A Novel Model for Inverse Data Envelopment Analysis in Multi-Stage Network

Sensitivity analysis is one of the purposes of data envelopment analysis (DEA). In inverse DEA, this question is always answered: If outputs of units under assessment are increased to a specified amount, by how much should the inputs of the unit be increased in order to maintain the efficiency of the unit? However, in two stage network DEA despite the intermediate indexes, the aforementioned question can be stated as follows: If outputs of unit under assessment are increased to a specified amount, by how much should the inputs and intermediate indexes of the unit be increased in order to maintain the efficiency of the unit, whether it is input oriented or output oriented? In most units with two stage structure, intermediate indexes are neglected and the system is considered as a single-stage network system with inputs and outputs. Though in these networks it is assumed that output value of stage 1 is considered the same as inputs of stage 2; however in the current sensitivity analysis, intermediate index changes were very significant. Eventually in this study, input and intermediate indexes changes were calculated by presenting the proposed model which is considered as novelty of this research. The arising question was answered using a numerical example. Afterwards, a comparison between the two models was investigated: a) considering intermediate indexes changes (two stage DEA model); b) neglecting intermediate indexes (single stage DEA model)

کلیدواژه ها:

Inverse data envelopment analysis ، Two-stage data envelopment analysis ، intermediate index ، minimum increase of input ، efficiency