国际清算银行-银行信贷损失预测：监管数据的经验教训（英）-2023.9-31页_340kb

Summary of Banks' Credit Loss Forecasts Study

This report analyzes banks' credit loss forecasts using confidential supervisory data from 2009 to 2022, comparing expected loss (EL) rates with actual loss (AL) rates. Key findings include:

• EL rates generally fail to capture the time evolution of AL rates, missing trends and spikes, supporting the need for conservative capital requirements under Basel III.
• However, EL rates accurately rank-order credit risk relative to AL rates across banks, indicating usefulness in risk comparison.
• Forecast optimism correlates with periods of high bank profitability and financial overheating, measured by the credit-to-GDP gap.
• Data inconsistencies between supervisory and accounting sources limit analysis, but robustness checks reinforce findings.
• Regulatory reforms, such as through-the-cycle PD estimates, mitigate procyclicality, but forecasting challenges remain due to uncertainty in real-time loss predictions.

The study recommends further investigation into counter-cyclical buffers and better integration of macro indicators to improve loss-absorbing resources. Practical applications highlight the importance of conservative capital mapping for regulatory resilience.