Abstract

This research evaluates the creditworthiness of Lebanese banks using the Bayesian Naïve Classifier (BNC) in the CAMELS framework. Using the CAMELS indicators—capital adequacy, asset quality, management efficiency, earnings, liquidity, and sensitivity to market risk—the study examines data from 2012 to 2022, a period also marked by financial instability. The complex interdependencies between these variables are modeled using the BNC, a machine learning technique that provides a probabilistic approach that improves prediction accuracy. In order to assess how well the BNC predicts banks’ ratings, training and testing datasets are created. The findings indicate that the most important elements influencing bank ratings are capital adequacy, management efficiency, and asset quality. Liquidity and sensitivity to market risk become more significant during economic downturns, especially following the 2019 financial crisis in Lebanon. With a predicted accuracy of more than 98%, the BNC proved its resilience and dependability in identifying patterns that traditional models would miss. By incorporating machine learning into the CAMELS framework, this study presents an innovative approach to credit risk assessment and offers insightful information to investors, regulators, and decision-makers who are keeping an eye on the stability of financial institutions. To further confirm this model’s resilience in many economic contexts, future studies should extend its use to more industries and geographical areas.

This work is licensed under a Creative Commons Attribution 4.0 License.