Early warning signals of bank distress: machine learning techniques and shapley values analysis

Grasso, Emmanuela; Fortugno, Filippo; Cheng, Long

Early warning signals of bank distress: machine learning techniques and shapley values analysis

License

Citation

Grasso, E, Fortugno, F, Cheng, L. Early warning signals of bank distress: machine learning techniques and shapley values analysis. 2023. handle: http://hdl.handle.net/10230/58428

This citation was generated automatically.

Abstract

The banking sector plays a crucial role in the economy. Therefore, being able to anticipate banks’ financial distress is key to the overall economic stability. In this study, using a dataset for European and United Kingdom banks for the period 2006-2021, we develop an early warning system for banks’ financial distress. To this aim, first, we employ several machine learning techniques and compare their performance in order to assess whether bank’s financial distress episodes are predictable and individuate the model that best serves this aim. Second, we complement the machine learning methods with the Shapley Values approach in order to identify the key drivers of a bank’s financial distress. Indeed, identifying the factors that signal a bank’s vulnerability to financial distress could ultimately serve as a guide for regulators and banks’ management in their preventive and corrective actions.