Problem:
A financial institution received a letter from a Federal regulator ordering a review and update of its credit risk models by a third party. The credit risk models in place had evolved over time and were owned by different parts of the institution.
Our solution:
The most severe problem was a lack of documentation of the models, inputs, uses, and performance. We started by instituting a common standard for documentation of the models, a common standard for creating of data dictionaries for the inputs, and interviews with users and potential users unknown to the developers. We then created a benchmark database and ran the hypothetical data through the system to highlight problems for users. We then systematized the credit risk modeling to create one overarching process that met the needs of multiple users.
Benefit to Our Client:
Three benefits followed: the Federal regulators received documentation for a rehabilitated system, “owners” of systems within the institution discovered “users” of which they hadn’t previously been aware, and risk managers were now able to negotiate data needs with operational groups.