chapter 2 - Bentham Science

144 Research Topics in Agricultural and Applied Economics, Vol. 2 Rezitis et al.
the procedure for changing the levels of inputs and outputs so that the branching efficiency can be increased and
performs a sensitivity analysis to determine the inputs and outputs to which the efficiency is more sensitive.
There is a small number of studies that have investigated the relative efficiency of Greek bank branches. The studies that
estimate the efficiency of different branches of the Commercial Bank of Greece are those by Vassiloglou and Giokas
(1990) based on 20 branches; Giokas (1991) on 17 branches; Donatos and Giokas (1995) on 187 branches; Giokas and
Athanassopoulos (2000) on 47 branches; and Donatos, Giokas and Athanassopoulos (2002) on 63 branches. However,
there is a growing international literature of DEA studies that investigates the efficiency of different bank branches. It is
considerable that most of them use a sample of fewer than 50 branches. In particular, the studies using the DEA
methodology to measure the efficiency of bank branches around the world are those by Sherman and Gold (1985) and
Golany and Storbeck (1999) based on bank branches in the US; Schaffnit, Rosen and Paradi (1997), Edelstein (2004)
and Paradi and Schaffnit (2004) in Canada; Lovell and Pastor (1997) in Spain; Rouatt (2002) and Yang (2002) in India;
and Jablonsky, Fiala, Smirlisand and Despotis (2004) in the Czech Republic. The remainder of this paper is organized as
follows. Section 2 outlines the methodological framework. The data sample is discussed in Section 3. Section 4 discusses
the empirical results obtained, while Section 5 offers a conclusion.
METHODOLOGY
The present study applies data envelopment analysis (DEA), which was initially developed by Charnes, Cooper and
Rhodes (1978), known as CCR, and was introduced to the banking sector by Sherman and Gold (1985). The detailed
formulation of the CCR model is given as:
Maxh
subject to
s
r 1
o m
uy
r rjo
vx
i ijo
i1
s
r 1
m
i1
r rj
i ij
1
u , v , r 1,..., s, i 1,..., m, j 1,...,
n
r i
uy
vx
where ho is the relative efficiency of branch o; o is the branch being assessed from the set of j=1,…,n bank branches; j is
the number of branches (j=1,…,n); r is the number of outputs (r = 1,….,s); i is the number of inputs (i = 1, ….,m); yrj is
the observed output r at branch j ( r=1,…,s); xij is the observed input i at branch j (i=1,…,m); ε is a small positive number
whose value in the present study is ε=10 -6 ; and ur, vi are virtual multipliers for input i and output r, respectively.
The above functional model is replaced with a linear programming equivalent through a series of transformations. In
the case where output enhancement is emphasized, the formulation is written as:
s
Maxh u y
(2)
subject to
o r rjo
r 1
m
vx
1
i ijo
i1
s m
uy vx 0
r rj i ij
r1 i1
u , v , j 1,...,
n
r i
(1)