Bayesian DEA framework for market power and efficiency analysis in banking operations

Abstract The accurate measurement of marginal costs and market power is a persistent challenge in production economics and operations research, particularly within the banking sector. This study addresses this problem by proposing a Bayesian likelihood-based approach to estimate the Lerner index under multiple inputs and outputs, extending the nonparametric Data Envelopment Analysis methods of Fukuyama and Tan (J Oper Res Soc 73:445–453, 2022a) with statistical inference. Using data from 36 Chinese banks (2011–2018), the results reveal that market power in loans is generally higher and more stable compared to securities, with significant variations across different bank ownership types. For instance, city banks demonstrate the highest market power in loans, while state-owned banks exhibit the lowest. Efficiency analysis indicates volatility across all bank types, with no clear efficiency patterns. These findings have critical policy implications, emphasizing the need for targeted strategies to enhance market power and efficiency sustainably. For example, rural, city, and joint-stock banks should focus on staff development in non-traditional banking, while state-owned banks could benefit from operational cost reductions. The proposed methodology also provides a robust framework for future studies on market power and efficiency in diverse industries.