Banque PSA Finance

Proportionality principle is absolutely necessary to adapt compliance exigencies to the real stakes and to avoid to make reviews an unbearable burden for institutions and NCAs
In a same manner, a proportionality principle is necessary to assess the independence of the validation function because it is very difficult in small and medium institutions to hire senior staff experienced on rating systems and completely separated validation and credit risk control unit sufficiently staffed and trained are difficult to afford.
The text is clear. However the necessity to cover a complete economic cycle should not lead to make an average of all the historics available, as it should not be representative of the current processes of origination of the institutions and all the improvements made for underwriting management.
Even though the arguments in favour of default-weighted LGD are real, in particular for Low Default Portfolios (cf. item (v)), it should be authorized to use exposure-weighted LGD calculation for RETAIL portfolios. We believe that the (i) argument is very strong, because the exposure-weighted LGD equals the LGD on a portfolio level, which has more economic sense. The use of default-weighted could create discrepancies in the case where an important pool of loans with very low EAD and low LGD can be find. A pool constituted with low EAD defaults may be difficult to be defined as homogeneous and may be difficult to be applicable to sound contracts because there is no model to assess for a sound contract the exposure at the moment of default. At least, if the default-weighted method was confirmed as compulsory, institutions should be allowed to assess average LGD after withdrawal of the non-material exposure defaults in the calculation basis. The application of this rule may trigger important developments for institutions as it may imply to redesign all LGD models and segmentations for those which apply exposure-weighted LGD.
Yes, they bring useful guidelines.
The majority of these rules should not have important impact on our models and RWA calculation. However, default-weighted LGD method may have significative implementation costs because it could lead to redesign numerous models.
Clarification of some important rules. We will be more prepared to the NCA's reviews which should lead to less recommandations as we may adapt our practices before their coming. It should also lead to assure consistency with competitors RWA.
Paul Quantin