Model risk: critical points (advanced analytics, data and automation)

What it is about

Hernán Enríquez, of Management Solutions, explained at the First Annual Risk Conference (September 22, 2023) the critical points of model risk management in an environment of advanced analytics, data and automation.

The 25 slides start from what counts as a model, the evolution of supervision toward a comprehensive approach, active model risk management and the prioritization of models to validate. They address five axes (documentation, data, methodologies, systems and processes, and integration into management), the potential for automation, non-regulatory models, the European Union artificial intelligence regulation and the phases for implementing a framework.

Objectives

• Explain what model risk is and why managing it has become more demanding.
• Present the components of a model risk management framework and how to prioritize validation.
• Identify where there is the most efficiency and automation potential in the model life cycle.

Key takeaways

1. Active model risk management requires a concrete framework with action plans and prioritizing the models to validate according to the risk associated with each.
2. The five axes of the framework are documentation, data, methodologies, systems and processes, and integration into management.
3. The greatest efficiency is found in data and documentation; a team of technical writers is one of the levers cited.
4. A mature model risk management culture involves the participation of those responsible across the three lines of defense.
5. The potential for automation varies by phase of the model life cycle, and non-regulatory models (collections, early warning, among others) also fall within scope.
6. There is still a long way to go in maturity, and efficiency is the next goal for advanced institutions.