classifying flexible measures in two-stage network dea

Standard data envelopment analysis (dea) supposes that the performance measure status from the point of view of input or output is known. nevertheless, in some situations, determining the status of a performance measure is not easy. measures with unknown status of input /output are called flexible measures. in this paper we propose new models based on the multiplier model under constant returns to scale (crs) and variable returns to scale (vrs) in basic two-stage network structures to classify flexible measures, in which each one of the flexible measures is treated as either input or output to maximize the overall network efficiency of the dmu under evaluation. the current paper develops an additive efficiency decomposition approach wherein the overall efficiency is expressed as a (weighted) sum of the efficiencies of the individual stages. this approach can be applied under both crs and (vrs) assumptions. the status of one flexible measure in two crs and vrs environments in basic two-stage network structures may have different results. in other words, a flexible measure is considered as input in crs environment, whereas it is selected as output in vrs environment. numerical examples are used to illustrate the procedures.