Risk Management, under the direction of the Firm’s CRO provides a Firmwide independent function for the management of risk. This function provides oversight of the risks inherent in the Firm’s businesses. Each line of business is responsible for decisions relating to risk strategy, policy, measurement and control within its respective business.
The Model Risk and Development Group (MRaD) provides oversight of the Firmwide Model Risk policy, guidance with respect to a model’s appropriate usage, and conducts independent reviews of models. In addition MRaD develops internal models related to measurement of counterparty risk, economic risk capital, and regulatory risk capital measures.
The Firm-wide Model Risk Governance Group focuses on monitoring and evaluating quantitative methodologies used throughout the firm. In particular, the group reviews economic and regulatory capital models, analyses complex model risk, and assesses appropriateness of risk measurement and reserve methodologies in major lines of business (Commercial Bank, Treasury Services, Asset Management, Corporate Treasury, Chief Investment Officer etc.). Team members have opportunities for exposure to a variety of business areas and research projects.
- Conduct reviews of models used in connection with pricing, loss forecasting, and capital calculation for wholesale products
- Evaluate conceptual soundness of model specification; reasonableness of assumptions and reliability of inputs; completeness of testing performed to support the correctness of the implementation; robustness of numerical aspects; suitability and comprehensiveness of performance metrics and risk measures associated with use of model.
- Design and implement experiments to measure the potential impact of model limitations, parameter estimation error or deviations from model assumptions; compare model outputs with empirical evidence and/or outputs from model benchmarks.
- Liaise with front office, Finance and Risk professionals to monitor usage and performance of the models and syndicate the findings of model validation.