摘要
Sensitivity of the returns toscale (RTS) classifications in data envelopment analysis is studiedby means of linear programming problems. The stability regionfor an observation preserving its current RTS classification(constant, increasing or decreasing returns to scale) can beeasily investigated by the optimal values to a set of particularDEA-type formulations. Necessary and sufficient conditions aredetermined for preserving the RTS classifications when inputor output data perturbations are non-proportional. It is shownthat the sensitivity analysis method under proportional dataperturbations can also be used to estimate the RTS classificationsand discover the identical RTS regions yielded by the input-basedand the output-based DEA methods. Thus, our approach providesinformation on both the RTS classifications and the stabilityof the classifications. This sensitivity analysis method caneasily be applied via existing DEA codes.