Backtesting for counterparty credit risk

Sebastian Schnitzler, Niklas Rother, Holger Plank and Peter Glößner

Abstract

ABSTRACT

Backtesting for counterparty credit risk (CCR) constitutes a major challenge for risk and trading departments in banks that use internal models or have an interest in calculating their credit default exposures for potential future exposure limitation or credit valuation adjustment trading purposes. A market standard for long-term exposure validation has not yet been established, despite the fact that several software solutions are already available. This may be due to the lack of legally binding regulations at present. In this paper, we discuss various problems around regulatory requirements for backtesting.We argue that it is possible to set up a backtesting framework that works for both full revaluation and American Monte Carlo techniques, and which achieves high performance via the reuse of previously simulated data. Our proposed framework combines market risk methodology (such as the traffic light approach) and statistical methods (such as goodness-of-fit tests for uniform distributions). Among other things, we investigate problems related to the long-term aspects of CCR and show that, due to long time horizons and overlapping data, it is necessary to enhance the existing market risk backtesting frameworks. Furthermore, we introduce several publications that discuss details of or alternative approaches to exposure backtesting. The results of this paper may be of interest to financial institutions and suppliers who intend to validate their long-term credit default exposure models.