Risk model development often faces the challenge of imperfect data. Histories can be short, quality may be lacking, and in some cases the number of test cases may be small – for example, with low-default portfolios.
Recent supervisory guidance in SR15-18 and SR 15-19 explains the importance of models being able to stand up to the rigours of validation. It is essential to have not only high quality primary ‘Champion’ models, but also high-quality Challenger/Benchmark models.
Protiviti’s Model Risk Management team has models to address a wide range of financial issues including credit risk, operational risk, market risk, and stress testing and forecasting. Our Models team can assist with development of Champion models as well as Challenger/Benchmark models. We have models for a variety of use cases, including stress testing (CCAR/DFAST) as well Scorecards, Credit Risk, Market Risk, Operational & Risk, Economic Capital/Solvency, and Liability Management.
Clients select us for the quality of our models, as well as our ability to overcome limitations and deal creatively with difficult datasets.
Our areas of expertise:
Stress Testing & Forecasting Models
Protiviti has built stress testing models for several large U.S. Banks for both CCAR and DFAST stress tests. Protiviti, evaluated correlation between economic and bank specific input variables and risk metrics to develop models for stress testing projections.
Credit Risk Scorecards
Protiviti has supported clients in developing and refining credit risk scorecards. A significant challenge in these projects has been addressing low default portfolios.
Allowance Models are subject to the new CECL/IFRS9 requirement. These models can be built using a variety of methodologies. Our Models team has led industry-wide presentations on CECL, IFRS9 and Allowance requirements, and has extensive experience with all leading modelling approaches.
Protiviti has led the development of operational risk models for Basel II mandatory U.S. bank and banks building oprisk models for other purposes. For these model builds, we influenced all key developmental decisions and drafted the documentation to support the approaches used. Our work has led to OCC approval of the framework and model for operational risk.
AML Customer Risk Scorecards
One recent ‘guideline’ from OCC regarding AML Customer Risk Scorecards is that the financial institutions should have a well-defined approach to quantify the BSA/AML risks of the bank’s customers. Our team has experience assisting client design global KYC standard, enrich customer KYC data, and develop centralized Customer Risk Rating (CRR) Model using statistic tool ‘SAS’ and ‘un-supervised’ machine learning techniques.