The impact of financial statement indicators on bank credit ratings: Insights from machine learning and SHAP techniques

초록

This study investigates the influence of financial statement indicators on bank credit ratings. We construct a dataset encompassing 53 banks and 28 key financial indicators and employ two machine learning models, GBR and LightGBM, to predict credit ratings based on these indicators. To understand the contributions of the individual indicators, we apply SHapley Additive exPlanations (SHAP) to interpret the forecasting results. The analysis reveals that indicators pertaining to a bank's revenue structure, particularly net interest income, have a significant impact on credit assessments. This finding underscores the critical role of a bank's debt repayment capacity and income stream diversification.

키워드