EBA Guidelines on the LGD under an economic downturn

On the 6th March 2019 the European Banking Authority (EBA) published its final Guidelines specifying how to quantify the estimation of loss given default (LGD) appropriate for conditions of an economic downturn.

The Guidelines set out requirements for the appropriate quantification of the calibration target used for downturn LGD estimates. As EBA have specified they include three types of approaches based on the level of sufficient amount of available data on losses:

• “Type-1 approaches can be applied when banks have sufficient loss data for the identified downturn period. In this case, institutions are allowed some modelling flexibility, but subject to a harmonised and prescriptive impact assessment;
• Type-2 approaches can be applied when banks do not have sufficient loss data for the identified downturn period. In this case, institutions are given the choice between two approaches, the so-called haircut or extrapolation approaches This will harmonise the approaches used by banks.
• Type-3 approaches can be applied in rare cases, where neither type-1 nor type-2 approaches can be used. In this case, banks have to apply a minimum margin of conservatism (MoC) requirement of 15 percentage points on LGD estimates.”

Moreover, a reference value has been introduced that plays the role of a non-binding challenger to the final downturn LGD estimation. It is calculated at least for each calibration segment as the simple average of the average realised LGDs from the two individual years with the highest observed economic losses. The reference value is not used as a calibration target but in the case of a material difference between the reference value and the downturn LGD estimates, it must be justified as specified in the Guidelines.

The Guidelines are part of a wider EBA effort to review the IRB approach aimed at restoring confidence and repairing IRB models. They also follow the RTS on the specification on economic downturn about which we have informed in a separate article.