Linear Models in Credit Risk Modeling

Linear models have been used in various applications. Credit risk analysis is an important area which relies on linear regression models. The objective of this article is to illustrate briefly the role of linear models in credit risk analysis. However, correlation is a crucial aspect within this context in general. On the one hand, we want to avoid correlation between the independent covariates within a regression-like context when developing a rating model on obligor level. On the other hand, we need to capture the correlation when estimating a loss distribution. This does not affect the two models themselves as they are separate models but not modules within a common framework. It is interesting to know how important correlation is and how it enters into the model development from a practical point of view. After a short note on measuring linear correlation, we give some ideas about the development of rating models from a practical point of view. The section about estimating the loss distributions will be more theoretical and is followed by some notes on the simulation of losses.