DEA with interval data: an illustration using the evaluation of branches of a Czech bank

In this paper we present an approach for dealing with interval data in DEA, which then we apply in assessing the efficiency of branches of a Czech bank. Unlike the original DEA models which evaluate efficiency in terms of past exact, we attempt to evaluate efficiency for a future period by assuming that the levels of inputs and outputs are not known exactly; the true input–output data are known only to lie within bounded intervals which are formed by pessimistic and optimistic estimates for the inputs and outputs. In such an interval data setting, we estimate upper and lower bounds for the efficiency scores of the units (branches). Moreover, we suggest a post DEA-model for discriminating, in terms of endurance, among the efficient units.