View article

Recent advances in default risk modelling for corporate lending offer many choices to model key credit risk measures such as forward looking cycle average default probability for different rating classes, dependence among defaults, and future variation of default probability over the credit cycle. These measures are used in several credit risk management applications including economic and regulatory capital calculations, and loss reserving. However, limitations in data quality and quantity are creating serious challenges in implementation. Within data constraints, practitioners have to choose the functional form of a model and estimate the parameters of the chosen model. Any choice of the model form is a postulation, and data constraints introduce significant uncertainty around the point parameter estimates. This article illustrates these eternal challenges with practical examples, and suggests that practitioners …