Setting Scenarios and Assessing Scenario Probabilities under IFRS 9 Accounting

An interpretation of the new standard (IFRS 9) on accounting for loan loss provisioning is that financial institutions are required to compute expected credit losses over a range of economic scenarios and then weigh them together according some probability based scheme. This poses substantial challenges theoretically as well operationally. In this short paper we address some related to the incorporation of economic outlooks. In particular, we pair the Merton approach to credit risk and Gauss-Hermite quadratures and propose a framework that enables the setting of the scenarios directly as well as indirectly with target "CDF"-view probabilities while achieving the "PDF"-view probabilities required from the computational point of view. We illustrate the approach numerically and demonstrate that the number of scenarios required for practical sufficiency is small. In addition, when the credit scenarios are set indirectly the scenarios have to be set quite far out in the tails