Efficiency of the Greek banking system in view of the EMU: a heteroscedastic stochastic frontier approach

Journal of Policy Modeling24 (2002) 813–829

Efficiency of the Greek banking system inview of the EMU: a heteroscedastic

stochastic frontier approach

Dimitris K. Christopoulosa,∗, Sarantis E.G. Lolosa,Efthymios G. Tsionasb

aDepartment of Economic and Regional Development, Panteion University, Athens, GreecebDepartment of Economics, Athens University of Economics and Business, Athens, Greece

Received 1 August 2000; received in revised form 17 July 2001; accepted 28 March 2002

Abstract

The paper estimates empirically cost efficiency of the Greek banking system for the pe-riod 1993–1998. The beginning of the examination period coincides with the acceleration ofliberalization and deregulation of the Greek financial system, in view of the country joiningthe EMU. The study uses a multi-input, multi-output technology and adopts a heteroscedas-tic frontier model instead of a commonly used homoscedastic one to measure cost efficiencyin the banking system. The empirical results show that larger banks are less efficient thansmaller ones. Also, it is found that economic performance, bank loans and investments arepositive related to the cost efficiency of the Greek commercial banking sector.© 2002 Society for Policy Modeling. Published by Elsevier Science Inc. All rights reserved.

JEL classification: C53; D24; G21

Keywords: Banking; Efficiency; Liberalization; Heteroscedastic frontier model

1. Introduction

Until the mid-1980s the Greek banking system operated in an environment char-acterized by selective controls and regulations which gradually led to inefficiency

∗ Corresponding author. Tel.:+30-19224948; fax:+30-19229312.E-mail address: [email protected] (D.K. Christopoulos).

0161-8938/02/$ – see front matter © 2002 Society for Policy Modeling.PII: S0161-8938(02)00174-6

814 D.K. Christopoulos et al. / Journal of Policy Modeling 24 (2002) 813–829

and to serious distortions in the functioning of the country’s financial system. Theneed for a modern, flexible and market-oriented financial system and the prospectsfor participating in the Single European Market and the EMU initiated efforts to-wards the deregulation of the financial system. In recent years banking activity inGreece was decisively affected by the harmonization of national regulations withinthe European Union and especially with the enactment of the Second Banking Di-rective (1992). In addition, consistent macroeconomic policies have been adoptedin view of the country’s prospects of joining the EMU, which gradually reducedinflation and the interest rates.1 Currently, the Greek banking system is faced withincreased competition and internationalization, while banking disintermediationhas been in evidence. These factors, together with the fast development of infor-mation technology, trigger major structural changes in the Greek banking system,which is making efforts at increasing efficiency, reducing costs of bank servicesand diversifying in other business areas. More changes are expected in the yearsto come.

In recent years, only one empirical work has focused attention on inefficienciesof Greek banking system. In particularNoulas (1997)used a data envelope analysis(DEA) methodology to evaluate the relative performance of state against privatebanks for 1992 to conclude that (i) technical efficiency decreased for both privateand state banks, and (ii) state banks experienced technical progress while privatebanks did not. However, his study limited the analysis to the pre-1993 period whenthe liberalization of the financial system was actually initiated.

The purpose of this paper is to extend earlier work on cost efficiency of thebanking system to a medium-sized country like Greece and study whether it hasbeen affected by the initiated liberalization process and deregulation measures thathave been taken. Up to the present the Greek banking system has not been studiedadequately due to data deficiencies and its features have not been thoroughlyanalyzed.2 We believe that the assessment and quantification of cost efficiencyof the Greek banking system over the liberalization period is necessary for theevaluation of its performance and future prospects within the EMU framework. Theconclusions drawn could prove useful for the analysis of the cost structure of thebanking sectors in other medium-sized economies that are undergoing structuralchanges.

The empirical evidence on Greece is arrived at using multi-input, multi-outputflexible cost function to represent the cost structure of the banking system. Tech-nical efficiency is estimated using stochastic frontier models. However, whenestimating stochastic frontiers in the banking sector it is quite probable that het-eroscedasticity is present in the data since banks of different “size” are pooled

1 Since 1993 an ambitious macroeconomic program (the “Greek Convergence Programme”) wassuccessfully implemented aiming at achieving the Maastricht criteria by the end of the 1990s. For adiscussion on this issue, seeTsionas (2000).

2 Since 1993 accounting and auditing standards for Greek banks were notably upgraded, where themost significant improvement was the introduction of consolidated financial statements.

D.K. Christopoulos et al. / Journal of Policy Modeling 24 (2002) 813–829 815

together. In this case, a heteroscedastic cost frontier approach should be used.Caudill and Ford (1993)andCaudill, Ford, and Gropper (1995)reported that het-eroscedasticity in the one-sided error affects materially parameter estimates in asingle factor frontier production function. Specifically, when the model is esti-mated by maximum likelihood, heteroscedasticity leads to an overestimation ofthe intercept and to an underestimation of the slope coefficients. Thus, they con-clude that inefficiency measures are affected by heteroscedasticity. In view of this,a heteroscedastic stochastic frontier approach is adopted here to evaluate technicalefficiency performance of the Greek banking sector in the deregulation period.

The remaining of the paper is organized as follows. Recent developments inthe Greek banking system are contained inSection 2. The econometric modelis presented inSection 3. In Section 4the empirical evidence is discussed andthe technical efficiency index is related to various economic variables. The finalsection summarizes the results and discusses implications for the banking activityin Greece.

2. Recent developments in the Greek banking system

A major structural feature of the Greek banking system in the past was its in-stitutional specialization required by law rather than dictated by market forces.Towards the late 1980s, there was a process of gradual and extensive liberalizationof the market, motivated by international developments and the need for partici-pation in the Single European Market for financial services. During the 1990s, theprocess of liberalization and deregulation of the banking system was carried out atan accelerating pace and the provision of the Second Banking Directive concern-ing establishment, operation and supervising of credit institutions was passed bythe Greek Parliament in August 1992. Various measures were also taken towardsthe modernization of the capital market.3

Commercial banks have been the dominant group in the Greek banking system.In 1998, besides the Central Bank, there were 49 credit institutions establishedand operating in the Greek banking market. The commercial banking system com-prises of 18 commercial banks4 and 20 branches of foreign banks, of which 12 areEU-based. Also there are seven specialized credit institutions, namely two invest-ment banks, three real-estate banks, a saving bank, a specific purpose bank, ninecredit co-operatives as well as 13 co-operatives banks.

Basic size characteristics of the Greek banking system are depicted inTable 1,below. Over the period 1993–1998, the presence of the large banks, in terms of

3 For a detailed discussion on the developments in the regulatory and institutional framework ofthe Greek financial system, see,inter alia, Christopoulos et al. (2000), Hondroyiannis, Lolos, andPapapetrou (1999)and the references cited therein.

4 Including the Agricultural Bank, which, although it is classified as a commercial bank since 1991,still operates rather as a specialized credit institution.

816D

.K.C

hristopoulosetal./Journal

ofPolicy

Modeling

24(2002)

813–829


assets and branches in the Greek banking system remained high, while most ofthe smaller banks expanded far more rapidly in relation to larger ones. Also, themajority of large banks and the fast growing smaller banks have, on the average, ahigher amount of deposits and products per branch. On the assets side, loans andadvances has been the most important output for all banks amounting to around40% of total assets, while the share of investments in shares is very low withthe exception of large banks and the faster growing smaller banks. The share ofbonds (mainly government securities) is fairly significant for most banks a factattributed to the obligation (until 1991) of banks to keep a high proportion (40%)of their portfolio in government securities. In subsequent years the purchasingof new treasury bills was gradually reduced but the stock of public debt held bythe banking system continued, keeping high interest rate margins between loansand deposits.5 However, the situation was gradually normalized along with thereduction of budget deficits and the liberalization of the banking system.

A specific structural feature of the Greek financial-credit system, which includesthe smallest number of credit institutions in the European Union, has been thedecisive presence of the state, the dominant role of a few large banks and thelimited share of foreign banks. Also, the relative weight of banking in the Greekfinancial system is still very high, although the depth of the capital and moneymarkets has gradually increased considerably in recent years. Recently, the Greekfinancial and banking landscape is changing rapidly and more major changes areexpected in the years to come, as a result of government measures and also as aresult of the dynamics of the domestic and the EU financial and credit markets.

In particular, the government decided at reducing the state’s presence in acontrolled way. Thus, four small and medium-sized banks (Attica, Cretabank,Macedonia-Thrace, Central Greece) and the third largest bank (Ionian) were pri-vatized. However, the state presence is still high (less so in the commercial bankingmarket), since all specialized credit institutions and two of the larger commercialbanks (National, Commercial) plus a medium-sized one (General) are directly orindirectly state-controlled. Also, a series of mergers and acquisitions were under-taken which changed substantially the structure of the Greek banking system. Inthe private sector, the third largest private bank (Eurobank) acquired two smallprivate banks (Interbank in 1996, Bank of Athens in 1998) and a major privatebank (Ergobank in 1998/1999). Also, the Bank of Piraeus bought two small banks(Xiosbank in 1998 and Prime in 1998) and a small bank (Dorian) was sold to aninvestment bank (Telesis in 1999). In the public sector, the National Bank mergedwith the two public mortgage banks (National Mortgage and Housing Bank) in1998. In 2000, the second major state-controlled bank (Commercial Bank) waspartially privatized.

Note that most of the above mentioned changes arenot taken explicitly intoaccount in our empirical analysis, with the exception of the cases of Interbank andthe Bank of Athens, since none of the mergers and acquisitions carried out over the

5 For a discussion on this point, seeHondroyiannis, Lolos, and Papapetrou (1998).


period 1993–1998 were practically realized within the examination period. Therespective banks continued to operate independently and the actual changes areexpected to come about in mid-2000.

3. The statistical model

Consider the following cost frontier

logCit = h(zit,�)+ vit + uit, i = 1, . . . , n, t = 1, . . . , T (1)

whereCit denotes the cost of the firmi of datet, zit = [p′ity

′it]

′ is ak × 1 vectorof explanatory variables which includes the logarithms of input prices and outputquantities,pit is theq × 1 vector of input prices,yit is thes × 1 vector of outputquantities (notice thatk = q+ s), � is ak×1 vector of parameters,h is a function,vit is a two-sided noise error term, anduit is a one-sided error term representingtechnical inefficiency.

In this paper, we make the following assumptions:

(i) vit ∼ IIDN (0,�2v),

(ii) uit ∼ IN(0,�2u,it), uit ≥ 0,

(iii) �2u,it = exp(w′

it�) wherewit is ag × 1 vector of explanatory variables ofthe variance of one-sided error term, and� is ag×1 vector of parameters,

(iv) vit anduit are mutually independent as well as independent ofzit .

The common practice after this step is to identify a parametric form of the costfrontier in Eq. (1). We choose the translog form because it imposes minimumapriori restrictions on the underlying production technology, and it approximatesa wide variety of functional forms. Considering that the specifiable factors ofproduction,6 are capital (K), labor (L) and deposits (D) andYL, YI andYA are thevalues of three categories of outputs, that is loans, investments and liquid assets,respectively, the cost function (1) can be written in translog form as follows

lnCrt =∑i

∑t

�∗itDt lnPirt + 1

2

∑i

∑j

�ijlnPirtlnPjrt

+ 1

2

∑i

∑j

�ijln Yi ln Yj +∑i

∑j

�iYln YirtlnPjrt

+∑

�∗t Dt ln Yirt +

∑t

�tDt + Urt + Vrt (2)

6 In this paper we follow the intermediation approach to banking, on the basis that it incorporatesall expenses (of which interest expenses are the most significant) and recognizes that deposits are moreaccurately inputs into banking activity rather than outputs (Berger, Leusner, & Mingo, 1997; Elyasiani& Mehdian, 1990). For the definition of the variables used, see theAppendix A.


with i, j = K, L, D, where�∗it = �i + �iA(t),�∗

t = �i + �iA(t), and�t = �o +A(t), Dt is a time specific dummy (t = 1,2, . . . , T ) andA(t) is an unobservablegeneral index of technical change. The main advantages of this index of technicalchange is that for constant prices and output there would not be an impact on techni-cal change as the commonly used trend models do, seeBaltagi and Griffin (1988).

For estimation purposes, the indexA(t) is scaled to zero in the initial year i.e.,A(1) = 0. By doing so, we are able to identify�o, �i , �i , as well as the indexA(t).

For the cost function (1) to satisfy the properties of neoclassical productiontheory, the following linear restrictions must be satisfied∑

i

�i = 1,∑i

�ij =∑j

�ji =∑i

�i =∑i

�ij = 0 (3)

Restrictions (3) imply that the cost function (2) is linearly homogeneous in factorprices.

If eit = vit + uit, the density function of the composed error term of the costfrontier is

p(eit) =(

2

�it

)�

(eit

�it

)�

(iteit

�it

)(4)

where� and� denote the pdf and cdf of the standard normal distribution,�2it =

�2v + �2

u,it, andit = �u,it/�v. The log-likelihood function can be written as

L(�,�,�v) =n∑i=1

T∑t=1

logp(eit) (5)

FollowingJondrow, Lovell, Materov, and Schmidt (1982), the conditional expectedvalue ofuit giveneit is

E(uit|eit) = sit

[eitit

�it+ �

(eitit

�it

)�

(eitit

�it

)−1]

(6)

wheresit = σvσu,it/σit.

4. Estimation results

The empirical investigation has been carried out using annual data from theBalance Sheet Accounts and Income Statements of the Greek banking system. Oursample coversall Greek commercial banks operating in Greece over the period1993–1998 (panel data). Thus, 19 banks are included in our sample in the period1993–1996; 18 banks in the year 1997 since one bank (Interbank) ceased operatingindependently and 17 banks in 1998 since yet another bank (Bank of Athens) ceasedindependent operation, both been absorbed by Eurobank.

The parameters ofEq. (2) were estimated by maximum likelihood. Estima-tion was accomplished using the algorithm described byBerndt, Hall, Hall, and


Hausman (1974)as implemented in TSP. To allow for heteroscedasticity in theone-sided error term several variables were included in the vectorz. However, onlythe number of bank branches was found to permit convergence of the iterative pro-cedure, although we have tried a great number of combinations of other variablesand several sets of initial conditions. Therefore, the number of bank branches isthe variable we choose to determine bank efficiency heteroscedasticity.

The estimated parameters of the heteroscedasticity model are reported inTable 2.The parameters in the vectorz, provide information on the presence of

Table 2ML estimates of the heteroscedastic translog model

Parameter Estimate t-ratio

�o −11.34∗ 1.62�L 1.70∗∗ 1.80�I −0.43∗ 1.64�A −0.20 0.42�D −1.44 0.71�L 5.93∗∗ 2.13�LL 0.11∗∗ 1.92�II 0.01∗∗ 1.84�AA −0.05∗∗∗ 3.00�DD −0.48 1.26�LL −0.57 0.79�DL 0.65∗ 1.38�LD −0.12 0.89�ID −0.03 −0.80�AD 0.16∗∗ 2.12�LL −0.49∗∗ 1.98�IL 0.10∗ 1.39�AL 0.13 1.02�LT 0.30 1.01�DT −0.12 0.74�LT −0.18 0.54�IT −0.03 0.50�AT 0.26 0.64A (2) −0.14 0.59A (3) 0.22 0.61A (4) 0.64 0.65A (5) 0.89 0.68A (6) 0.67 0.63

Likelihood 45.2570

Variance parameter estimateszo −5.10∗ 1.36z1 0.016∗∗ 1.94�2u 1.72∗∗∗ 15.46

Note: The signs (∗∗∗), (∗∗) and (∗) indicate statistical significance at 1, 5 and 10% level of signifi-cance, respectively.


heteroscedasticity. The slope coefficient is significantly different from zero, al-though marginally so. At first sight this may seem to imply an agnosticism regard-ing the presence or absence of heteroscedasticity. But we will see later on, thataccounting for heteroscedasticity has dramatic implications for efficiency mea-surement and efficiency dependence on certain banking characteristics. Therefore,from the economic viewpoint we cannot say that this agnosticism is correct.

Bank efficiency estimates for the Greek banking system for the period 1993–1998 are presented inFig. 1. Our results indicate that for the small and medium-sized banks the average efficiency indices over the 1993–1998 period are almost100%, while in large banks they range from 60 to 95% (Fig. 1a). Furthermore, forlarger banks it appears that efficiency is inversely proportional with size in termsof assets (see alsoTable 1). Thus, if large banks used their inputs as efficiently aspossible, they could reduce their production cost by roughly 5–45%. The overallpicture does not change if instead of the 1993–1998 averages, the year estimatesare considered (Fig. 1b), the levels of efficiency remaining relatively stable overthe liberalization period.

It seems to be the case that larger banks are not as efficient as their smaller coun-terparts. This fact could be attributed to traditional characteristics of the Greekbanking system, more pronounced in larger and mostly state-controlled banks,such as inefficient management, inadequate staff motivation and strict labor re-lations. Furthermore, large banks with an extended branch network burden thecost of bank operation the most since the sole function of the great majority ofbranches is still the mobilization of savings, while the offering of products (e.g.,loans and advances) is limited to only few central branches located in metropoli-tan areas. Note that, over the examination period, the growth of bank products inrelation to costs was faster in smaller banks compared to those of larger banks.However, the already initiated expansion of housing credit and consumer bank-ing activities resulting from market liberalization will enable bank branches toexpand production, thus making a better use of their production factors and im-prove on efficiency. This positive effect will be more pronounced in banks witha greater number of branches, placing them in an advantageous positionvis-à-vistheir smaller competitors. Nevertheless, substantial investment in technology andmarketing are essential to developing and selling these services.

However, the efficiency picture of the banking system changes if we adopt asimple homoscedastic approach to estimate bank efficiency instead of the het-eroscedastic frontier model used. The results derived from the homoscedasticmodel are presented inFig. 2.

As seen fromFig. 2, no substantial difference seems to exist between large andsmall banks in terms of efficiency and for the great majority of banks cost efficiencyranges between 85 and 95%. In other words, larger banks are as efficient as theirsmaller counterparts. This is due to the fact that not accounting for heteroscedas-ticity (when in fact we should) leads to overestimation of inefficiency of smallbanks and underestimation of inefficiency for large banks. Thus, efficiency mea-sures derived from homoscedastic frontiers in the presence of heteroscedasticity


Fig. 1. Results of heteroscedastic model.


Fig. 2. Results of simple homoscedastic model.


Table 3Regression analysis of technical efficiency

Parameters Coefficient t-value

Constant .665∗∗∗ 3.73Size −.44∗ 10−07∗∗∗ 6.16ROA .0083∗∗∗ 3.66I/Y .281∗ 1.67L/Y .347∗ 1.87R2 .81F(4, 117) 119

Note: The signs (∗∗∗), (∗∗) and (∗) indicate statistical significance at 1, 5 and 10% level of signifi-cance, respectively.The regression results are corrected for heteroscedasticity using the White procedure.

are biased. Note thatCaudill and Ford (1993)and Caudill, Ford, and Gropper(1995)arrived to comparable results for a sample of US commercial banks.

Having derived and discussed efficiency measures we now concentrate on pos-sible factors that are related to the efficiency of the Greek commercial bankingsector. To this end, we performed regression analysis aiming at establishing sta-tistical evidence between the bank-specific efficiency measures and various bankcharacteristics.

In particular, the efficiency measure derived from our heteroscedastic model isrelated to the bank size, bank performance and to the mixture of bank output. Thebank size (Size) is measured by the value of total deposits;7 bank performance isapproximated by the returns on assets variable (ROA); and the mixture of outputis measured by the share of the most profitable categories of bank products in totalbank output, that is loans to output (L/Y) and investments to output (I/Y).8

As can be seen fromTable 3, the independent variables account for a significantshare in the variation of the technical efficient index. In particular, the negativestrong correlation between the efficiency measure and bank size indicates that largebanks are less efficient than smaller banks. This finding contradicts the results ofother studies in which a positive correlation exists between efficiency and variousmeasures of bank size. See, for example, the study byFavero and Papi (1995)for the Italian banking sector and also the works ofAly, Grabowski, Pasurka, andRangan (1990)andMester (1996)for US banks. However, all these studies proceedunder the assumption that homoscedastic frontiers are valid.

Also, the ROA variable is strongly positively correlated with the efficiencyindex, indicating the expected strong relationship between the improvement of a

7 We also used alternative indicators of bank size, such as total assets and the number of bankbranches but they did not improve our results.

8 Note that liquid assets, mainly government securities, have not been profitable for banks overmost of the examination period since they were used to finance budget deficits on a compulsory basisand not competitive terms (seeSection 2).


bank’s economic performance and the achievement of higher efficiency. Note thatMester (1996)reached the opposite conclusion for US banks.

Finally, our results show that a greater share of investments in total bank output(I/Y) and especially a greater share of loans in total output (L/Y) are related tothe achievement of higher efficiency scores. This implies that the compositionof output has implications for bank efficiency. More specifically, it suggests thatthere may be “efficiency economies of scope” in those banks that have achieveda level of output diversification, are more efficient because they can utilize sharedresources to produce more outputs in a more efficient way. It also suggests thatdiversified banks may have better performing management because of the skills ittakes in order to produce and supervise the production of more financial products.This higher level of skills may be responsible for the observed positive associationbetween output composition and efficiency.

5. Conclusions

In this study we have examined cost efficiency for the Greek banking systemover the period 1993–1998, a period characterized by an acceleration of liberaliza-tion and deregulation of the financial system. We used a multi-input, multi-outputflexible cost function to represent the technology of the sector, and a heteroscedas-tic frontier approach to measure technical efficiency. The results suggest not onlythe presence of heteroscedastic inefficiency in the banking sector, but also thatignoring heteroscedasticity leads to misguided conclusions about the efficiency oflarge versus small banks.

Our empirical results showed that the small and medium-sized banks are almostfully efficient, while in large banks efficiency measures range from 60 to 95%.Furthermore, for larger banks it appears that efficiency is inversely proportionalwith the bank size in terms of assets. Thus, if large banks used their inputs asefficiently as possible, they could reduce their production cost by roughly 5–45%.It appears that over the period under review the large Greek banks enjoy fairlylow cost efficiency. This fact could be attributed to traditional characteristics ofthe Greek banking system, more pronounced in larger and mostly state-controlledbanks, such as inefficient management, inadequate staff motivation and strict laborrelations.

Furthermore, our empirical results showed that bank efficiency measures arerelated to various bank-specific economic factors. In particular, the improvement ofbank performance is strongly related to the achievement of high efficiency scores,while increased shares of investments and especially loans to total bank outputwere found to be quite important for the improvement of cost efficiency. Thus,efforts would be mainly directed to better economic performance. It seems that therole of these efforts is crucial for further improvements in bank efficiency and, toa large extent to creation of a competitive banking system. On the contrary, banksize is found to be inversely related to the efficiency index.


Over the second half of the 1990s, Greek banks started to reconsider theirstrategies and extensive restructuring in terms of mergers and acquisitions tookplace, in view of Greece entering the Euro-zone in 2001. The observed mergersand acquisitions, mostly between smaller banks over the reviewed period, maybe attributed to two main factors. The first is related to a wave of privatizationinitiated by direct government action, in line with EU developments. The secondfactor directing acquisitions has been market discipline, that is to obtain cost andefficiency gains. The latter is justified by the empirical findings ofChristopoulos,Lolos, and Tsionas (2000), since the great majority of banks involved in mergersand acquisitions exhibited increasing returns to scale.9 Since larger banks are lessefficient it would appear puzzling to find them engaged in mergers and acquisi-tions as buyers. However,Resti (1998), who analyzed 67 bank mergers in Italyobtained the same result. To explain this result, it should be noted that differencesin efficiencyas well as economies of scale and scope constitute the most impor-tant premerger cost incentives for acquiring another bank. If the acquiring bank ismore cost efficient than the acquired bank, there is an incentive for merger, sincetransferring management skills to the acquired bank can create additional profit.

The EMU entry is affecting the Greek banking market in several respects. Incr-eased cross-country competition and the intensification of competition in the dome-stic market has put pressures on bank profitability and led to a reduction in interestrate margins, although they still remain high in relation to the EU average. Also, thegradual reduction of interest rates along with specific government actions are lead-ing to increased lending to the private sector and the offering of new differentiatedproducts. These actions, including the reduction of the required bank deposits at theCentral bank, in line with the EU harmonization of the minimum reserve require-ments ratios, and the abolition of the banks’ obligation to keep a high proportionof their portfolio in government securities for the financing of public deficits, ledto increased liquidity of banks. In addition, market liberalization enabled banksfor a greater utilization of the Interbank market. Thus, restructuring of the assetand liability sides of banks’ balance sheet is taking place and converges to the pat-tern characterizing the EU markets. Finally, bank disintermediation also occurs,although bank financing is still the most dominant source of enterprises’ funding.

In the years to come it is expected a new round of mergers, acquisitions, con-glomeration and strategic alliances in the Greek banking market. Greek bankswill be aiming at repositioning in the EMU markets and achieving the necessarylarge-scale operations in order to share the high information technology costs andto reap economies of scale.10 They will also try to exploit market segments within

9 The cost reduction impact of this restructuring of the Greek banking system, strongly expectedby most officials of banks involved will come off in future (To Vima, 1999).

10 Indicative of future developments and prospects in the Greek banking system is the recentco-operation agreement between the largest private bank, the largest state-controlled bank and thenational telecommunications organization in an effort to use effectively the most up-to-date informa-tion technology.


the EU or in regional markets (e.g., in the Balkans), but they will be constrainedto achieve a more efficient use of resources owing to their small size.

Acknowledgments

This paper has greatly benefited from discussions with Dionyssis Christopoulosand Yannis Filippakis of the Bank of Greece. Thanks are also due to an anonymousreferee. However, all errors and omissions remain ours.

Appendix A

All variables are defined according to the Consolidated Banking AccountingSystem.

The three output variablesYL (loans)YI (investments) andYA (liquid assets)are defined in real terms as follows:

• YL: loans and advances. Include short-term and long-term loans and advances toindustry and customers; also doubtful and in arrears (not excluding provisions).

• YI : investments. Include shares and other variable-income securities and partic-ipation in affiliated and nonaffiliated companies.

• YA: liquid assets. Include all investments in fixed income securities as well asgovernment securities.

The three input variablesL (labor),K (capital) andD (deposits) are as follows:

• L: labor. Defined as the total number of full-time employees.• K: capital. Defined as fixed assets, including tangible fixed assets (land, lots,

buildings and installations, furniture, office equipment, etc. less depreciation),as well as intangible fixed assets (goodwill, software, restructuring expenses,research and development expenses, minority interests, formation expenses,underwriting expenses, etc.).

• D: total deposits. Include bank bonds and sight, saving, time and restricteddeposits of the private and the public sector.

The unit prices of the three respective inputs (PL, PK andPD) are defined asfollows:

• PD: interest expenses to total deposits (unit price of deposits). Interest expensesinclude interest paid on deposits and commission expenses and payments underLaw 128/175.11

11 Law 128/175 refers to obligations of commercial banks to finance priority sectors at preferentialinterest rates. Note that these are payments to the Bank of Greece, but they are also included in theinterest income as revenues.


• PL: personnel expenses to total labor (unit price of labor). Personnel expensesinclude wages and salaries, social security contributions, contributions to pen-sion funds and other related expenses.

• PK : capital expenses to fixed assets (unit price of capital). Capital expenses referto depreciation expenses on a historical cost-basis balance sheet.

• BR: number of branches to total number of branches. The ratio of a bank’snumber of branches to the number of branches of the whole banking system.

• ROA: net profits to total assets (return on assets).Net profits is defined as grossoperating profit less total operating results. Gross operating profit is definedas the sum of net interest revenue (interest revenue minus interest expenses)and noninterest revenue (participation revenue, revenue from trading portfolio,commissions, foreign currency transactions revenue, net capital gains and otherrevenue). Total operating results include wages and personnel expenses, gen-eral expenses, depreciation expenses and provisions.Total assets include cashin hand and deposits with the Bank of Greece, government and other securi-ties acceptable for refinancing with the Bank of Greece, loans and advances tocredit institutions and customers less provisions, bonds and other fixed incomesecurities, shares and other variable-income securities, participation in affiliatedand nonaffiliated companies, intangible assets, tangible fixed assets and otherassets, as well as prepaid expenses and accrued income.

References

Aly, H. Y., Grabowski, R., Pasurka, C., & Rangan, N. (1990, May). Technical scale and allocativeefficiencies in US banking: An empirical investigation.The Review of Economics and Statistics,211–218.

Baltagi, H. B., & Griffin, J. M. (1988). A general index of technical change.Journal of PoliticalEconomy, 96, 21–41.

Berger, N. A., Leusner, H. J., & Mingo, J. J. (1997). The efficiency of bank branches.Journal ofMonetary Economics, 40, 141–162.

Berndt, E. R., Hall, B. H., Hall, R. E., & Hausman, J. A. (1974). Estimation and inference in nonlinearstructure models.Annals of Economic and Social Measurement, 4, 653–665.

Caudill, S. B., & Ford, J. M. (1993). Biases in frontier estimation due to heteroscedasticity.EconomicsLetters, 41, 17–20.

Caudill, S. B., Ford, J. M., & Gropper, D. M. (1995). Frontier estimation and firm-specific ineffiicencymeasures in the presence of heteroscedasticity.Journal of Business and Economic Statistics, 13,105–111.

Christopoulos, K. D., Lolos, S. E., & Tsionas, E. G. (2000, January).Efficiency, productivity growthand cost structure of the Greek banking system on the threshold of EMU. Panteion UniversityDiscussion Paper.

Elyasiani, E., & Mehdian, S. (1990). Efficiency in the commercial banking industry: A productivityapproach.Applied Economics, 22, 539–551.

Favero, C. A., & Papi, L. (1995). Technical efficiency and scale efficiency in the Italian banking sector:A nonparametric approach.Applied Economics, 27, 385–395.

Hondroyiannis, G., Lolos, S. E., & Papapetrou, E. (1998). Market structure and performance in theGreek banking system. In J. J. Choi & J. A. Doukas (Eds.),Emerging capital markets: Financialand investment issues (pp. 179–191). Westbit, CT: Quorum Books.


Hondroyiannis, G., Lolos, S. E., & Papapetrou, E. (1999). Assessing competitive conditions in theGreek banking system.Journal of International Financial Markets, Institutions and Money, 9,377–391.

Jondrow, J., Lovell, C. A. K., Materov, I., & Schmidt, P. (1982). On the estimation of technical ineffi-ciency in the stochastic frontier production model.Journal of Econometrics, 19, 233–238.

Mester, J. L. (1996). A study of bank efficiency taking into account risk preferences.Journal of Bankingand Finance, 20, 1025–1045.

Noulas, A. (1997). Productivity growth in the Greek banking industry: State versus private banks.Applied Financial Economics, 7, 223–228.

Resti, A. (1998). Regulation can foster mergers. Can mergers foster efficiency? The Italian case.Journalof Economics and Business, 50, 157–169.

To Vima. (1999, December 5). New income sources for Greek banks.To Vima, 27.Tsionas, E. G. (2000). P-STAR analysis in a converging economy: The case of Greece.Economic

Modelling, 18, 46–60.

Documents

Efficiency of the Greek banking system in view of the EMU: a heteroscedastic stochastic frontier approach