OSFI Proposes Revisions to Guideline on Model Risk Management

The Guideline E-23 on model risk management was first issued in September 2017 for deposit-taking institutions. Since then, the supervisory work of OSFI has identified opportunities to provide greater clarity for deposit-taking institutions on certain elements: model risk management guidance at the enterprise-wide level, the scope of models to which the guideline applies, and the application of the proportionality principle toward smaller institutions. OSFI is now proposing to revise the scope of the guideline on model risk management by extending its application from the deposit-taking institutions only to include all federally regulated insurance companies and federally regulated pension plans. OSFI also plans to expand the scope of this guideline to address the emerging risks from digitalization and use of advanced analytics (including artificial intelligence and machine learning). OSFI is seeking input from stakeholders on the expanded scope of application and models, along with any other element of the current Guideline E-23 where additional detail or greater clarity would be beneficial. OSFI plans to launch a consultation on Guideline E-23 in March 2023, with final guidance planned for publication by the end of 2023 and target implementation by June 2024.

With the increase in use of models, the scope of Guideline E-23 will be enhanced to include models used beyond capital calculation and risk management, following a risk-based approach. The updated guidance will reflect the extent to which model governance structures and frameworks may need to be enhanced in terms of lines of accountability to cover multidisciplinary model risk management, interrelationships between models and data, technology advancements and evolving model risks, and potential opacity of models and third-party dependencies and their effect on model outcomes and results. Below are some of the aspects of model risk that OSFI is considering in the update to the Guideline E-13:

• Expanded model risk challenges, suggesting stronger coverage of controls and governance through data lineage
• Strengthening the rigor employed by model owners, users and validators
• Appropriate frequency and intensity of monitoring depending on the risk of models
• Different types of bias that could manifest in models; for example, unwanted bias that can lead to fairness considerations, which is one of the principal evolving topics in the artificial intelligence space
• Appropriate level of documentation, commensurate with model risk, while being sensitive to Industry trends toward agility in model development and the opportunity to leverage platforms as part of the model lifecycle
• Enhance the scope to include models used beyond capital calculation and risk management, following a risk-based approach

Related Links

• Proposed Revisions
• Guideline E-23, September 2017

Keywords: Americas, Canada, Banking, Insurance, Artificial Intelligence, Guideline E-23, Modeling Risk, Regtech, Model Governance, Credit Risk, Lending, Basel, Regulatory Capital, Stress Testing, OSFI, Headline