Identifying Stress Test Scenarios

The usefulness of stress tests as risk management tool crucially depends on the choice of stress scenarios: if the scenario is too implausible, the stress test results will be ignored by management. Unfortunately, there is no generally accepted standard how to choose scenarios. Many institutions apply some standardized scenarios derived from extreme market moves in the past. A substantial drawback of such scenarios is the fact that they usually ignore the characteristics of the portfolio under consideration. Here we make two points: (1) We argue for the importance of taking into account portfolio characteristics when choosing stress scenarios, and (2) we compare four different methods in terms of the loss incurred by the resulting scenarios and the plausibility of these scenarios. These methods are: the standard scenarios proposed by the Derivatives Policy Group, naive modelling of historical crisis, portfolio-dependent modelling of historical crisis, and Monte Carlo search for worst case scenarios. Portfolio -dependent methods of scenario identification – in particular Monte Carlo search – produce scenarios which are at the same time more severe and more plausible. In passing we introduce a new measure of plausibility for stress scenarios. ∗ The statements made herein are the authors' opinions, and do not necessarily reflect the views of the Fachhochschule Vorarlberg or of the Oesterreichische Nationalbank. † e-mail: [email protected] ‡ e-mail: [email protected] T. Breuer, G. Krenn Identifying Stress Test Scenarios