Considering the dependence between the credit loss severity and the probability of default in the estimate of portfolio credit risk: an experimental analysis

Abstract

In this paper a simple and innovative model for measuring more accurately the credit tail risk of a banking book is presented. This is a Monte Carlo simulation model in which the credit loss severity (LGD) is a stochastic variable and it is correlated to the default event. Specifically, LGD is assumed to be distributed as a conditional beta function and its two parameters a and b are estimated assuming a mean value of LGD linked to the value of the PD conditional to the value of the macro-economic risk factor generated in every Monte Carlo simulative scenario. The linkage between the average LGD and the conditional PD is obtained by a simple linear regression analysis calibrated by using the time series of easily available financial historical data (Moody’s, 2011; Standard & Poor’s, 2012).