On 7 August 2023, the European Banking Authority (EBA) updated its roadmap for the implementation of internal ratings based (IRB) model requirements to limit compliance costs for institutions.
In July 2019, the EBA issued a progress report and set a new timeline for the implementation of the requirements of its so called ‘IRB repair program’. The EBA extended the deadline until the end of 2023 for loss given default (LGD) and credit conversion factor (CCF) models that cover portfolios that will no longer be eligible for the advanced internal ratings based (AIRB) approach under the Basel III framework.
The EBA has now stated that due to the upcoming implementation of the final Basel III standards in the EU framework, the implementation of the IRB repair requirements for LGD and CCF models that cover portfolios no longer eligible for the revised AIRB approach may be postponed to the date of entry into force of the future Capital Requirements Regulation (CRR 3). Within that period, institutions may also choose to apply for permission to return to a less sophisticated IRB approach or for the permanent partial use of the standardised approach for those portfolios, according to Articles 149 and 150 of the CRR.
The possibility of postponing the implementation of the IRB roadmap provisions does not apply to any probability of default models, or to those LGD or CCF models that have in their scope of application exposures that may remain under the AIRB approach.
The EBA has also published an IRB validation handbook which provides guidance on the validation functions, as set out in Article 185 of the CRR. The handbook provides an overview of the validation framework and describes the elements where the validation function is expected to form an opinion, without prescribing any specific methodology. It also covers the relationship of the validation function with other functions related to corporate governance and the work of the validation function in the model cycle, as well as when using external data, outsourcing validation tasks, and in a situation of data scarcity.