A coherent economic framework to model correlations between PD, LGD and EaD, and its applications in EaD modelling and IFRS-9

This paper proposes an economic framework recognising EaD as a stochastic variable and capturing the PD–LGD, PD–EaD and LGD–EaD correlations. It explains how these correlations can be estimated from historical data, and how PD, LGD and EaD can then be simulated in determining credit VaR. The framework allows credit losses to be more accurately captured, both in terms of the expected credit losses (ECL under IFRS-9 and CECL) and the unexpected tail events in measuring Credit VaR. The framework quantifies the potential underestimation of the tail risk in Credit VaR and the IFRS-9 ECL if the full correlation structure is not captured. By explicitly modelling EaD in a correlated fashion with PD and LGD, lenders can understand and model the increase in funding requirements during downturns. Application in back-testing IFRS-9 ECL is discussed and supplemented by a numerical example.