LGD Credit Risk Model: Estimation Of Capital With Parameter Uncertainty Using MCMC
This paper investigates the impact of parameter uncertainty on capital estimate in the well-known extended Loss Given Default (LGD) model with systematic dependence between default and recovery. The authors demonstrate how the uncertainty can be quantified using the full posterior distribution of model parameters obtained from Bayesian inference via Markov Chain Monte Carlo (MCMC). Results show that the parameter uncertainty and its impact on capital can be very significant. They have also quantified the effect of diversification for a finite number of borrowers in comparison with the infinitely granular portfolio.