Bank Examination Data as Predictors Of Bank Net Loan Losses

A study was conducted to evaluate ways of forecasting the level of loan losses. Three models were used in the study. The Benchmark Model assumes a relationship between loan losses and the aggregate loan levels as a means of forecasting loan loss levels. The Prediction Model using Classified Loan Data bases the loan loss forecast on a relationship using data from on-site examination by banks with respect to loans which have a greater than normal risk of default. Finally, Ratio Models base forecasts on the ratio of net loan losses to total loans and can be applied to almost any relationship. The study showed models based on the use of classified loan data provided more accurate forecasts of loan loss levels. However, the simplier models were more accurate in forecasting loan loss rates. These general results did vary to some degree depending on the size of the bank.