Reiriew of Islamic Economics, Vol. 8, No. 2, 2004
ɪʒ
The relative merits and limitations of these approaches have
often been discussed in the literature. Put briefly, one often-mentioned
disadvantage of SFA models is that they impose a function form and
distribution assumptions on the data prior to estimation.10 In
contrast, DEA does not require any assumptions about the functional
form, and is, therefore, less open to misspecification. Also, DEA is
non-parametric; it does not take into account random errors. As such,
there is no problem in subsequently assuming an underlying
distribution form for the error term. However, for this very ability to
avoid statistical noise, the efficiency estimates the approach provides
could well be biased, if the production process is largely characterized
by stochastic elements.
The SFA and DEA techniques are mostly used in two stages.
First, each seeks to estimate efficiency or inefficiency scores relevant
to the objectives of the study. These scores can be used to place the
PUs on the efficiency ladder to indicate their relative performance."
In the second stage, an attempt is made to search for the nature and
extent of causal relationship between inefficiency estimates and other
relevant variables such as size of the firms, their location, taxation,
public expenditure, stability, investment, profits, stock prices, and so
on. Use is made of logarithmic conversion of data that straightens the
selected production function. The application can focus on individual
entities e.g. firms or farms or their aggregation. The frontier approach
is a tool that can be used in any field of inquiry where variables can
be subjected to measurement.
Using the SFA for estimating cost efficiency requires the
specification of a function form. The translog cost frontier is often
considered appropriate for studying efficiency issues in the banking
sector. It is a very general and flexible function form and encompasses
some other approaches also like the Cobb-Douglas. SFA looks at the
question of efficiency in terms of cost minimization for given outputs,
rather than maximization of output from given inputs.
SFA models vary according to the type of data used for the study
i.e. cross-sectional or panel, and are oriented to inputs or to output.
Studies on efficiency of Islamic banks we shall be evaluating use cross-
sectional input oriented data i.e. they seek resource conserving cost
minimization, given the output to be produced. Early cost oriented
efficiency studies typically included more than one output and jointly