Simple Logit Credit Risk Model*

This point-in-time credit risk classification model is relatively simple, easy to interpret and stable model for use across a broad spectrum of non-financial corporate entities.  The model uses measures of leverage, cash flow, liquidity and size, all frequently cited parameters used in many credit risk models.  The model was developed using twenty-three years of data from merged Moody's and Compustat databases.  This large volume of data sets covering an extensive time period are responsible for the stability of the model across in-same and out-of-sample testing and also for the tight balance obtained between Type I and Type II errors.

While the model was not developed with the objective of maximizing its accuracy in predicting actual defaults, it appears to be reasonably good in classifying entities using a five point scale of risk grades.

As a logit model, its score can be translated into a probability of default extending its usefulness as a broadly applicable front-end process for portfolio credit risk processes.

 

* See Parameterizing Credit Risk Models With Rating Data, October 18, 2000; Mark, Carey; Hrycay, Mark.