outputs that considers off-balance sheet items and credit risk.

F .Pasiouras /Research in International Business and Finance 22(2008)301–318

311

Table 4

Purely domestic vs.international Greek banks

N PDBs IGBs

Model 4Model 5Model 4Model 5

TE (VRS)SE TE (VRS)SE TE (VRS)SE TE (VRS)SE

Mean Mean Mean Mean Mean Mean Mean Mean

Panel A:operations abroad through subsidiaries (SIGBs)

200411PDBs/7SIGBs 0.9210.9950.9140.9690.939 1.0000.9420.985200310PDBs/7SIGBs 0.9260.9840.9780.9650.9920.9770.9720.999200211PDBs/6SIGBs 0.9760.9690.9240.9410.9990.9630.9890.98520018PDBs/6SIGBs 0.9870.9930.9150.889 1.000 1.0000.9840.97120006PDBs/6SIGBs 0.9650.9850.9340.9920.9990.974 1.0000.960Overall (2001–2004)46PDBs/32SIGBs 0.9520.9850.9330.9510.9840.9830.9760.981Panel B:operations abroad through subsidiaries and branches (SBIGBs)

200410PDBs/8SBIGBs 0.9130.9940.9050.9660.946 1.0000.9490.98720039PDBs/8SBIGBs 0.9340.9830.9760.9610.9740.9790.9750.99920029PDBs/8SBIGBs 0.9700.9660.9360.9290.9990.9680.9590.98820017PDBs/7SBIGBs 0.9850.9920.9030.874 1.000 1.0000.9860.97520005PDBs/7SBIGBs 0.9580.9820.9200.9910.9990.978 1.0000.965Overall (2001–2004)40PDBs/38SBIGBs 0.9490.9830.9300.9430.9830.9850.9730.984Notes .PDBs,purely domestic banks;IGBs,international Greek banks;TE,technical ef?ciency;SE,scale ef?ciency;VRS,variable returns on scale;Model 4is estimated with ?xed assets,customer deposits and short term funding,number of employees and loan loss provisions as inputs,and loans,other earning assets and off-balance sheet items as outputs;Model 5is estimated with employee expenses,other non-interest expenses and loan loss provisions as inputs,and net interest income,net commission income and other operating income as outputs.

312F.Pasiouras/Research in International Business and Finance22(2008)301–318

tions from each group by year,the test is performed on the scores of the pooled sample of the 78observations.The results of the K–W test indicate that under the intermediation approach IGBs,both SIGBs and SBIGBs,are more ef?cient than PDBs,in terms of TE that is statistically signi?cant at the10%and5%level,respectively.14Hence,our results appear to be in line with the ones of Isik and Hassan(2002)for the Turkish banking sector.There are several possible explanations for these?ndings.First,these banks may transfer resources such as technology or employees with increased skills gained abroad to the home market,hence increasing their technical ef?ciency.They can also retain their customers by following them abroad or experience gains due to increased reputation and diversi?cation that reduces their risk.As for the insigni?cant differences in the case of scale ef?ciency,Isik and Hassan(2002)mention that exploitation of scale economies cannot be the motivation for these large banks’foreign expansion as multina-tional banks would have exhausted their scale economies when they were small,pure domestic banks.

5.3.Stage2—Tobit analysis

In order to investigate the determinants of ef?ciency we construct an econometric model with technical and scale ef?ciency as dependent variables.As in previous studies,due to the ef?ciency measures ranging between0and1we use Tobit analysis.As Saxonhouse(1976)points out,heteroscedacity can emerge when estimated values are used as dependent variables in the second stage analysis.Hence,following Hauner(2005),QML(Huber/White)standard errors and covariates are calculated.

We examine the effect of two groups of factors on ef?ciency.First,we analyze the in?uence of various bank?nancial characteristics.We follow previous studies and examine the following variables:equity to assets(EQAS),return on average assets(ROAA),loan to assets(LOANS), and market power(POWER)as measured by the relative size of bank(i.e.market share in terms of assets).

Second,we examine the in?uence of bank’s strategies in terms of investments in technology (i.e.ATMs and branches)and internationalization of operations.We include the number of ATMs (ATMs),the number of branches(BRANCH),and two dummy variables indicating whether banks are offering their services abroad through subsidiaries(SUB ABR,that takes the value of1if yes and0otherwise)or branches(BR ABR,that takes the value of1if yes and0otherwise).

Considering the small number of observations in our sample,we estimate two speci?cations of the Tobit model with each one of the two sets of variables examined sequentially.We follow this approach in order not to overload the regressions.The?ndings are reported in Table5.Panel A presents the results of the regressions with the bank?nancial characteristics,while Panel B presents the ones with the variables that proxy for the strategic decisions of the banks.

EQAS is statistically signi?cant and positively related to ef?ciency in all our speci?cations. Hence,well-capitalized banks are also more ef?cient,both in terms of technical and scale ef?-ciency.These results are in line with Isik and Hassan(2003a)in Turkey,Casu and Girardone (2004)in Italy,Rao(2005)in United Arab Emirates and Kwan and Eisenbeis(1997)in the US among others,all reporting a positive relation between capitalization and various measures of

14In the?rst case,the chi-square equals3.520whereas in the latter case it equals4.930.Differences in SE were not signi?cant for neither of the two models,as they were not signi?cant in the case of TE obtained from Model5.The chi-square and p-values are not reported here but are available from the author upon request.

F.Pasiouras/Research in International Business and Finance22(2008)301–318313 Table5

Tobit censored regression results

Model4Model5

TE SE TE SE

Coef.p-Value Coef.p-Value Coef.p-Value Coef.p-Value Panel A:TE and SE regressed over bank?nancial characteristics

EQAS0.0670.0000.0430.0000.0510.0000.0190.002 ROAA0.0040.874?0.0090.7130.1150.0010.0700.062 LOANS0.0060.0030.0100.0000.0090.0000.0140.000 POWER0.0390.0000.0210.0000.0240.0000.0170.000 Panel B:TE and SE regressed over strategic decisions related variables

ATM0.0020.2610.0000.7660.0010.317?0.0000.816 BRANCH0.0060.0810.0050.0020.0090.0120.0070.006 BR ABR?0.6390.429?0.4200.549?1.6040.040?0.4790.491 SUB ABR0.7680.0290.7310.0020.6790.1180.3200.218 Notes.N=78observations;TE,technical ef?ciency;SE,scale ef?ciency;Model4is estimated with?xed assets,customer deposits and short term funding,number of employees and loan loss provisions as inputs,and loans,other earning assets and off-balance sheet items as outputs;Model5is estimated with employee expenses,other non-interest expenses and loan loss provisions as inputs,and net interest income,net commission income and other operating income as outputs; EQAS,equity to assets;ROAA,return on average assets;LOANS,loans to assets;POWER,market share in terms of total assets;ATM,the number of bank’s ATMs;BRANCH,the number of bank’s branches;BR ABR,dummy variable that equals1if the bank has branches abroad and0otherwise;SUB ABR,dummy variable that equals1if the bank has subsidiaries abroad and0otherwise;QML(Huber/White)standard errors and covariates have been calculated to control for heteroscedacity.

ef?ciency.One potential explanation for these?ndings is that since EQAS re?ects the degree to which shareholders have their own capital at risk in their institution it also re?ects their incentives to monitor management and assure that the bank operates ef?ciently(Eisenbeis et al.,1999). Hence,as Isik and Hassan(2003a)mention these results are in favour of the conjectures of moral hazard theory.

ROAA is positively related to the ef?ciency measures in almost all cases however,it is sta-tistically signi?cant only in the case of the pro?t-oriented approach(i.e.Model5).Although Christopoulos et al.(2002)report a positive and signi?cant relationship between pro?tability and ef?ciency in the Greek banking sector between1993and1998,the results from studies in other countries are mixed.For instance,Ataullah and Le(2006)report both negative and positive statistically signi?cant impacts of return on assets on ef?ciency measures in India depending on the speci?cation of the model.Casu and Molyneux(2003)examine a sample of banks from the principal EU banking sectors15and?nd a positive relationship between pro?tability ef?ciency, which is however statistically signi?cant in only two of the?ve years of the analysis.Isik and Hassan(2002)report a positive and signi?cant correlation between both return on equity and return on assets and ef?ciency in Turkey.However,Casu and Girardone(2004)report a negative and statistically signi?cant relationship in Italy.

LOANS carries a positive sign that is statistically signi?cant in all cases and is consistent with Isik and Hassan(2003a).Casu and Girardone(2004)also report a positive relationship although 15The principal EU banking sectors are:France,Germany,Italy,Spain,UK.

314F.Pasiouras/Research in International Business and Finance22(2008)301–318

not statistically signi?cant.Isik and Hassan(2003a)argue that the positive relationship between loan activity and ef?ciency can be attributed to the ability of relatively ef?cient banks to manage operations more productively,that enables them to have lower production costs and consequently to offer more reasonable loan terms allowing them to gain larger share in the loan market segment. In contrast to the above studies,Havrylchyk(2006)?nds a negative relationship between the loans to assets ratio and ef?ciency,which however becomes positive once the off-balance sheet items are omitted from the output vector.

POWER is also statistically signi?cant and positively related to TE and SE in both models (i.e.4and5).Since this variable re?ects the relative size of the bank and its market power, our?ndings seem to support the arguments in favour of size as well as market share.While our results contradict those of Christopoulos et al.(2002)who report a negative relationship between size and ef?ciency,they are in line with the studies of Halkos and Salamouris(2004)and Rezitis(2006)in Greece and Berger et al.(1993)and Miller and Noulas(1996)in the US that report a positive relationship between size and ef?ciency.When we interpret our variable as an indicator of market power rather than size,we tend to support the ef?cient structure hypothesis. As Isik and Hassan(2003a)explain,relatively ef?cient?rms,due to their low costs of production, might have competed more aggressively,made higher pro?ts and ultimately gained larger market share.

Turning to the variables that are related to the strategic choices of the banks,the results indicate that only BRANCH is signi?cant in all cases(although signi?cant only at the10%level in the case of TE estimated with Model4).On the other hand,ATM does not have a statistically signi?cant impact on ef?ciency in any of our speci?cations.One potential explanation is that the Greek banking system relies heavily on branches as a distribution network with the number of branches increasing from year to year contrary to other EU countries,where a declining trend takes place. As mentioned in the summary of the Annual Report of the Bank of Greece(2005)the continued increase in the number of bank branches in Greece is associated with the fast growth of retail banking and Greek customers’continued preference for transaction through branches.While the number of ATMs has also increased(e.g.5468in2003;5787in2004)such channels are supplementary to branches,which retain their key role as points of sale(Annual Report of Bank of Greece,2004).Attracting new customers and maintaining the existing ones is a main strategic choice for banks,which,as mentioned in the2004Annual Report of the Bank of Greece,can be achieved more effectively through personal contact at branches.Furthermore,an extensive branch network also supports the expansion of Greek banks via cross-selling such as bank assurance products(summary of Annual Report of Bank of Greece,2005).These characteristics of the Greek banking sector obviously explain why the number of branches has a signi?cant impact on the ef?ciency of banks,and why the alternative distribution networks of ATMs do not seem to signi?cantly in?uence bank ef?ciency.

Finally,with respect to the variables that are related to the international presence of banks, the results are mixed.Operating abroad through branches appears to be negatively related to the ef?ciency of banks,which is,however,signi?cant only in the case of TE estimated with Model 5.In contrast,operating abroad through subsidiaries appears to have a positive impact on both technical and scale ef?ciency,although this is statistically signi?cant only for Model4.

6.Conclusions

In the present study,we estimate both the technical and scale ef?ciency of Greek banks over the period2000–2004.We use input oriented data envelopment analysis with variable returns to

F.Pasiouras/Research in International Business and Finance22(2008)301–318315 scale and estimated?ve models to examine several issues not considered in the study of the Greek banking sector in the past.

In particular,we estimate the ef?ciency of the banks in our sample with and without off-balance sheet items and loan loss provisions to account for different levels of off-balance sheet activities and credit risk.In all cases,the models are estimated following the traditional intermediation approach and the recently proposed(in the context of DEA)pro?t-oriented approach.We also compare the ef?ciency of banks that have an international presence with the ones that offer their services only in the domestic market,an issue that has received relatively little attention in the bank ef?ciency literature.Finally,we use Tobit analysis to regress the ef?ciency scores obtained from DEA on several variables re?ecting bank?nancial characteristics and strategic decisions.

The results indicate that the inclusion of off-balance sheet items in the outputs does not have an impact on the ef?ciency scores,while the inclusion of loan loss provisions in the inputs contributes to higher ef?ciency https://www.360docs.net/doc/4d3293899.html,paring scores obtained from the models estimated by the intermediation approach with those obtained from the pro?t-oriented approach,year by year,provided mixed results.However,we?nd that in terms of the overall mean,the pro?t-oriented model provided lower ef?ciency scores than two of the models estimated under the intermediation approach and almost identical scores with the other two models.In any case,the differences between the two approaches were much smaller than the ones reported in Drake et al. (2006).

Banks with international operations appear to be more ef?cient than those operating only at the national level,consistent with Isik and Hassan(2002).We obtain similar results whether we de?ne as international,banks that offer their services abroad either through subsidiaries or both subsidiaries and branches.However,the differences are statistically signi?cant only in the case of technical ef?ciency estimated under the intermediation approach.

Finally,we regress the scores obtained from the pro?t-oriented model and the full interme-diation model over banks’?nancial characteristics and variables re?ecting strategic decisions. Capitalization,loan activity and market share in terms of total assets are statistically signi?cant and positively related to the ef?ciency measures in all cases.Pro?tability is positively related to the ef?ciency measures in all cases;however it is statistically signi?cant only in the case of the pro?t-oriented approach.Turning to the variables related to the strategic choices of the banks, the results indicate that the number of branches is signi?cant in all cases,while the number of ATMs does not have much impact in any of our speci?cations.Finally,with respect to the dummy variables indicating whether the banks are operating abroad through branches or subsidiaries,the results are mixed.

Future research could extend the present study in numerous directions.First,commercial banks could be compared with cooperative banks,as the latter have received little attention in past studies for Greece.Second,domestic banks could be compared with foreign banks,which have also received limited attention(and not in the context of ef?ciency).Finally,it would be worthwhile to consider a longer time period and examine the impact of environmental factors such as GDP,in?ation and stock market capitalization on the ef?ciency of the Greek banking sector.

Acknowledgments

I would like to thank an anonymous reviewer and Sailesh Tanna for valuable comments that helped me improve an earlier version of the manuscript.Any remaining errors are,of course,my own.

316F.Pasiouras/Research in International Business and Finance22(2008)301–318 References

Akhigbe,A.,McNulty,J.E.,2003.The pro?t ef?ciency of small US commercial banks.J.Bank.Financ.27,307–325. Altunbas,Y.,Chakravarty,S.P.,2001.Frontier cost functions and bank ef?ciency.Econ.Lett.72,233–240. Altunbas,Y.,Gardener,E.P.M.,Molyneux,P.,Moore,B.,2001.Ef?ciency in European banking.Eur.Econ.Rev.45, 1931–1955.

Altunbas,Y.,Liu,M.-H.,Molyneux,Ph.,Seth,R.,2000.Ef?ciency and risk in Japanese banking.J.Bank.Financ.24, 1605–1628.

Apergis,N.,Rezitis,A.,2004.Cost structure,technological change,and productivity growth in the Greek banking sector.

Int.Adv.Econ.Res.10(1),1–15.

Ataullah,A.,Cockrill,T.,Le,H.,2004.Financial liberalization and bank ef?ciency:a comparative analysis of India and Pakistan.Appl.Financ.Econ.36,1915–1924.

Ataullah,A.,Le,H.,2006.Economic reforms and bank ef?ciency in developing countries:the case of the Indian banking industry.Appl.Financ.Econ.16,653–663.

Bank of Greece,2004.Annual Report2004.Athens.Available at:http://www.bankofgreece.gr/en/publications/report.asp. Bank of Greece,2005.Summary of the Annual Report2005.Athens.Available at:http://www.bankofgreece.gr/en/ publications/report.asp.

Banker,R.D.,Charnes,A.,Cooper,W.W.,1984.Some models for estimating technical and scale inef?ciencies in data envelopment analysis.Manage.Sci.30,1078–1092.

Berg,S.,Forsund,F.,Jansen,E.,1992.Malmquist indices of productivity growth during the deregulation of Norwegian banking,1980–89.Scand.J.Econ.94,211–228.

Berger,A.N.,DeYoung,R.,1997.Problem loans and cost ef?ciency in commercial banks.J.Bank.Financ.21,849–870.

Berger,A.N.,Humphrey,D.B.,1997.Ef?ciency of?nancial institutions:international survey and directions for future research.Eur.J.Operat.Res.98,175–212.

Berger,A.N.,Hunter,W.C.,Timme,S.G.,1993.The ef?ciency of?nancial institutions:a review and preview of research past,present and future.J.Bank.Financ.17,221–249.

Berger,A.N.,Mester,L.J.,2003.Explaining the dramatic changes in performance of US banks:technological change, deregulation,and dynamic changes in competition.J.Financ.Intermed.12,57–95.

Bos,J.W.B.,Colari,J.W.,https://www.360docs.net/doc/4d3293899.html,rge bank ef?ciency in Europe and the United States:are there economic motivations for geographic expansion in?nancial services.J.Bus.78(4),1555–1592.

Casu,B.,Girardone,C.,2004.Financial conglomeration:ef?ciency,productivity and strategic drive.Appl.Financ.Econ.

14,687–696.

Casu,B.,Girardone,C.,2006.Bank competition,concentration and ef?ciency in the single European market.Manch.

Sch.74(4),441–468.

Casu,B.,Molyneux,Ph.,2003.A comparative study of ef?ciency in European banking.Appl.Econ.35,1865–1876.

Charnes,A.,Cooper,W.W.,Rhodes,E.,1978.Measuring the ef?ciency of decision making units.Eur.J.Operat.Res.2, 429–444.

Charnes,A.,Copper,W.W.,Huang,Z.M.,Sin,D.B.,1990.Polyhedral cone-ratio DEA models with illustrative application to large commercial banks.J.Econom.46(1/2),73–91.

Christopoulos,D.K.,Lolos,S.E.G.,Tsionas,E.G.,2002.Ef?ciency of the Greek banking system in view of the EMU:a heteroscedastic stochastic frontier approach.J.Policy Model.24,813–829.

Christopoulos,D.K.,Tsionas,E.G.,2001.Banking economic ef?ciency in the deregulation period:results from het-eroscedastic stochastic frontier models.Manch.Sch.69(6),656–676.

Coelli,T.,1996.A guide to DEAP version2.1:a data envelopment analysis(computer)program.CEPA Working Paper 1996/08.Available at:https://www.360docs.net/doc/4d3293899.html,.au/econometrics/cepa.htm.

Coelli,T.,Prasada Rao,D.S.,Battese,G.E.,1999.An Introduction to Ef?ciency and Productivity Analysis.Kluwer Academic Publishers,USA.

Cooper,W.W.,Seiford,L.M.,Tone,K.,2000.Data Envelopment Analysis.A Comprehensive Text with Models,Appli-cations,References and DEA-Solver Software.Kluwer Academic Publishers,USA.

Drake,L.,2001.Ef?ciency and productivity change in UK banking.Appl.Finan.Econ.11,557–571.

Drake,L.,Hall,M.J.B.,2003.Ef?ciency in Japanese banking:an empirical analysis.J.Bank.Financ.27,891–917.

Drake,L.,Hall,M.J.B.,Simper,R.,2006.The impact of macroeconomic and regulatory factors on bank ef?ciency:a non-parametric analysis of Hong Kong’s banking system.J.Bank.Financ.30,1443–1466.

F.Pasiouras/Research in International Business and Finance22(2008)301–318317 Eisenbeis,R.A.,Ferrier,

G.D.,Kwan,S.

H.,1999.The informativeness of stochastic frontier and programming frontier ef?ciency scores:cost ef?ciency and other measures of bank holding company performance.Working Paper No.23.

Federal Reserve Bank of Atlanta.

Girardone,C.,Molyneux,Ph.,Gardener,E.P.M.,2004.Analysing the determinants of bank ef?ciency:the case of Italian banks.Appl.Econ.36,215–227.

Goddard,J.A.,Molyneux,Ph.,Wilson,J.O.S.,2001.European Banking:Ef?ciency,Technology and Growth.John Wiley &Sons,England.

Halkos,G.E.,Salamouris,D.S.,2004.Ef?ciency measurement of the Greek commercial banks with the use of?nancial ratios:a data envelopment analysis approach.Manage.Account.Res.15,201–224.

Haslem,J.A.,Scheraga,C.A.,Beding?eld,J.P.,1999.DEA ef?ciency pro?les of U.S.banks operating internationally.

Int.Rev.Econ.Financ.8,165–182.

Hauner,D.,2005.Explaining ef?ciency differences among large German and Austria banks.Appl.Econ.37,969–980.

Havrylchyk,O.,2006.Ef?ciency of the Polish banking industry:foreign versus domestic banks.J.Bank.Financ.30, 1975–1996.

Hughes,J.P.,Mester,L.J.,1993.A quality and risk-adjusted cost function for banks:evidence on the‘too-big-to fail’doctrine.J.Prod.Anal.4,293–315.

Isik,I.,Hassan,M.K.,2002.Technical,scale and allocative ef?ciencies of Turkish banking industry.J.Bank.Financ.26, 719–766.

Isik,I.,Hassan,M.K.,2003a.Ef?ciency,ownership and market structure,corporate control and governance in the Turkish banking industry.J.Bus.Financ.Account.30(9/10),1363–1421.

Isik,I.,Hassan,M.K.,2003b.Financial deregulation and total factor productivity change:an empirical study of Turkish commercial banks.J.Bank.Financ.27,1455–1485.

Karafolas,S.,Mantakas,G.,1996.A note on cost structure and economies of scale in Greek banking.J.Bank.Financ.

20,377–387.

Kasman,A.,Yildirim,C.,2006.Cost and pro?t ef?ciencies in transition banking:the case of new EU members.Appl.

Econ.38,1079–1090.

Kwan,S.,Eisenbeis,R.A.,1997.Bank risk,capitalization,and operating ef?ciency.J.Financ.Serv.Res.12(2/3),117–131.

Laeven,L.,Majnoni,G.,2003.Loan loss provisioning and economic slowdowns:too much,too late?J.Financ.Intermed.

12,178–197.

Lang,G.,Welzel,P.,1998.Technology and cost ef?ciency in universal banking.A“Thick Frontier”—analysis of the German banking industry.J.Product.Anal.10,63–84.

Luo,X.,2003.Evaluating the pro?tability and marketability ef?ciency of large banks.An application of data envelopment analysis.J.Bus.Res.56,627–635.

Maudos,J.,Pastor,J.M.,Perez,F.,Quesada,J.,2002a.Cost and pro?t ef?ciency in European banks.J.Int.Financ.Markets, Inst.Money12,33–58.

Maudos,J.,Pastor,J.M.,Perez,F.,https://www.360docs.net/doc/4d3293899.html,petition and ef?ciency in the Spanish banking sector:the importance of specialization.Appl.Financ.Econ.12,505–516.

Mester,L.J.,1996.A study of bank ef?ciency taking into account risk-preferences.J.Bank.Financ.20,1025–1045.

Miller,S.M.,Noulas,A.G.,1996.The technical ef?ciency of large bank production.J.Bank.Financ.20,495–509. Morck,R.,Yeung,B.,1991.Why investors value multinationality.J.Bus.64,165–187.

Noulas,A.G.,1997.Productivity growth in the Hellenic banking industry:state versus private banks.Appl.Financ.Econ.

7,223–228.

Pastor,J.M.,Serrano,L.,2005.Ef?ciency,endogenous and exogenous credit risk in the banking systems of the Euro area.

Appl.Financ.Econ.15,631–649.

Rao,A.,2005.Cost frontier ef?ciency and risk-return analysis in an emerging market.Int.Rev.Financ.Anal.14,283–303.

Rezitis,A.,2006.Productivity growth in the Greek banking industry:a non-parametric approach.J.Appl.Econ.9(1), 119–138.

Saxonhouse,G.R.,1976.Estimated parameters as dependent variables.Am.Econ.Rev.66,178–183.

Sturm,J.-E.,Williams,B.,2004.Foreign bank entry,deregulation and bank ef?ciency:lessons from the Australian experience.J.Bank.Financ.28,1775–1799.

Thanassoulis,E.,2001.Introduction to the Theory and Application of Data Envelopment Analysis.A Foundation Text with Integrated Software.Kluwer Academic Publishers,USA.

318F.Pasiouras/Research in International Business and Finance22(2008)301–318

Tortosa-Ausina,E.,2003.Nontraditional activities and bank ef?ciency revisited:a distributional analysis for Spanish ?nancial institutions.J.Econ.Bus.55,371–395.

Tsionas,E.G.,Lolos,S.E.G.,Christopoulos,D.K.,2003.The performance of the Greek banking system in view of the EMU:results from a non-parametric approach.Econ.Model.20,571–592.

Williams,B.,2002.The defensive expansion approach to multinational banking:evidence to date.Financ.Markets Inst.

Instrum.11,127–203.