Supplier selection: comparison of DEA models with additive and reciprocal data

Supplier evaluation is one of the most important fields of application for data envelopment analysis (DEA). Criteria may include negative data in both input and output factors. Data translation can solve this problem, but interpretation is not evident from the literature. Use of an additive model is one method of managing the problem of negative data. This paper addresses this issue in relation to the supplier ranking problem. It describes the development of a ranking with cross-efficiency that incorporates negative data into the additive model. The additive model we describe is compared with previously used DEA models in which data is replaced with reciprocal values when necessary. We present a supplier-evaluation-related example of this case. After the efficiency evaluation, a supplier ranking system is constructed. To do this, we use the cross-efficiencies obtained from the additive model. Aggregate efficiencies help display the suppliers in descending order of efficiency. Finally, the paper compares self- and peer-appraisal indicators for reciprocal and additive DEA models.

Automated summary

Supplier evaluation is crucial in DEA, where negative data can be handled using additive models. This paper focuses on supplier ranking with cross-efficiency, incorporating negative data into the model and comparing it with traditional DEA models. A supplier-evaluation example is provided to demonstrate the methodology.