Data envelopment analysis (DEA), occasionally called frontier analysis, was originated by
Charnes, Cooper and Rhodes in 1978. It is a performance measurement technique, can be used
for evaluating the relative efficiency of the decision-making units (DMU's) in the organizations.
It is a method for identifying efficient points in the mixed case. That is, when there are both “less
is better” and “more is better” measures. An attractive feature of DEA is it does produce an
efficiency score between 0 and 1. It does this by making slightly stronger assumptions about how
efficiency is measured. Specially, DEA assumes each performance measures can be classified as
either an inputs or an output. For outputs, more is better, whereas for inputs, less is better. The
score of a point or a decision-making unit is then the ratio of an output score divided by an input
score. DEA is concerned with measuring the relative efficiency of a sample of producers,
referred to as decision-making units (DMU). Another commonly use DEA is Banker, Charnes
and Cooper (BCC) model. BCC version is more flexible and allows variable return to scale. That means if an increase in a unit’s inputs does not produce a proportional change in its outputs then
the unit exhibits variable returns to scale. As the unit changes its scale of operations its efficiency
would either increase or decrease. The main advantage of the variable return to scale is that scale
inefficient companies are only compared to efficient companies of a similar size. However, when
imposing variable returns to scale the company may be technically efficient but not operating at
its optimal scale. Under the assumption of variable returns to scale a unit found to be inefficient
has its efficiency measured relative to other units in the data-set of a similar scale size only. As a
result no unit will obtain a lower efficiency score using variable returns to scale and some units
are likely to achieve higher efficiency results. The number of 100% efficient units is also likely
to be higher under the assumption of variable returns to scale as all units with the lowest value
for any of the inputs or highest value for any of the outputs are rated as efficient.