Recalibrating credit risk models – a theoretical perspective with practical implications

Lawrence G. Antioch

Abstract

ABSTRACT

A standard recommendation in the literature is for banks to monitor model performance and, where appropriate, recalibrate models to ensure assigned probabilities of default (PDs) are consistent with actual PDs. Precisely when models should be recalibrated has been left largely for banks to determine and yet such decisions have a profound impact on their capital adequacy. In this paper, we develop a theoretical model that attempts to elucidate the underlying dynamics of the PD and the inherent challenges of recalibrating credit models and also offer some tentative decision rules.