Page 319 - Contributed Paper Session (CPS) - Volume 4
P. 319
CPS2245 Azrie Tamjis
The aims of this study are: first, to carry out a cost-efficiency analysis of
Malaysian banks for the years 2000– 2011 using stochastic frontier approach
and examining how changes in the financial services affected efficiency,
productivity and the market structure of the banking industry in Malaysia.
Second, to examine the impact of market liberalisation initiatives, via the
FSMP, on efficiency and productivity in Malaysian banks.
2. Methodology
In this study, frontier measurement is employed to measure the efficiency
of Malaysian banks for the years 2000– 2011. For better estimation of cost-
efficiency, and taking into account the effect of heterogeneity (e.g. ownership
structure, banks specialisation, inherent risks, and size), this study uses Battese
and Coelli’s (1995) one-stage approach, which may have an impact on the
efficiencies. In this one-stage approach, a set of control variables (e.g. capital
adequacy, asset quality and liquidity) and environmental variables (e.g.
ownership, specialisation, financial liberalisation and size) are included into the
specification of cost- and profit-efficiency functions. These different sets of
control and environmental variables are tested in several stages using
statistical testing (i.e. the log-likelihood ratio test), searching for the best fitting
model that is later utilised for the estimation of efficiency scores in Malaysian
banks.
The SFA model assumes that in producing a certain level of output, firms
face various technical inefficiencies and a given combination of input levels.
The firm’s production is influenced by the sum of a parametric function of
known inputs, with unknown parameters, and a random error (associated with
the measurement error of the level of production and inefficiency). SFA
requires a functional form, such as cost or profit, with a two-component error
terms: random error and inefficiency. By way of illustration, the single-
equation stochastic cost function model is shown below:
= + (1)
where is the natural logarithm of output for the i-th bank at time t,
is a vector of inputs of i-th bank at time t, is a vector of unknown
parameters to be estimated and is the error term. Following Aigner et al.
(1977), the assumption of the composed error term is:
= + (2)
where and are independently distributed; represents random
uncontrollable error and is assumed to be normally distributed with zero mean
and variance is drawn from a one-sided distribution that is assumed to
308 | I S I W S C 2 0 1 9