Abstract
Uncertainty is an important issue to consider when evaluating entities in both public and private sectors. On the other hand, many operations have more than one stage process when some inputs are fed to the system to produce a number of intermediate measures. The intermediate measures are then transformed into final products in the subsequent stages. The composition method in network data envelopment analysis (NDEA) is a popular method for measuring the efficiency of a two-stage process. The composition method is fractional bi-objective programming that is solved by non-linear programming techniques such as bisection search. In this paper, the two-stage NDEA is extended with negative data and undesirable outputs. First, we propose an alternative linear model based on the goal programming technique to avoid complex non-linear calculations. Then, we use a method to transform negative data into positive and undesirable outputs into desirable ones. Finally, we develop the proposed model using the fuzzy α-cut approach in order to incorporate data uncertainty in the linear goal programming (GP) model. To validate the accuracy of the proposed model, a numerical example is solved. To show the applicability of the proposed model, a real case of 22 insurance companies is examined. We also perform a comparative analysis to specify the benchmark and inefficient companies. Comparative analysis can help managers to recognize where improvement should be investigated with priority.