Application of Data Envelopment Analysis Models in Evaluating the Efficiency of Bank Branches

Although digitalization is a major topic in the banking industry today and clients are able to use banking applications to handle most service and transaction requests and to use a wide range of products, clients still want to consult their bankers when making decisions about long-term commitments and their financial future. This is why the role of bank branches is still indispensable today. However, given the significant costs of running a branch network, the efficiency and performance of individual branches is a concern for every bank. One possible approach to evaluate efficiency is data envelopment analysis models. However, basic DEA models may not be sufficient when modelling more complex production processes such as those in a bank branch network. For this reason, network data envelopment analysis models have been developed to capture the more complex structure of the production process. This paper focuses on their application to assess the efficiency of bank branches in 2019–2021, where branches are modeled as two-stage production processes. To this purpose, a novel approach is used to address the problem of the absence of efficient units in many practical applications. Furthermore, super-efficiency models are applied and modified for two-stage production processes. In the context of their application, the thesis addresses the problem of infeasibility of solution when considering variable returns to scale and using approaches that do not suffer from the infeasibility problem. Considering the use of different approaches to obtain measures of efficiencies and measures of super-efficiencies, and thus the resulting ranking of the branches, the differences in the results obtained are discussed, as well as the changes in the results over time.