Reverse Stress Testing

Abstract

This article outlines a framework for the analysis of extreme events based on forward-looking reverse stress testing. We carry out a portfolio simulation and identify stress scenarios which are critical for bank solvency as the ones contributing the most to cost of capital, as expressed by KVA scenario differentials.

Applications include model validation, trading limits, model risk management and hedging. A pricing model is invalid if it breaks on a path leading to stress conditions, causing alpha leaks that go undetected in market risk models such as value-at-risk (VaR), expected shortfall and stressed VaR. Trading limits are best predicated on incremental cost of capital and model risk capital can be assessed by computing cost of capital with Bayesian averages. Stress scenarios also have the potential to suggest risk-reducing hedges.

Keywords: Model Validation, Stress Testing, Reverse Stress Testing, Capital Models, Risk Margins, Trading Limits, Cost of Capital, KVA, Model Risk, Short Rate Models, PFE

JEL Classification: D81, G13, G28, G32