Keijsers, Bart, "Relating LGD for Bank Loans to the State of the World", MET, Vol. 20, No. 1, (2013), 4-9.
Abstract: Credit risk is arguably the most important risk to banks. The borrowers they lend to can default, such that they are not able to repay their debts. This can be quantified ex-post by the default rate (DR) and the loss given default (LGD). Using the PECDC database, the world’s largest bank loans LGD database, we show that the LGD for bank loans is related to macroeconomic variables. Secured loans are related to interest rates and unsecured loans to the oil price index. The relation with these variables implies there is in fact downturn LGD for bank loans. We use state space methods to construct a latent (unobserved) variable from the observed DR and LGD. We can relate the latent variable to macroeconomic variables, so it seems to reflect the state of the world. It can be used to adjust the LGD estimates to account for the cyclicality.