The European Banking Authority (EBA) has published new standards to make approval processes for IRB model changes more efficient, reducing the number of changes classified as material and aligning with CRR3 updates.
The European Banking Authority (EBA) has introduced targeted amendments to its Regulatory Technical Standards (RTS) on material model changes for banks using Internal Ratings Based (IRB) models. These amendments aim to reduce the number of changes classified as material, enabling a more risk-based supervisory approach.
The revisions address the high volume of material model change applications, which have caused delays in approval timelines and hampered banks’ ability to implement improvements. The updated RTS rely more on quantitative thresholds, significantly decreasing the number of changes requiring prior approval, while maintaining supervisory oversight.
Qualitative triggers are now limited to changes involving model redevelopment, risk parameter re-estimations, or significant changes to default definitions. Routine model maintenance changes generally require notification unless they exceed the quantitative thresholds.
The RTS have been aligned with updates under the Capital Requirements Regulation III (CRR3), removing references to approaches no longer part of the prudential framework, such as the IRB approach for equity exposures and the Advanced Measurement Approach (AMA).
These efforts complement the European Central Bank’s (ECB) work to simplify approval processes for IRB model changes, supporting more efficient supervisory procedures across the EU and enabling quicker approvals for banks.
The EBA’s work is mandated under Article 143(5) of Regulation (EU) No 575/2013, aiming to specify conditions for assessing the materiality of IRB rating system changes. The revisions also support broader improvements in the EU regulatory and supervisory framework, as outlined in the EBA’s reports on supervisory efficiency.
Coordination with supervisory authorities, including the ECB, continues to promote more efficient supervisory processes for IRB models.