Forecasting bankruptcy and physical default intensity

This report presents two of our investigations: one is to obtain an accurate forecast for the corporate bankruptcy; the other is to obtain a physical default intensity. Both investigations were based on the hazard model, using only firm-specific accounting variables as predictors. Different methods, such as the list-wise deleting, closest- value imputation and multiple imputation, were applied to tackling the problem of missing values. Our empirical studies showed that the multiple imputation performed the best amongst these methods and led to a forecasting model with economically reasonable predictors and corresponding estimates.