Performance evaluation of Chinese commercial banks by an improved slacks-based DEA model

The national economy heavily relies on the banking sector, and researchers have been focused on discovering the most effective method of enhancing bank's overall efficiency. Data Envelopment Analysis (DEA) is a commonly utilized tool for measuring banking efficiency. We proposed an improved slacks-based measure model (SBM) with undesirable outputs using a by-production framework. In such case, the bank's operating process is considered as a parallel system. Inputs are categorized into interest costs and operating costs, which are then evaluated in two separate parallel stages. In the first stage, outputs are categorized into interest income and non-interest income, which are shared and considered as final outputs. In the second stage, non-performing loans are deemed as undesirable outputs. This proposed approach splits bank efficiency into two parallel stages, enabling us to identify the source of overall inefficiency, and effectively capture the efficiency differences between the two types of costs. To demonstrate the feasibility of the proposed approach, we have successfully applied it to 36 commercial banks in China from 2016 to 2021. The empirical results of the research prove that the changes of the banks' overall efficiencies were mainly due to the changes of stage 2. And this approach provides a rich source of information to assist in decision making.