DEA is used widely nowadays and has become an important tool in
evaluating the performance of manufacturing firms as well as service
organizations. It is a non-parametric methodology which uses linear programming
methods to optimize the use of inputs or outputs of every bank in the data set. If
the aim is to minimize inputs while outputs are constrained, we have an outputoriented
DEA model; else if it is to maximize outputs while inputs are constrained,
we have an input-oriented DEA model. In this case, banks that are already at their
optimization levels will form an (optimal) piece-wise surface (or frontier) for the
whole set, and then efficiency of a certain bank now can be measured by comparing
its current level to the frontier. Practically, we can measure the efficiency of a certain j0-th bank using
the equation proposed by Charnes, Cooper and Rhodes (1978) under the assumption that there is no difference in scale among Thai banks4.