Objective

To test and validate scenario computation models used by a European Global Bank to compute PnL under shocks to various market factors.

CRISIL's Solution

• Assessed methodology documents quality in accordance with the SR 11-7 requirements
• Conducted input data assessment spanning data quality, data fit-for-purpose, data governance, and data assumptions and limitations
• Verified model implementation at the risk-factor level, product level, and book level
• Identified 32 material products across asset classes, and conducted performance testing using GMAG FO-RMS pricers
• Performance testing included comparing the three computation approaches (full reval., partial reval., and SBA). This was done using a range of hypothetical shocks to various risk factors encompassing regulatory scenarios
• Verified the appropriateness and adequacy of model governance and ongoing monitoring activities
• Validated associated models, such as error correction models for IR, FX, and equity, along with de-arbitrage models, such as FX skew de-arb model and IR smoothing model
• Conducted the annual revalidation of material risk factors, such as equity spot and vol., FX spot, and credit spot risk factors
• Established minimum standard guidelines and acceptance thresholds for model use
  

Client Impact

• Conducted the first comprehensive validation of all the market-risk stress-testing models across the bank's legal entities
• Established minimum standard guidelines for the use of partial revaluation and sensitivity-based approaches for different asset classes, product types, risk factors, and scenarios
• Successfully conducted IT regression testing of the Excel-based pricers with front office systems for 35 material products
  

Questions

Looking for high-end research and risk services? Reach out to us at: