In this class we deal with CreditMetrics™, an interesting structural model of default, which can be seen as a further derivation of Merton’s one.

Introduced in 1997 by JP Morgan, CreditMetrics has some interesting features:

• The default threshold is defined through credit ratings, and not using liabilities. These ratings can be external or internal.
• Therefore, CreditMetrics overcomes the problems related to the correct definition of the liability threshold, as outlined in Merton’s and Moody’s KMV models.
• Since thresholds are defined through credit ratings, CreditMetrics allows for the computation of both the PD and the probability of credit deterioration.
• As we will see with an example, the computation of the thresholds is rather simple, provided that we accept the Normality assumption.

Naturally, CreditMetrics also has some points of weakeness. For example, the fact that it heavily relies on the Normality assumption may cause an underestimation of tail risks. In fact, as we have said when discussing Value-at-Risk, the financial world is rarely Gaussian, given the non-negligible probability of extreme events.

In this course we will just consider the univariate version of CreditMetrics: the multivariate case would require some more advanced knowledge of probability.
However, I can guarantee that you will grasp all the basic features of the model.
For those of you interested in a more complete understanding of the model, the original CreditMetrics™ report is available in the Extra Materials session of this week.