The bank turned to PwC to develop, document, and implement credit loss stress testing models for its unsecured consumer products (credit cards and various unsecured line and loan products) in various regions across the globe for use in the 2017 CCAR submission.
PwC deployed a comprehensive team of experienced credit modeling specialists to assist the client with end-to-end model development activities including:
- Assessing data quality, suitability, and sufficiency for stress testing.
- Discussing suitable model variables with the business experts followed by comprehensive variables selection process.
- Developing competing risk statistical models for Probability of Default at the account level, as well as various types of Exposure at Default/Loss Given Default models and discussing candidate models with the client’s modeling team and business experts.
- Performing comprehensive statistical and performance testing including statistical diagnostic tests, back-testing, scenario analysis, stability testing, and sensitivity testing of the final candidate model(s).
- Developing comprehensive model documentation using the client’s documentation template.
- Developing code implementing the models into production environment and performing comprehensive system testing.
- Developing a comprehensive ongoing performance monitoring plan, process, and code.
- Supporting the client modeling team during the internal model validation review process and regulatory review.
- Providing end to end project management support across a large number of different modeling work streams.
Impact on client’s business
The delivered models advanced the bank’s 2017 CCAR submission by replacing existing models with stronger models in terms of methodology, development practice, documentation, and performance. The new models were well received by all client internal stakeholders including model owners and users, model validation and review, and senior business and CCAR workstream leadership.