A mixed-integer network DEA with shared inputs and undesirable outputs for performance evaluation: Efficiency measurement of bank branches

Abstract

Conventional DEA performs like a “black box” and provides no information about sub-processes. In some cases, such as banks, providing services made up of interactive and interdependent processes. Also, in real-world applications, inputs could be shared among these sub-processes. Moreover, due to the characteristics of some variables, such as number of employees, only integer values could be assigned to them. Hence, to address these shortcomings, in this study, a mixed-integer network DEA (MI-NDEA) with shared inputs and undesirable outputs has been proposed to evaluate the efficiency of decision-making units. The proposed model considers integer values for some of the input variables. Also, it assumes that some inputs are shared among different stages of the production process. To illustrate the capability of the model, the efficiency of “Internet banking”, “profitability”, “production”, and “overall” performance of a set of bank branches have been evaluated and results are discussed. The results indicate that the mean of overall efficiency for all branches is high. However, some branches are not efficient enough in the “Production” stage or “Profitability” stage. To identify the source of inefficiency in such branches, projection values have been calculated and recommendations have been made for policy makers.

Keywords:

• Data envelopment analysis
• banking
• efficiency
• integer programming

Acknowledgements

The article was prepared within the framework of the Basic Research Program at HSE University. The authors would like to thank Professor Said Salhi, The Editor-in-Chief of Journal of the Operational Research Society, associate editor, and two anonymous reviewers for their insightful comments and suggestions. As result, this article has been improved substantially.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.