Held in Madrid and streamed online, the event brought together leading risk and regulatory experts to analyse developments in credit risk modelling, with a focus on the balance between IRB and the Standardised Approach, model governance, and the integration of advanced modelling techniques.
Keynote Address – Stephen Woulfe (European Central Bank)
Stephen Woulfe presented upcoming supervisory reforms aimed at reducing complexity and reinforcing a risk-based approach within model oversight. Notably, he outlined changes including a tiered classification of findings and the delegation of closure responsibilities for less severe issues—representing a significant operational shift.
- New tiered classification of findings: From July 2025, ECB Joint Supervisory Teams will implement a four-tier structure: F1/F2 (low severity) and F3/F4 (material severity).
- Autonomous closure of minor findings: Banks will be permitted to close F1 and F2 findings directly within ECB systems without prior supervisory validation, provided evidence is retained for five years.
- Focus on material risk: Supervisory efforts will concentrate on F3/F4 issues.
- Trust and accountability: Woulfe remarked that "trust is swiftly followed by responsibility".
- Cultural shift and operational relief: The approach aims to reduce administrative burden and foster more efficient supervision.
IRB Strategy and Simplification
Panellists discussed how shifting supervisory expectations and internal strategic considerations are reshaping the use of IRB. Rather than seeking maximum coverage, institutions are now adopting a more selective, portfolio-specific approach justified by modelability, data availability and cost-effectiveness.
Simplification emerged as a central theme, with many institutions reconsidering the value of IRB in volatile or non-material portfolios. Article 494d was noted as a helpful mechanism for reverting to the Standardised Approach, although obtaining formal supervisory approval remains challenging, particularly for cross-border banking groups.
Model Governance and Monitoring
Monitoring was highlighted as the cornerstone of model lifecycle management. Panellists agreed that redevelopment should be driven by clear performance evidence and that unnecessary redevelopment should be avoided to conserve resources and reduce supervisory scrutiny.
Practices are becoming more sophisticated, with greater attention to sub-model behaviour, data drift, and user interaction, including overrides. While automation is seen as key to improving efficiency, it should focus on routine tasks, leaving space for human judgement in critical areas such as scenario review and interpretation of results.
Advanced Techniques and External Data
While machine learning and generative AI hold great promise, their application in IRB models remains limited by regulatory requirements around explainability and control. These techniques are more commonly used in less regulated contexts such as early warning tools and scoring models.
FinTech competition is a key driver of modernisation. Unlike traditional banks, FinTechs can deploy models faster and leverage advanced techniques more readily. In response, banks are accelerating automation in development, validation and monitoring—aiming to complement, not replace, human expertise.
Supervisory Dialogue and Strategic Decisions
Panellists underlined the value of maintaining a constructive dialogue with supervisors. After the effort invested in IRB repair and model redevelopment, many institutions are reflecting on the broader benefits of IRB coverage.
Although simplification offers operational gains, it must be implemented carefully to maintain governance standards and model quality. Even without IRB, models must remain robust and well-documented for use in provisioning, credit decisions and capital planning.
The need for consistent supervisory implementation across jurisdictions was also discussed. Ultimately, any strategic shift away from IRB should be a board-level decision supported by a strong internal risk management framework.
Contribute to the Conversation
To explore these topics further and understand how institutions are adapting their modelling frameworks, we invite you to participate in our industry survey on the future of internal models. Your input will help build a representative picture of current practices, challenges and emerging priorities in IRB strategy, model governance and advanced modelling techniques.