DEA problems and limitations

Question 1

The efficiency of a given DMU depends, therefore, on:

Answer 1

the inputs and outputs that have been included in the model.

Question 2

the more inputs and outputs are included in the model, the more:

Answer 2

data is needed to obtain reliable results; see Pedraja et al (1999).

Question 3

DEA problems:

Answer 3

extreme values of inputs or outputs and the choice of variables for the model.

Question 4

The DEA problem of extreme values of inputs or outputs ( the lowest value of an input and the highest value of an output) could result in:

Answer 4

DMUs that become 100% efficient.

Question 5

The DEA problem of “the more inputs and outputs are included in the model”, could result in:

Answer 5

the more units will be efficient. Taking a metaphor from a different context, we would find that the “naughty boy” who puts the least effort in the class and gains low marks would be efficient under a DEA model that includes amount of effort as an input.

Question 6

If too many inputs and outputs are included:

Answer 6

some of them may be highly correlated and, therefore, redundant.

Question 7

The DEA problem of “removing inputs or outputs from a model”:

Answer 7

will decrease efficiency estimates, which will, at best remain constant. This decrease would affect some DMUs more than others.

Question 8

The same characteristics that make DEA a powerful tool can also create problems. An analyst
should keep these limitations in mind when choosing whether or not to use DEA.

Answer 8

1. Since DEA is an extreme point technique, noise (even symmetrical noise with zero mean) such as measurement error can cause signicant problems.
2. DEA is good at estimating “relative” efficiency of a DMU but it converges very slowly to
   “absolute” efficiency. In other words, it can tell you how well you are doing compared to your
   peers but not compared to a “theoretical maximum.”
3. Since DEA is a nonparametric technique, statistical hypothesis tests are dicult and are the
   focus of ongoing research.
4. Since a standard formulation of DEA creates a separate linear program for each DMU, large
   problems can be computationally intensive.