Multi-Period Credit Default Prediction: A Survival Analysis Approach

摘要: The book deals with the problem to estimate credit default probabilities under a flexible multi-period prediction horizon. Multi-period predictions are naturally desirable because maturity of loans usually spans several periods. However, single-period models largely prevail in literature so far due their simplicity. Predicting over multiple periods indeed entails certain challenges that do not arise within view. Among main contributions this work is show there relatively simple solutions these available. From methodological point view, survival analysis approach used. In context, time until (or lifetime) central variable investigation as opposed traditional reducing information binary representing event. Modeling has advantage both timing events and censored data utilized. Since issues gain importance horizon grows it no coincidence selected for problem. The results following. First, new index measuring predictive accuracy proposed its advantages commonly used indices shown theoretically by an empirical analysis. This part second chapter which further includes methods statistical inference index. third chapter, case panel datasets time-varying covariates dealt with. A developed simpler than available far. study concerning North American public firms, we provide evidence delivers more accurate competitors well. final assigning probability estimates given rating grades examined. If rare, standard approaches have drawbacks. As alternative, Bayes presented mitigates effects sparseness. estimator applied comprehensive sample sovereign bonds. findings capital requirements bonds likely be underestimated using but when estimator.