EFFICIENCY OF BANKS: RECENT EVIDENCE FROM THE TRANSITION

ECONOMIES OF EUROPE –1993-2000*

H. Semih Yildirim**

College of Commerce
University of Saskatchewan
Saskatoon, SK S7N 5A7, Canada
Tel.: +1 (306) 966-2503
Fax: +1 (306) 966-2515
E-mail: yildirim@commerce.usask.ca

and

George C. Philippatos

432 Stokely Management Center

Department of Finance
University of Tennessee
Knoxville, TN, 37996-0540, USA,
Tel.: +1 (865) 974-1719,
Fax: +1 (865) 974-1716
E-mail: gphilip@utk.edu

Original Version: August 2001

Revised: April 2002
Revised: October 2003

* Earlier versions of this paper were presented various university workshops, and the 2002 EFMA Conference in
London, England, June 26-29, 2002. We wish to thank to R. E. Shrieves, M. C. Collins, H. Chang, G. Koutmos,
S. Ouandlous, Henri Pages, and various participants for their constructive comments and suggestions on earlier
versions of this paper. Financial support for data acquisition and summer research from the University of
Tennessee is also acknowledged.
** Corresponding author.

1
 EFFICIENCY OF BANKS: RECENT EVIDENCE FROM THE TRANSITION
ECONOMIES OF EUROPE –1993-2000

Abstract

This study examines the cost and profit efficiency of banking sectors in twelve transition
economies of Central and Eastern Europe (CEE) over the period 1993-2000, using the
stochastic frontier approach (SFA) and the distribution free approach (DFA). The managerial
inefficiencies in CEE banking markets were found to be significant, with average cost
efficiency level for 12 countries of 72 and 77 percent by the DFA and the SFA, respectively.
The alternative profit efficiency levels are found to be significantly lower relative to cost
efficiency. According to the SFA, approximately one-third of banks’ profits are lost to
inefficiency, and almost one-half according to the DFA. The results of the second-stage
regression analyses suggest that higher efficiency levels are associated with large and well-
capitalized banks. The degree of competition has a positive influence on cost efficiency and a
negative one on profit efficiency, while market concentration is negatively linked to
efficiency. Finally, foreign banks are found to be more cost efficient but less profit efficient
relative to domestically owned private banks and state-owned banks.

Keywords: Banking; Efficiency; Transition Economies; Stochastic Cost Frontier.

JEL Classification: G21, P20

2
 EFFICIENCY OF BANKS: RECENT EVIDENCE FROM THE TRANSITION
ECONOMIES OF EUROPE –1993-2000

1. Introduction

Over the last decade the banking markets in the transition economies of Central and

Eastern Europe (CEE) have gradually evolved from the traditional monobank system of the

central-planning period to a western-style, geographically and sectorally diversified, two-

tiered system of today. The governments, with financial and strategic support from

international organizations, spent enormous amounts of effort to develop a competitive and

efficient banking system, based on market principles, for the transformation from central

planning to market economies and for converging toward the criteria for European

integration.1 New financial markets, institutions and channels of intermediation have been

established almost from scratch. Although many prudential regulations have been enacted and

supervisory systems have been formed to smooth the transformation and sustain the financial

stability, promoting banking sector efficiency still remains an important concern for many of

the CEE governments and the various international organizations that support the transition

process.

The performance of CEE banks is of research interest and policy relevance due to

several reasons: First, the importance and dominant role played by banks in the provision of

intermediation services and the capital formation process in transition economies. Since their

money and capital markets are still in a developing stage, the transition countries have

primarily bank-based financial systems and performance of the banking system plays a key

1
The first group of CEE countries including the Czech Republic, Estonia, Hungary, Poland, and Slovenia
started negotiations with the EU in March 1998 while the second group of CEE countries–Bulgaria, Latvia,
Lithuania, Romania, and Slovakia – began negotiations in February 2000

3
role on economic growth and sound functioning of the industrial sectors. Secondly, the

competitive structure of the financial sector has significantly changed due to recent

deregulation and liberalization, accompanied by large-scale privatization and foreign

participation.2 Thanks to less restrictive licensing policies, the number of foreign and

domestically owned private banks has substantially increased in transition countries over the

recent years. These developments in the region certainly have some efficiency implications

for operating performance and productivity in the financial systems. Finally, as the banks and

other financial institutions in the region are preparing for increased competition under the

European Economic and Monetary Union (EMU), it is important to evaluate their ability to

compete and survive in an integrated European financial landscape. Therefore, for both

managerial and public policy concerns, it is extremely important to know the efficiency level

of banking operations in determining how the banking industry will respond to these

challenges and which banking firms are likely to prevail during the ongoing transition and

integration phase.

In light of the above discussion, our study evaluates the efficiency levels of

commercial banks in 12 CEE countries attained through the recent liberalization,

restructuring, and deregulation process. Although research on the efficiency of banks and

other financial institutions in the US and other developed countries is voluminous, there has

been very little previous research that focus on transition economies. Our study intends to fill

this gap. Despite the remarkable structural changes in transition economies this limited

academic attention can partly be explained by the unavailability of reliable data until recently.

Our study utilizes unbalanced panel data from 12 CEE countries over the period 1993-2000

2
The share of foreign ownership in terms of both total assets and capital currently exceeds two-thirds in many of
these countries, making the CEE banking sector the most open emerging market to foreign participation.

4
and employs the “stochastic frontier approach” (SFA) and the “distribution free approach”

(DFA) to estimate bank-specific efficiency relative to the predictive cost and profit functions

of banks that are efficient. We also analyze the potential correlates of efficiency by regressing

the inefficiency scores on numerous bank-specific and market-structure variables. Given the

current restructuring and deregulation efforts in these countries, the empirical results of this

study should be timely and helpful to the policymakers of the above countries, as well as to

banking scholars.

The remainder of this paper is organized as follows. Section 2 briefly reviews the

efficiency concepts and some salient literature. Section 3 discusses the data and methodology

employed. Section 4 presents the estimates of efficiency and Section 5 summarizes the results

and inferences of the study.

2. Efficiency Concepts and Review of Relevant Literature

The performance of depository financial institutions has been the subject of many

empirical studies, after the seminal works of Benston (1965) and Bell and Murphy (1968).

Early research on bank performance has given the most emphasis to the analysis of scale and

scope economies employing a cost function where all banks are implicitly assumed to operate

at approximately the same efficiency levels. Research on scale efficiency in the banking

industry usually detects a relatively flat U-shaped average cost curve and reports higher

efficiency levels for medium-sized banks than either large or small banks (Berger et al.,

1999).

However, the academic research focus on bank performance has recently shifted to

frontier efficiency or X-efficiency, which measures inefficiency as the deviation from the

5
efficient frontier where best-practice firms operate. That is, the cost efficiency of an institution

is measured by its performance relative to an estimated performance of the best firm

characterized by minimum costs or input usage in the industry, holding scale and output mix

constant. Prior research concludes that cost inefficiencies in the range of 20 to 30 percent

dominate the effect of scale and scope diseconomies, generally found to be in the range of 5 to

10 percent (Berger and Humphrey 1991). Yet, there is no consensus among the researchers on

the efficiency concept, functional form, and estimation technique that yield the most accurate

efficiency measure.

The extant literature has employed various efficiency concepts and measurement

methods to estimate operational efficiency and to analyze the cross sectional determinants of

efficiency differentials across banks. With regard to estimation techniques, measuring

efficiency through the estimation of stochastic frontiers has received increasing attention

among the researchers over the last two decades. Literature offers numerous parametric and

non-parametric methods for the empirical estimation of such a frontier. The parametric

methods involve defining the frontier through a functional form such as Translog, which is

estimated by several econometric techniques. The non-parametric methods, on the other hand,

do not presume any explicit functional form for the frontier and construct it from the observed

input-output ratios using mathematical programming techniques. The non-parametric

methods, by not allowing any shocks to production and cost, consider any deviation from the

frontier as inefficiency. Due to this restrictive assumption, stochastic frontiers based on a

composed error model, which allows us to distinguish between inefficiency and other

stochastic shocks, are considered superior to non-parametric frontiers in measuring efficiency.

6
 The literature on the efficiency of financial institutions in the US and other well-

developed countries contains a large number of articles. Since this is not intended to be a

comprehensive review, we refer the readers to a handful of surveys in this respect.3

Nevertheless, it is beneficial to focus directly on some relevant studies that provide some

insights on cross-country comparisons of efficiency.

In an application of frontier efficiency approach to international data, Allen and Rai

(1996) employ both the "stochastic frontier" (SFA) and the "distribution free" (DFA)

approaches to estimate a standard translog functional form for 15 developed countries over the

period 1988-1992. Initially they divide their sample into two groups of banks based on

whether a bank is headquartered in a country which allows the practice of “universal” or

“functionally-separated” banking, and then they further classify each group into "large" and

"small" sub-samples. They report that the cost inefficiency measures range from 15 percent

for large banks in universal banking countries to 27.5 percent for large banks in functionally-

separated banking countries. On average, functionally separated banks are found to be

operating less cost efficiently than their universal counterparts. They also find that, greater

efficiency is associated with higher profitability, lower total cost, smaller bank size and higher

level of loans.

Vander Vennet (1996) studies the effect of domestic and cross-border mergers and

acquisitions on the efficiency of credit institutions in Europe over the period 1988-1993 and

finds that cross-border acquisitions improve the efficiency levels of acquiring banks while

domestic consolidation does not. Bikker (1999) applies the stochastic cost frontier approach

to the European banking industry in order to measure the effect of increased competition on

3
See, among others, Berger, Hunter and Timme (1993), Berger and Humphrey (1997), and Berger and Mester
(1997) for an extensive review of literature on efficiency of financial institutions.

7
bank efficiency. He finds that, on average, Spanish, French and Italian banks appear to be less

efficient than those in Germany, the Netherlands and the UK, while banks in Luxembourg,

Belgium and Switzerland are the most efficient. Berg et al. (1993) use data envelopment

analysis to study the efficiency in Scandinavian banking markets and find the average

Swedish bank (78%) to be more efficient than the average Norwegian bank (57%), which, in

turn, is more efficient than the average Finnish bank (53%). Finally, Maudos et al. (2002)

analyzes the cost and profit efficiencies of banks for ten countries of The European Union by

employing panel data frontier approaches. They report average cost and profit efficiency

levels of 82.7% and 45% respectively, for ten countries considered. In their study, Austria and

Germany emerge as the most cost efficient countries while Luxembourg and Portugal stand

out as the most profit efficient.

In their extensive international literature survey, Berger and Humphrey (1997)

emphasize the fact that, out of 130 efficiency analyses of depository financial institutions,

covering 21 countries, only about 5 % examine the banking sectors of developing countries.

They note that the vast majority of the efficiency literature focus on the banking markets of

well-developed countries with particular emphasis on the U.S. markets (about 75%). The

relatively scant literature on bank efficiency in emerging markets focuses mainly on the

efficiency differentials among banks with different ownership status and asset size. This is

due to the fact that the banking systems in these countries are still in their infancy and markets

are usually characterized by high state-ownership, newly privatized domestic banks, and rapid

entry by foreign banks. The policy issues examined in these studies usually address the

questions regarding privatization of state-owned banks, elimination of restrictions for

8
domestic and foreign bank entry and operations, and existence of scale economies associated

with mergers and acquisitions.

Regarding the effect of ownership on bank performance, previous research produced

contradicting results between developed and emerging banking markets. Domestic banks in

developed countries are generally found to be more efficient relative to their foreign-owned

counterparts (Berger et al., 2000). For example, prior research has found that domestically-

owned U.S. banks are substantially more cost efficient than are foreign-owned banks (Hasan

and Hunter, 1996; Mahajan et al.,1996; Chang et al., 1998). These results are consistent with

the notion that foreign banks with a driving motive of obtaining a quick market share

expansion will rely heavily on purchased funds, a relatively more costly way of financing

their investments compared to core deposits, which require setting up an extensive delivery

capacity and establishing a broad customer base (DeYoung and Nolle, 1996). An alternative

explanation is provided by Peek et al. (1999) who conjecture that the observed inefficiency of

foreign banks that enter the U.S. market through acquisition could be attributed perhaps to the

lower performance and efficiency of target banks compared to other domestic banks prior to

the acquisition. In contrast, the efficiency studies in emerging banking markets generally find

that foreign banks are more efficient than either state-owned banks or domestic private banks.

For example, Bhattacharya et al. (1997), Srivastava (1999) for Indian banks, Hasan and

Marton (2000) for Hungarian banks, and Isik and Hassan (2002) for Turkish banks find that

foreign banks are more cost efficient than domestic banks. Similarly, Ozkan-Gunay (1998)

and Hasan and Marton (2000) find foreign banks are more profit efficient than domestic and

state-owned banks in Turkey and Hungary, respectively. Claessens et al. (2001) find that

foreign banks in emerging markets are more profitable than their domestic counterparts.

9
Leightner and Lovell (1998) and Laeven (1999) also find that foreign banks exhibit higher

operational efficiency in the Thai banking system.

Regarding bank size, Hasan and Marton (2000) show that larger banks are more cost

and profit efficient than smaller banks in Hungary. Similarly, Srivastava (1999) reports the

existence of economies of scale in the Indian banking market. Leaven (1999) presents

evidence that small Thai banks are less profit efficient than medium and large banks. By

contrast, Isik and Hassan (2002) find a negative relationship between banks size and cost

efficiency for Turkish banks. Accordingly, Leightener and Lovell (1998) find that size is

negatively associated with productive efficiency for both domestic and foreign banks in

Thailand.

3. Empirical Design for Efficiency Estimation and Database

Our analysis comprises two stages. In the first stage, we employ the standard translog

specification to obtain efficiency estimates for individual banks in the sample. Since the

production technologies of banks are unknown a priori, we estimate the efficiency measure as

the deviation from the efficient frontier where best-practice firms operate. In this approach, a

frontier is established from the estimated cost or profit function of banks in the data while the

inefficiencies or deviations from the frontier are represented by the error terms. In the second

stage, we explore the determinants of efficiency with certain bank-specific and industry-

specific factors by a series of regression analyses. We evaluate the performance of CEE

banks using two different optimization concepts—cost minimization and alternative profit

maximization. In this respect x-efficiency refers to the degree of managerial success on using

inputs and outputs in a manner that will minimize costs and maximize profits. Under cost

minimization, the inefficiency arises from sub-optimal choices of input quantities given input

10
prices and output quantity, whereas under profit maximization, the inefficiency originates

from sub-optimal choice of output quantities given output prices or sub-optimal output prices

given quantities. The following sub-section outlines the basic model, which is estimated under

two different frontier specifications.

3.1 The cost and profit frontiers

Cost efficiency scores measure the performance of a banking firm relative to the best-

practice bank that produces the same output bundle under the same exogenous conditions. The

cost frontier is derived by estimating the following cost function:

C =C (y,w,z,u,e) (1)

where, C measures total costs for bank, including both operating and financial costs; y is a

vector of outputs; w is a vector of input prices; z represents the quantities of fixed bank

parameters (bank capital, fixed assets, off-balance sheet items, etc.); u is the inefficiency term

that captures the difference between the efficient level of cost for given output levels and

input prices and the actual level of cost; and e is the random error term. Assuming the

inefficiency and random error term are multiplicatively separable from the rest of the

parameters, the cost function can be expressed in logarithmic form as:

ln C= f(y,w,z) + ln u + ln e (2)

where f denotes a functional form. After estimating a particular cost function, the cost

efficiency for bank i is measured as the ratio between the minimum cost (Cmin) necessary to

produce that bank’s output and the actual cost (Ci):

Cmin exp[ f ( y, w, z )]x exp(ln umin ) umin

COSTEFFi = = = (3)
Ci exp[ f ( y, w, z )]x exp(ln ui ) ui

11
where umin is the minimum ui across all banks in the sample. Under this formulation, an

efficiency score, say 0.90, implies that the bank would have incurred 90 percent of its actual

costs had it operated in the cost frontier.

Profit efficiency measures how close a bank is to attaining the maximum possible

profit as a best-practice firm on the frontier for given levels of input and output prices

(quantities) and other exogenous market variables. Previous literature offers two different

specifications for the profit maximization objective, namely “standard” and “alternative”

(non-standard) profit functions. (Humphrey and Pulley, 1997; Berger and Mester, 1997). The

standard profit function assumes that output markets are perfectly competitive so that banks

are price-takers in both output and input markets while alternative profit specification assumes

that banks can have some power in determining output prices. Thus, standard profit function

is specified as a function of input and output prices, whereas alternative profit function is

specified as a function of input prices and output quantities.

Berger and Mester (1997) note that the alternative profit specification is preferred over

standard profit specification when (1) there are differences in the quality of banking services,

(2) markets are not perfectly competitive so that banks might have some market power in

pricing their outputs, (3) outputs are not completely variable, so that banks can not achieve

every output scale and product mix, and (4) output prices are not available. Since banks in our

sample belongs to a diverse group of countries that differ in market structure, regulation, and

quality of banking services provided, it will be more appropriate to employ alternative profit

specification for cross-country comparisons. Furthermore, perfectly competitive markets

assumption of standard profit function does not reflect the realities of transition banking since

12
markets are still considered to be in their developing stage.4 Finally, specifying profits as a

function of output quantities rather than output prices avoids the problem of having to

measure output prices, which are basically not available in our case.

The alternative profit specification employs the same set of exogenous variables as the

cost function in Equation (1) with the only difference that profit replaces total cost as the

dependant variable in the frontier regression. Therefore, the alternative profit frontier is given

by

P = P ( y,w,z u, e) (4)

where P is the variable profits of the firm, which includes all the interest and fee income

earned less total costs, C, used in the cost function. The profit function can be written in log

terms:

ln (P+θ) = f(y,w,z) + ln e - ln u (5)

where θ is a constant added to every bank’s profit to make it positive so that the natural log

can be taken. Profit efficiency is measured by the ratio between the actual profit of a bank and

the maximum possible profit that is achievable by the most efficient bank.

Pi exp[ f ( y, w, z )]x exp(ln ui ) − θ

PROFEFFi = = (6)
Pmax exp[ f ( y, w, z )]x exp(ln umax ) − θ

where umax is the maximum ui across all banks in the sample. For example, profit efficiency

score of a bank, say, of 80% means that the bank is losing about 20% of its potential profits to

managerial failure in choosing optimum input quantities and outputs prices.

4
This assumption is not also consistent with the findings of Yildirim and Philippatos (2003a) that banks in
majority of the CEE countries operated under monopolistic competition over the years 1993-2000.

13
3.2 Estimation Techniques

The biggest challenge in frontier efficiency studies has been separating inefficiencies

from random error. Four main frontier approaches, each with different sets of arbitrary

assumptions on the probability distributions of inefficiency and random error terms, have been

utilized. These approaches are (a) data envelopment analysis (DEA), (b) thick frontier

approach (TFA), (c) stochastic frontier approach (SFA), and (d) distribution free approach

(DFA).5 Berger, Hunter, and Timme (1993) report that there is no simple rule for selecting

which of these methods can be best used to describe banking data and these methods do not

produce consistent estimates of inefficiencies.

The DEA approach is a non-parametric linear programming technique, where the set

of frontier observations are those using proportionally less of each input for a given set of

outputs and using the correct proportion of inputs given their respective prices. The DEA,

which focuses primarily on technical efficiency, does not allow for random errors, assuming

that all deviations from the estimated frontier represent inefficiency. Under the TFA, a lowest

and highest cost quartiles of banks are estimated from a cost function. The maintained

assumption here is that the error terms within the lowest and highest cost quartiles reflect

statistical noise, while the differences between the lowest and highest cost quartiles represent

inefficiency measures (Berger and Humphrey, 1991).

In order to measure the efficiency level of CEE banks, this study employs two

different parametric techniques: the stochastic frontier approach (SFA) and the distribution

free approach (DFA). The SFA asserts that managerial or controllable inefficiencies can only

increase costs (reduce profits) above (below) best-practice frontier and that random

5
See Bauer et al. (1997) for a comprehensive review of these approaches.

14
fluctuations or uncontrollable factors can increase or reduce costs (profits). Therefore, the

model assumes that inefficiency measures, (ln u), which represent the departure from the

efficient frontier follow an asymmetric half-normal distribution, while random fluctuations are

distributed as two-sided normal with a zero mean and variance σe2.6 Jondrow et al. (1982)

propose a model, which specifies a functional form of the distribution of the one-sided

inefficiency component, and derive the conditional distribution (ui | ui + ei).

The DFA tries to avoid the arbitrary assumptions of the stochastic frontier approach,

where panel data are available. This approach also separates the composite error term into

inefficiency and statistical noise components. However, it assumes that there exists a core

inefficiency for banks, which persists over time while the random error part vanishes out over

time (Berger, 1993). According to the DFA, inefficiency estimate of a bank is determined by

the difference between average residual of the bank i, (lnui), and the average residual of the

bank on the frontier (lnumin), assuming that the random errors will cancel out over time. The

estimated average residual is then transformed into a measure of efficiency as follows:

EFFi = exp (lnumin - lnui) (7)

where lnumin is the average residual for the bank with the lowest average cost residual. The

most efficient bank is represented by an efficiency score of 1 and the other banks are assigned

efficiency scores between 1 and 0. Under this framework, an efficiency score (EFF) of 0.90

implies that the bank would have incurred 90 percent of its actual costs had it matched its

operations with those of the best-practice bank.

6
Economic theory does not say much about the distribution of the inefficiency component of composite error
term. Various distributions have been proposed in the SFA literature for ui, which can only be observed
indirectly. The most widely used distribution assumption for ui is half normal distribution, which is also
employed in this study. Alternate model specifications include the exponential distribution and the gamma
distribution.

15
 Berger (1993) emphasizes the fact that the random error component, ln e, may not

cancel out completely during the sample period. Therefore, some banks might have extreme

average residuals in both ends of the distribution. To mitigate the problem caused by the

extreme values, prior research calculated additional efficiency measures derived from

truncated distributions of the ln ui. That is, observations that fall below the pth percentile are

set equal to the pth percentile value, and observations that exceed the (1-p)th percentile are set

equal to the (1-p)th percentile value. Keeping up with the literature, we truncate the

distribution at the three values of p: 1%, 5% and 10%.

3.3 Functional form

We employ the multiproduct translog functional form to estimate the cost and

alternative profit frontiers and derive the efficiency measures. The cost frontier function is

represented by:

2 2 2
ln(C / w3 z) = α0 + ∑αl ln(wi / w3 ) + 0.5 ∑ ∑ω lh ln(wl / w3 ) ln(wi / w3 )
l =1 l =1 h =1
3 3 3
+∑ β k ln(yk / z) + 0.5 ∑ ∑β kj ln(yk / z) ln(y j / z)
k =1 k =1 j =1
3 2
+ ∑ ∑δ
k =1 l =1
lk ln(yk / z) ln(wl / w3 ) + ϕ1 ln Z + 0.5ϕ2 (ln Z )2
3 2
+ ∑ τ k ln(yk / z) ln Z ) + ∑ ς l ln(wl / w3 ) ln Z + ln eti + ln uit (8)
k =1 l =1

where wi and yi are input prices and output amounts and z is the equity capital. The

dependent variable, total cost, is the sum of interest expenses, personnel expenses and other

16
operating expenses.7 We impose the regular restrictions of symmetry and linear homogeneity

for input prices in estimating the parameters of Equation (8) as the following:

3 3 3
β kj = β jk , ωlh = ω hl ; ∑α l = 1,
l =1
∑ωlh = 0 ,
h =1
∑δ
l =1
lk = 0. (9)

Cost and input prices are normalized by the price of capital before taking logarithms to

impose linear input price homogeneity. Since we do not decompose the efficiency measure

into technical and allocative components, the cost functions are not estimated using the input

share equations.

The alternative profit frontier estimation employs essentially the same specification in

cost equation with some minor changes. For the profit frontier estimation the dependent

variable ln(C / w3 z ) is replaced with ln( P / w3 z ) and the inefficiency term is -u. Cost, profit

and output variables are normalized by equity capital (Z). This normalization controls for

heteroscedasticity, scale biases, and other estimation biases in addition to providing a more

economic meaning since the dependant variable in profit function essentially becomes ROE, a

common measure of performance.8

Still a controversial issue in banking literature is the definition of banking inputs and

outputs, and more specifically the treatment of deposits in this respect. The two competing

approaches that determine what constitutes inputs and outputs of banks are the production and

7
Some authors assert that fitting a single translog function over a sample of banks that vary widely in size and
product mix will create a specification bias. They propose the use of non-parametric techniques such as kernel
regression, spline-augmented, or Fourier flexible approximations. (See McAllister and McManus 1992 , Mitchell
and Onvural 1996, Berger, Leusner, and Mingo 1997) However, Berger and Mester (1997) find that choosing a
fourier flexible form over standard translog model does not have important effect on the measure of average
industry efficiency or on the rankings of individual banks. They report that the average efficiency measures are
only 1 percent lower when Fourier specification employed. The limited number of observations in the present
data prevents us from employing Fourier approximations. One approach to mitigate this problem would be to
estimate separate cost functions over subsets of the firms in the sample (See Dowling and Philippatos, 1990).
8
For detailed discussions of this normalization see Berger and Mester (1997,1999) and Berger and DeYoung
(2001).

17
the intermediation approaches (Sealey and Lindley, 1977). Production approach considers

labor and physical capital as inputs and number of processed accounts as outputs. The

intermediation approach considers deposits as inputs (and thus considers interest on deposits

as a component of total costs, together with labor and capital expenses) and defines loans and

investments as outputs. We adopt the intermediation approach and following Berger and

Humphrey (1992) and specify deposits as both inputs and outputs of banks. Accordingly, in

this study, banks are modeled as multi-product firms that produce 3 outputs (loans,

investments, and deposits) and employ 3 inputs (borrowed funds, labor, and physical capital).9

Loans (Y1) are measured as the sum of all loan accounts intermediated by banks less non-

performing loans.10 Investments (Y2) is the sum of total securities, equity investments and

other investments. The third output (Y3) is the produced deposits (the sum of demand,

savings, and time deposits). These outputs are produced by using three inputs: borrowed funds

(X1), labor (X2), and physical capital (X3). The price of borrowed funds (W1) is estimated as

interest expenses divided by customer and short term funding plus other funding. The price of

labor (W2) is defined as the ratio of personnel expenses to total assets.11 The price of physical

capital (W3) is measured as the ratio of other operating expense to fixed assets. Table 1

provides the summary statistics for model parameters in thousands of US dollars for 2000 and

12 countries.12

9
For each output category, zero values are replaced by a small positive number 0.001 to allow logarithms to be
taken, as in several previous studies.
10
Due to the unavailability of asset-specific data we cannot disaggregate loans into their various components.
11
Due to the unavailability of data on the number of employees we cannot employ the ratio of personnel
expenses to number of workers as unit price for labor. Using the ratio of personnel expense to total assets as
labor cost is a common approach in efficiency studies that employ BankScope data.
12
Similar statistics for individual countries are not reported here to save space but available from authors upon request

18
 The equity capital (Z) is included in the model for several reasons. First, it controls for

managerial risk preferences in solving maximization and minimization problems. In order to

absorb financial shocks, risk-averse managers might hold higher level of equity than that of

cost minimizing level. Therefore, failure to account for risk preference might lead to label

otherwise optimally behaving bank, given its managerial risk preference, as inefficient.

Furthermore, higher level of equity implies lower default risk, all else being equal. The default

risk in return, affects the cost and profit of the bank through the risk premium the bank has to

pay for borrowing (Mester, 1996). Given the reality of high insolvency risks due to significant

non-performing loans, including equity becomes very important for the study of transition

banking. Berger and Mester (1997) also notes that inclusion of equity in the analysis may

account for the scale bias arising from differences in bank dependence on borrowed funds

since large banks usually depend more on borrowed funds. Finally, equity constitutes an

alternative to deposits in funding loans and investments, and therefore can significantly affect

costs and profits. Since equity financing is generally the most costly way of funding assets,

neglecting the equity level will bias the efficiency scores in favor of banks with high reliance

on equity.13

3.4 Database

The banks in our sample constitute a fairly large proportion of the banks in the

transition economies of Europe over the period of 1993-2000. Financial statement data for the

banks were taken from Fitch-IBCA's BankScope date set. The countries included are the

13
Since equity values reported for some banks were negative we applied the following transformation. The
absolute value of the minimum equity value in the sample plus 1 is added to each bank’s equity so that the
logarithm is defined.

19
Czech Republic, Estonia, Croatia, Hungary, Latvia, Lithuania, FYR of Macedonia, Poland,

Romania, Slovenia, the Slovak Republic, and the Russian Federation. Albania, Bulgaria, and

Yugoslavia were excluded form the initial sample due to lack of complete information. The

initial sample consisted of 2364 observations on 562 financial institutions. To be included in

the final sample, banks had to be classified as commercial banks or cooperative banks in the

BankScope data set and they must have all the model variables available at least for three

years. Bank holding companies, investment banks and securities houses, saving banks, real

estate and mortgage banks, non-banking credit institutions, and other specialized

governmental credit institutions are excluded from the initial sample to make the data more

comparable across countries. 96 % of the firms in the sample are consisted of commercial

banks, and the remaining 4 % were cooperative banks. Due to the log linear specification in

the estimated model, observations that have negative value on any of the explanatory

variables are also dropped from the sample. The selection process yielded a unbalanced panel

with 2042 observations belonging to 325 banks over the sample period. Not all the banks were

in continuous operation over the entire period due to failures, mergers, and de novo entry.

Table 2 presents the descriptive statistics of bank characteristics under investigation for each

country for 1999. All data are reported in US$ as the reference currency and corrected for

inflation. Differences in the average banks size are substantial. The average Czech bank has

more assets ($2.08 billion in 1999) than does the average bank in other CEE countries,

followed by Polish and Hungarian banks. However, the average Polish bank has generated

more loans than the average bank in other countries, followed by Czech and Estonian banks.

On average, banks in the Czech Republic and Poland have the highest equity capital on their

balance sheets, followed by Hungarian banks.

20
4. Empirical results

The mean cost and profit efficiencies obtained from the frontiers that were estimated

by two different methods, i.e. SFA and DFA, are presented in Tables 3 and 4.14 Furthermore,

we reported the results in four different truncation points (0, 1, 5, and 10%). The frontier

models we estimated all had R-squared values of 0.95 or higher and plausible parameter

estimates. The simple correlation between total assets and the absolute cost residual for each

bank indicated no monotonic patterns of heteroscedasticity.

As expected, the level of efficiency changes significantly with the level of truncation

applied. For example, the average cost efficiency measured by the DFA for the overall sample

changes from 45% to 66% when only 1% of extreme values are replaced by the value of

truncation point. The inefficiency level increases to 72% with 5% truncation. This might be

due to persistent random factors that do not factor out completely over the sample period as

assumed by the DFA approach. Following the prior literature, we will focus our analysis on

the efficiency measures obtained by 5% truncation level.

According to the SFA results at 5% truncation point, the average cost efficiency level

for 12 CEE countries under examination is 77%. This result suggests that, on average, about

one-fourth of the bank resources are wasted during the provision of banking services in

transition economies. The results of the analysis are comparable to that of previous studies of

developed banking markets that are typically in the vicinity of 20% to 30%. According to

these results Poland and Slovenia appears to be the most efficient countries while the Russian

Federation and the three Baltic States (Lithuania, Latvia, and Estonia) are the least efficient.

14
For the SFA, we estimated the frontier cost function and the inefficiency measures using the algorithm, based
on the model developed by Aigner et al. (1977). The residuals are computed following the approach by Jondrow
et al. (1982).

21
 The cost efficiency levels estimated by the DFA at 5 % truncation are slightly lower

than those estimated by the SFA. The overall efficiency measure is 0.72 for the entire sample.

The 72% efficiency measure means that the average bank needs 28% more resources to

produce the same output as the average efficient bank. Based on the DFA results, Poland has

the highest average efficiency level (82%) and Lithuania has the lowest (63%). Given the

relatively well-developed nature of the Polish banking industry this result does not come as a

surprise. This result might partly be attributed to the increased foreign participation with more

efficient operating techniques in Poland. The highly concentrated structure of the banking

markets and the lack of competition might be the reasons for such low scores of efficiency in

the Baltic States. For example, the Estonian banking market is highly concentrated and the

three largest domestic banks controlled more than 95% of total assets at the end of June 2000.

The efficiency scores of other countries range between 65% and 81%, implying that an

average bank in these countries can reduce its operating costs by 19% to 35% if it can adjust

its operations according to the bank at the frontier. Overall, the results imply that, banks in

transition economies can significantly reduce their production costs if they can utilize their

productive inputs more efficiently.

The results of the alternative profit efficiency estimation are presented in Table 4. Here

too, the levels of efficiency measures significantly vary with the level of truncation chosen.

As in many previous researches, the alternative profit estimates are lower than those of cost

efficiency (Berger and Mester (1997) for US banks; Maudos et al. (2002) for 10 EU member

countries). According to the SFA, approximately one-third of banks’ profits are lost to

inefficiency, and almost one-half according to the DFA.

22
4.1 Correlates of cost and profit efficiency

Having documented the efficiency scores of each national banking industry, the next

step is to determine whether the efficiency levels can be explained by bank-specific or country

specific factors. For this purpose, we provide an explanatory analysis through examining the

cross-sectional determinants of bank-specific efficiency scores from the SFA, by regressing

these measures against a number of financial and structural variables.

Several bank- and industry-specific factors may influence the efficiency of a particular

bank. Some of these factors may be neither inputs nor outputs in the production process, but

rather circumstances faced by a particular bank. The variables consist of two groups-the first

representing firm-specific attributes, and the second encompassing the nature of the market

and regulatory structure in effect over the period examined. The bank-specific variables

included in the regressions are: size (LNTA=log of total assets measured in thousands of US

dollar); capitalization (EQTY= book value of stockholders’ equity as a fraction of total

assets); risk (LOANS/TA= total loans over total assets, LLR/TL= Loan loss reserves as a

fraction of gross loans); funding (CSTF=customer and short term funding over total funds;

IBDP/TOTDEP= interbank deposits over total deposits) and off-balance sheet activity

(OBSI/TA=off-balance sheet items over total assets) variables.

The LNTA and EQTY variables are included in the model to examine the effect of

bank size and capitalization on efficiency. Strong capital structure is essential for the banks in

transition economies since it provides additional strength to withstand financial crises and

increased safety for depositors during unstable macroeconomic conditions. Furthermore,

lower capital ratios in banking imply higher leverage and risk, and therefore greater borrowing

costs. Thus, the efficiency level should be higher in better-capitalized banks. Since efficient

23
banks are likely to be better in credit evaluation we expect to get a negative coefficient on the

LLR/TL variable. We do not have any a priori expectation on the signs of the coefficients of

other bank-specific variables.

There are several ways by which market structure can affect the performance of

financial intermediaries. Recent theoretic work in the economics of information postulates

that under the setting of asymmetric information and uncertainty, competitive pressures serve

as the most effective instrument in fostering productive efficiency (Hart, 1983). Competition

motivates management to operate closely to their production frontier and also, under the

agency framework, provides the principals with relevant information for monitoring

effectively the agent’s activities. Incumbent banks in an industry characterized by entry

barriers are subject to a lesser degree of market discipline since the degree of competition they

face will be relatively low. Therefore, we hypothesize that increased competition will improve

bank efficiency. In order to account for the effect of the degree of competition (COMP), we

use the Panzar and Rosse (1987) H-statistic as computed previously in Yildirim and

Philippatos (2003a).15

It is widely acknowledged that due to the lack of potential competition, banks in

concentrated markets can exercise market power to earn monopoly profits and enjoy the

luxury of operating in inefficient levels. The ICR3 variable, which represents the market share

of the largest three banks in the industry, is used to capture the effect of market concentration

on bank efficiency. Two competing theories that explain the linkage between market

concentration and firm performance are the “structure-performance paradigm” (Mason, 1939;

15
H statistic is the sum of the elasticities of the reduced-form revenues with respect to input prices. Panzar and
Rosse (1987) show that H statistic is equal to 0 in monopoly, 1 in perfect competition and between 0 and 1 under
monopolistic competition.

24
Bain, 1951) and the “efficient structure hypothesis” (Demsetz, 1973). Structure-performance

paradigm asserts that concentration is the result of market power and conjectures a negative

relation between concentration and efficiency. The efficient structure hypothesis, in contrast,

argues that efficient firms grow large at the expense of inefficient firms, and therefore

anticipates a higher efficiency in concentrated markets. Hence, we do not have a strong ex-

ante expectation on the sign of the concentration coefficient.

The GDP variable represents the growth rate in country domestic product and is used

as a proxy for local economic conditions. Favorable economic conditions will affect positively

the demand and supply of banking services, and will possibly improve bank efficiency. To

distinguish between foreign and domestic banks we included the FOREIGN dummy variable.

We expect to find higher efficiency measures for foreign-owned banks compared to

domestically owned private or state banks. Similarly SPEC variable is used to distinguish

between commercial and cooperative banks and accounts for the effect of bank specialization.

We also specified the LISTED dummy variable to account for any systematic differences in

efficiency levels of the publicly traded banks and private banks.

4. 2 Results of the regression Analysis

The second stage regressions were estimated using GLS fixed-effects estimators,

where the standard errors were calculated using White’s (1980) correction for

heteroscedasticity. Table 5 reports the results of the estimation. Both regressions include

country and time-specific fixed effects which are not reported. Overall, most of the

coefficients are significant and are in line with our expectations.

The coefficient on the size variable is positive and statistically significant at the 1 %

level, indicating that, on average, larger banks attain a higher level of cost efficiency in their

25
operations. This might be the result of the relaxation of asset restrictions in the banking

system that allowed the banks to grow and venture into different banking business practices,

and to accrue some economies of scale and scope. Profit efficiency on the other hand does not

seem to be linked to asset size at any conventional significance levels. The level of equity

capital is positively related with efficiency in both models. This finding is consistent with the

results of the previous research that usually report higher cost and profit efficiency levels for

well-capitalized banks. Banks with higher ratio of loans to assets are found to be more cost

efficient. A negative and statistically significant LLR/TL coefficient, as expected, indicates

that a higher level of problem loans is associated with lower cost and profit efficiency levels.

This result might suggest that efficient banks are very effective in evaluating credit risk

(Berger and DeYoung, 1997). Cost-efficient banks also appear to have higher customer and

short tem funds in total funding and low interbank deposit ratios. These two measures,

however, have no significant correlation with profit efficiency. Finally, banks with higher

level of off-balance sheet activities are found to be significantly more cost and profit efficient.

Among the market structure variables, the degree of competition has a positive

influence on cost efficiency and a negative one on profit efficiency. These results suggest that

the banks operating in more competitive and contestable markets are under more pressure to

control their costs and cannot earn higher profits by exercising their potential market power.

Consistent with this result, the negative and highly statistically significant relationship

between industry concentration ratio and profit efficiency implies that profitability is not the

result of concentration or market power. The GDP variable is positively linked to cost

efficiency but negatively linked to profit efficiency. Regarding the impact of bank ownership,

results suggest that foreign banks operating in transition countries appear to be more cost

26
efficient but less profit efficient relative to domestically owned private banks and state-owned

banks. The dummy variable representing bank specialization is significant only in the cost

efficiency case indicating that commercial banks are less cost efficient relative to cooperative

banks. Finally, public trading dummy (LISTED) did not yield any significant relationship in

both cost and profit efficiency models.

5. Conclusion

The prospects for the financial services world have substantially changed over the last

decade. This is particularly true for the transition economies of Eastern Europe that have

gone through a historic transformation from central planning to a market-based system. This

study has examined the cost and profit efficiency of banking sectors in twelve transition

economies of Europe, by employing the stochastic frontier approach (SFA) and the

distribution-free approach (DFA). An unbalanced panel of CEE banks is studied over the

period 1993-2000. We first estimated the average cost and profit efficiency levels for a panel

of 325 banks over an 8-year period for twelve countries. We then regressed these efficiency

scores to see whether they depend on any particular bank-specific or industry-specific factors.

As in most previous studies of bank efficiency, we find that the average bank deviates

substantially from the best-practice frontier. The managerial inefficiencies in CEE banking

markets were found to be significant, with average cost efficiency level for 12 countries 72

and 77 percent by DFA and SFA. Overall, these average estimates suggest that an average

bank would have incurred 23 to 28 percent less of its actual costs had it matched its

performance with the best-practiced bank. According to our results Poland and Slovenia

appears to be the most cost efficient countries while the Russian Federation and the three

27
Baltic States (Lithuania, Latvia, and Estonia) are the least efficient. The alternative profit

efficiency levels are found to be significantly lower relative to cost efficiency. According to

SFA, approximately one-third of banks’ profits are lost to inefficiency, and almost one-half

according to the DFA.

The results of the second-stage regression analyses suggest that higher efficiency

levels are associated with large and well-capitalized banks. Further, banks that heavily rely on

core deposits in funding their assets are found to be more efficient. Consistent with most prior

research, higher level of problem loans is associated with lower efficiency levels. Regarding

the effect of market structure on bank performance, the higher level of competition in banking

markets is associated with lower cost and higher profit efficiency, while market concentration

is negatively linked to efficiency. Favorable economic conditions seem to improve only cost

efficiency. Finally, foreign banks operating in transition countries are found to be more cost

efficient but less profit efficient relative to domestically owned private banks and state-owned

banks.

As a caveat, we note that these results should be interpreted with great care since

previous research found that average efficiency estimates differ substantially across different

estimation approaches and efficiency concepts even for the same sample. Like other

stochastic frontier approaches the estimation results from SFA and DFA may be affected by

the distributional assumptions of the annual residuals of each bank. (Bauer et al. 1998).

We believe that this research contributes to the body of knowledge concerning the

process of transition from central planning to a market-based system, and it helps

policymakers to review the results of their efforts in promoting financial stability and creating

an efficient, properly supervised financial system for both the transitional economic reforms

28
and the preparations for accession to the European Union. Regarding the future path of the

CEE banking markets we would like to make the following remarks. It is clear that, the

integration of the European markets as well as the ongoing global financial liberalization will

force the CEE banking firms to improve their efficiency. As a result of the elimination of

regulatory barriers to foreign entry, the banking firms will feel heightened competitive

pressures from large European financial institutions that are currently operating at relatively

low margins. In response to growing domestic and international competition, banks will be

forced to enhance their performance by minimizing the costs of operations, employing

optimal production plans, upgrading their operations through new technology, and reducing

excess capacity through merging with more efficient banks. Banks will also have to search for

alternative ways to increase production capacity such as offering new services and products

and focusing on non-interest income-generating activities. Since only the most efficient

institutions will survive these challenges, the inefficient banks will either be acquired or

eventually be driven out of market by the competition. Increased efficiency in the banking

industry will affect positively the customers of those banks since reduction in costs will be

reflected as lowered prices and improved service quality. Increased efficiency will also

improve the capacity of the banking industry to generate financial innovations and meet better

the changing needs of the consumers of financial services.

Estimates of efficiency provide valuable information on the performance of banks and

the reasons for underperformance in CEE banking industries. The next stage of research can

be to link measures of efficiency to other factors such as corporate governance, product

quality and regulatory issues. It will also be of interest to extend the analysis to other sectors

such as insurance and other financial institutions.

29
 References

Aigner, D., K. Lovell and P. Schmidt (1997), “Formulation and Estimation of Stochastic Frontier
Production Function Models", Journal of Econometrics, 1977, 6, pp. 21-37.

Allen, Linda and Anoop Rai (1996), "Operational Efficiency In Banking: An International
Comparison," Journal of Banking and Finance, 1996, v20(4,May), 655-672.

Bain, Joe, S., (1951), “Relation of profit rate to industry concentration”, Quarterly journal of
economics, 65 pp. 293-324

Bauer, P.W., A.N. Berger, G.D. Ferrier and D.B. Humphrey (1998), ‘Consistency conditions for
regulatory analysis of financial institutions: a comparison of frontier efficiency methods’, Journal
of Economics and Business, 50, pp. 85-114.

Bell, F.W. and N.B. Murphy, 1968, “Costs in commercial banking: a quantitative analysis of bank
behavior and its relation to bank regulation”, Research Report No.41, Federal Reserve Bank of
Boston.

Benston, G.J., 1965, “Branch banking and economies of scale”, The Journal of Finance 20, no.2, 312-
332

Berg, S. A., F. R. Frrsund, L. Hjalmarsson and M. Suominen, (1993), ‘Banking efficiency in the Nordic
countries’, Journal of Banking and Finance, Vol. 17, Nos. 2-3, April, 371-388.

Berger, A.N. and D.B. Humphrey, 1991. “The Dominance of Inefficiencies over Scale and Product
Mix Economies in Banking”, Journal of Monetary Economics 28(1): 117-148.

Berger, A.N., Humphrey, D.B., 1992. “Measurement and efficiency issues in commercial banking”, In:
Griliches, Z. (Ed.), Output Measurement in the Service Sectors. The University of Chicago Press,
Chicago, pp. 245-479 (Chapter 7).

Berger, A.N., 1993. “Distribution-Free Estimates of Efficiency in the U.S. Banking Industry and Tests
of the Standard Distributional Assumptions”. Journal of Productivity Analysis 4(3): 261-292.

Berger, A. N., W. C. Hunter, and S. G. Timme.,1993, “The Efficiency of Financial Institutions: A

Review and Preview of Research Past, Present, and Future,” Journal of Banking and Finance,
1993, v17, 221-249.
Berger, A N., Leusner, J H., Mingo, J J.,1997, “The efficiency of bank branches”. Journal of Monetary
Economics, Vol. 40 (1), pp141-162

Berger, A.N. and D.B. Humphrey. 1997. “Efficiency of Financial Institutions: International Survey and
Directions for Future Research.” European Journal of Operational Research 98: 175-212.

Berger, A.N. and L.J. Mester. 1997. “Inside the Black Box: What Explains Differences in the
Efficiencies of Financial Institutions?” Journal of Banking and Finance 21: 895-947.

30
Berger, A.N. and L.J. Mester. 1999. “What Explains the Dramatic Changes in Cost and Profit
Performance of the U.S. Banking Industry?” Wharton school Center for Financial Institutions,
Working Paper No. 99-1

Berger, A. N., Rebecca S. Demsetz, Philip E. Strahan, 1999, “The consolidation of the financial
services industry: Causes, consequences, and implications for the future”, Journal Of Banking And
Finance (23)2-4 (1999) pp. 135-194

Berger A. N., R. Deyoung, H. Genay, and G. F. Udell, 2000, “Globalization of Financial Institutions:
Evidence from Cross-Border Banking Performance”, Brookings-Wharton Papers on Financial
Services 2000 (2000) 23-120

Berger A. N., and R. Deyoung, 2001, “The Effects of Geographic Expansion on Bank Efficiency”,
Journal of Financial Services Research, 19 (2-3): 163-184

Bhattacharya, A., Lovell, C.A.K., and Sahay, P., 1997. “The impact of liberalization on the productive
efficiency of Indian commercial banks”. European Journal of Operational Research 98, 332-345.

Bikker, J.A., 1999, “Efficiency In The European Banking Industry: An Exploratory Analysis To Rank
Countries”. Working Paper, De Nederlandsche Bank

Chang, C.E., Hasan, I. and Hunter, W.C., 1998. “Efficiency of multinational banks: An empirical
investigation”. Applied Financial Economics 8, pp. 1¯8.

Claessens, S., A. Demirguc-kunt and H. Huizinga, 2001, “How Does Foreign entry Affect Domestic
Banking Markets?”, Journal of Banking &Finance, vol.25, 2002, pp. 891-911.

Demsetz, H.: “Industry structure, market rivalry and public policy.” Journal of Law and Economics, 16
(1973), 1-9.

DeYoung, R. and Nolle, D., 1996. “Foreign-owned banks in the US: Earning market share or buying
it?”. Journal of Money, Credit, and Banking 28, pp. 622¯636.

Dowling W.A. and George C. Philippatos, 1990. “Economies of scale in the US savings and loan
industry:1973-83”, Applied Economics, 22, 237-248

Hart, O. S., “The Market Mechanism As an Incentive Scheme," Bell Journal of Economics l4 (Autumn
l983) 366-382

Hasan, I. and Hunter, W.C., 1996. “Efficiency of Japanese multinational banks in the United States.”
Research in Finance 14, pp. 157¯173.

Hasan I. and K. Marton, “Development and Efficiency of the Banking Sector in a transitional
Economy: Hungarian Experience”, BOFIT Discussion Papers No 7 / 2000

Humphrey, D.B. and Pulley, L.B., 1997. , “Banks' responses to deregulation: Profits, technology, and
efficiency”. Journal of Money, Credit, and Banking 29, pp. 73¯93

Isik Ihsan, Hassan M. Kabir, “Technical, scale and allocative efficiencies of Turkish banking industry”
Journal Of Banking And Finance, (26)4 (2002) pp. 719-766

31
Laeven L., “The Efficiency of the Thai Banking System”, World Bank, July 1999.

Leightner, J.E., and C. A. Knox Lovell, “The Impact of Financial Liberalization on the Performance of
Thai Banks”, Journal of Economics and Business, No.50, 1998 pp. 115-131

Jondrow, J., I. Materov, K. Lovell and P. Schmidt, "On the Estimation of Technical Inefficiency in the
Stochastic Frontier Production Function Model," Journal of Econometrics, 1982, 19, pp. 233-238.

Mahajan, A., Rangan, N. and Zardkoohi, A., 1996. “Cost structures in multinational and domestic
banking.” Journal of Banking and Finance 20, pp. 283¯306.

Mason, Edwards S., 1939, “Price And Production Policies Of Large-Scale Enterprise”, American
Economic Review, 29, Pp.61-74

Maudos J., J. Pastor, F. Perez, and J. Quesada, 2002, “Cost and Profit Efficiency in European Banks”,
Journal of International Financial Markets, Institutions, and Money 12, (2002) 33-58

McAllister, P.H. and D.A. McManus. 1993. “Resolving the Scale Efficiency Puzzle in Banking”.
Journal of Banking and Finance 17(2-3): 389-405.

Mester L.J., 1996, “A Study of Bank Efficiency Taking into Account Risk-Preferences”, Journal of
Banking and Finance, 20, pp. 1025 - 1045.

Mitchell, K. and N. M. Onvural. 1996. “Economies of Scale and Scope at Large Commercial Banks:
Evidence from the Fourier Flexible Functional Form” Journal of Money,Credit,and Banking
28(2): 178-99.

Ozkan-Gunay, E.N., 1998, “Economies of Scale and Scope in the Turkish Banking Industry: The Effect
of the Financial Liberalization program” The Journal of Economics, Volume XXIV, No. 1, 1998

Panzar, J.C., Rosse, J.N., 1987. “Testing for Monopoly equilibrium”. Journal of Industrial Economics
35, 443-456.

Peek, J., Rosengren, E.S., Kasirye, F., 1999. “The poor performance of foreign bank subsidiaries: Were
the problems acquired or created?” Journal of Banking and Finance 23, 579¯604.

Sealey, C.W., Jr. and J.T. Lindley. 1977. “Inputs, Outputs, and a Theory of Production and Cost at
Depository Financial Institutions”, Journal of Finance 32(4): 1251-1266.

Srivastava, P., “Size, Efficiency and Financial Reforms in Indian Banking”, Indian Council for
Research on International Economic Relations, July 1999

Vander Vennet, R., 1996. “The effect of mergers and acquisitions on the efficiency and profitability of
EC credit institutions”. Journal of Banking and Finance 20, pp. 1531¯1558.

Yildirim, H.S., and George C. Philippatos, 2003a, “Competition and contestability in CEE banking
industry”, Working paper, University of Tennessee, April 2003.

32
Table 1 Summary Statistics for Model Parameters*
Variable Mean ($000) Std Dev Minimum Maximum
Y1 (Loans) 375745.14 887784.54 0 8631164.93
Y2 (Investments) 397894.68 977042.86 11.07 8968847.69
Y3 (Deposits) 550507.11 1530905.94 0 16408895.6
X1 (Borrowed Funds) 757116.5 1818418.05 0 17826757.81
X2 (Labor) 15185.77 44077.24 0.5 819737.22
X3 (Physical Capital) 16663.7 48526.43 4.36 649857.76
W1 (Price of Borrowed Funds) 31.75 1331.32 0.0003 56042.37
W2 (Price of labor) 2.20836 25.757 0.001 1038.18
W3 (Price of Physical capital) 0.0233 0.0330431 0.0003 1.112
C (Total cost) 107544.95 400582.02 13.08 13397311.31
P (Earnings before tax) 13499.63 60255.81 1 615735.09
Z (Equity Capital) 83677.68 161045.35 15.604 1599000
*
All measures are adjusted for inflation in 2000 US$, except for unit costs.

33
Table 2 Descriptive statistics of banks in sample by country average for 1999
Assets Deposits Other Equity Loans Investment
Country N funding securities
Czech Rep. 23 2080406.6 1624421.6 133993.74 181037.25 824995.62 1022069.07
Estonia 4 853797.76 596246.13 77872.34 117707.83 497291.66 247507.28
Croatia 35 371027.14 239724.81 47188.73 55213.97 197176.16 165395.22
Hungary 30 955514.74 747333.03 47471.45 82570.47 407301.88 458503.72
Lithuania 9 283078.5 233102.49 8186.64 29750.64 146475.77 71032.55
Latvia 20 153242.88 132194.78 1500.94 14950.45 69846.75 62257.78
FYR of Macedonia 10 97645.31 63878.15 496.53 16536.54 41953.39 47531.04
Poland 43 1912469.2 1630925.3 7222.85 172547.94 967080.4 779226.86
Romania 27 318874 252160.32 3128.1 48889.57 97394.2 170034.02
Russian Federation 80 535235.56 378220.24 14861.45 69694.85 228825.34 226358.92
Slovenia 20 687551.78 560575.39 24718.57 70852.66 377501.9 274026.27
Slovakia 18 846286.35 738380.42 11147.26 46441.51 484277.61 356381.25
All quantity variables are in thousands of US dollars and corrected for inflation.
Total number of banks N = 319 for 1999.
Total observations over 9 years (1993-2000) = 2042

34
Table 3 Cost Efficiency Measures
Stochastic Frontier Approach (SFA) Distribution Free Approach (DFA)

Country SFA(0) SFA(1) SFA(5) SFA(10) DFA(0) DFA(1) DFA(5) DFA(10)

Czech Republic 0.725 0.748 0.769 0.802 0.478 0.627 0.705 0.756
Estonia 0.698 0.715 0.742 0.764 0.403 0.666 0.685 0.724
Croatia 0.768 0.792 0.814 0.837 0.597 0.697 0.724 0.785
Hungary 0.730 0.754 0.788 0.803 0.457 0.587 0.712 0.748
Lithuania 0.602 0.654 0.692 0.735 0.415 0.541 0.628 0.674
Latvia 0.658 0.711 0.732 0.779 0.387 0.596 0.667 0.726
FYR Macedonia 0.728 0.752 0.768 0.794 0.432 0.712 0.745 0.774
Poland 0.745 0.801 0.854 0.835 0.444 0.738 0.824 0.841
Romania 0.714 0.735 0.748 0.792 0.450 0.700 0.733 0.785
Russian Fed. 0.629 0.702 0.723 0.750 0.410 0.624 0.650 0.714
Slovenia 0.738 0.792 0.827 0.844 0.472 0.766 0.814 0.854
Slovakia 0.656 0.755 0.775 0.800 0.409 0.642 0.687 0.755
Overall 0.699 0.743 0.769 0.795 0.446 0.658 0.715 0.761

35
Table 4 Profit Efficiency Measures
Stochastic Frontier Approach
(SFA) Distribution Free Approach (DFA)

Country SFA(0) SFA(1) SFA(5) SFA(10) DFA(0) DFA(1) DFA(5) DFA(10)

Czech Republic 0.642 0.671 0.695 0.712 0.331 0.416 0.429 0.521
Estonia 0.624 0.666 0.701 0.714 0.342 0.382 0.435 0.530
Croatia 0.542 0.592 0.648 0.660 0.300 0.444 0.560 0.684
Hungary 0.435 0.522 0.573 0.594 0.274 0.530 0.615 0.671
Lithuania 0.601 0.650 0.680 0.699 0.357 0.371 0.489 0.580
Latvia 0.635 0.674 0.704 0.716 0.401 0.420 0.456 0.555
FYR Macedonia 0.667 0.680 0.698 0.719 0.380 0.394 0.461 0.568
Poland 0.564 0.609 0.664 0.670 0.310 0.430 0.548 0.667
Romania 0.375 0.461 0.550 0.566 0.294 0.482 0.554 0.608
Russian Fed. 0.472 0.540 0.595 0.608 0.275 0.557 0.571 0.605
Slovenia 0.580 0.627 0.670 0.676 0.314 0.403 0.523 0.633
Slovakia 0.635 0.673 0.692 0.701 0.370 0.348 0.456 0.557
Overall 0.564 0.614 0.656 0.670 0.329 0.431 0.508 0.598

36
Table 5 Regression Analysis of potential correlates of efficiency
Variable Cost Efficiency Profit Efficiency
Coefficient t-value Coefficient t-value
Intercept 0.889 13.12*** 0.726 12.47***

LNTA Logarithm of Total Assets 0.012 4.74*** 0.003 0.81

EQTY Shareholder's Equity /Total Assets 0.126 5.03*** 0.058 1.34

LOANS Total Loans / total assets 0.057 3.14*** 0.027 1.23

LLR/TL Loan Loss Reserves/Gross Loans -0.002 -3.87*** -0.006 -6.77***

CSTF Customer and ST funding/total funds 0.051 1.81* -0.028 -0.58

IBDP Interbank Deposits/Total Deposits -0.027 -4.45*** -0.013 -1.76*

OBSI Off-balance sheet items/total assets 0.017 1.73* 0.045 2.57**

Degree of Competition by Panzar and Rosse
COMP 0.0426 2.24** -0.154 -4.55***
H-statistic
ICR3 Market share of largest 3 banks in the industry -0.000 -0.51 -0.003 -7.08***

GDPGROWTH Growth rate in state real domestic product 0.004 6.12*** -0.021 -10.17***
Dummy variable that equals 1 if bank is a
SPEC -0.065 -4.94*** 0.024 1.12
commercial bank, 0 otherwise
Dummy variable that equals 1 if more than
FOREIGN 50% of the bank assets are owned by foreign 0.033 3.98*** -0.079 -5.12***
banks; 0 otherwise
Dummy variable that equals 1 if bank is
LISTED -0.004 -0.46 0.008 0.75
publicly traded, 0 otherwise
Adjusted R2 0.34 0.26
*,**,*** represent significance at the 10%, 5%, and 1% levels respectively
All regressions also include country and time dummy variables which are not reported.

37