Model validation methodologies for corporate probability of default (PD) modeling

The views and opinions expressed in this article are those of the thought leader as an individual, and are not attributed to CeFPro or any particular organization.

By Michael Jacobs, Jr., Lead Quantitative Analytics and Modeling Expert, PNC

In your opinion, what are some key characteristics of a good model validation methodology for corporate probability of default (PD) modeling?

A robust methodology should strike a balance between statistical or theoretical purity on the one hand, versus practical considerations on the other, examples being that PD models have to be acceptable to different types of end users (e.g., lines of business vs. regulators vs. portfolio managers), or that there are limitations to data, and that development as well as validation has to be conducted in the context of awareness regarding resource constraints.

Michael will be presenting at the 10th Annual Risk Americas Convention, which is taking place virtually on May 25-26. Click here to view the full agenda, and register for your complimentary* pass to the event.

*Risk Americas is a free to attend event for all who register from regulated financial institutions or government bodies. If you are representing a service provider, vendor or consultant there is a nominal fee to attend. CeFPro has the final word on all registrants.