Description:
When using historical data for Loss Given Default (LGD) modelling, it is advisable to check the homogeneity of obligors and exposures assigned to the same grade or estimate – said differently: it is important to prove that no further relevant risk differentiation is possible. Large Corporate unsecured exposures, given their global cash flow and assets, are often argued to be a global and homogeneous asset class. For which –due to the relative lack of internal historical data within banks– is often produced a set of base estimates of the unsecured LGD risk parameter. In this document, we provide the results of an analysis performed to investigate whether this asset class can be considered as homogeneous for LGD modelling. We performed univariate analyses by relevant potential risk differentiation drivers: Large (GSIFI) banks vs. other lenders, Exposure size, Lenders’ regions and Borrowers’ regions; for which no significant potential for risk differentiation was found. However, using bivariate analyses crossing Lenders and Borrowers’ regions – i.e. out of region lending, we found potential for further risk differentiation.
