21 June 2022
The European Banking Authority (EBA) today published four principles to support supervisory efforts in assessing the representativeness of COVID-19-impacted data for banks using internal ratings based (IRB) models. These principles will be part of a supervisory handbook, which the EBA will publish later in 2022 with the objective to ensure a harmonised approach in the use of COVID-19 data, especially where the use of moratoria and other public measures may have led to changes in default rates.
The EBA has received requests for a timely clarification of the requirements covering the representativeness and usage of IRB-relevant data impacted by the COVID-19 pandemic and the countering measures in terms of validation and calibration. The four principles are also of general interest, especially in the context of other modelling practices, such as IFRS9, and will allow supervisors and banks to include them in their current considerations.
The first principle clarifies that the guidance on the assessment of data representativeness laid down in the EBA Guidelines on PD and LGD should be applied also in the case of COVID-19-impacted data. The second principle clarifies that a significant decrease in applied IRB risk parameters compared to the pre-crisis levels indicates a potential lack of representativeness and should be analysed in more depth. The third principle deals with the default and loss rates observed during the pandemic and clarifies that in case of non-representativeness of such rates, a recalibration should be postponed to lower long-run averages. Finally, the fourth principle tackles the validation and recalibration of downturn LGD in the context of the COVID-19 pandemic. Here, the EBA recommends that potential downward recalibrations be postponed at least until the effects of the crisis have fully materialised in the observed loss rates.
The EBA is currently working on a supervisory handbook, which aims to clarify the existing requirements applicable for the validation of regulatory credit risk models, i.e. under the IRB approach. This handbook will clarify the expectations applicable to the so-called ‘validation function’, both in terms of assessment of the model performance and modelling environment, as well as in terms of process and governance.