Stress Testing with Volatility-Weighted Historical Simulation: A Comparison of HS Methods

Abstract

The stress tests recommended by most central banks, for their supervisory review process, do not consider the probability of occurrence of the abnormal shocks. They also focus on worst-case scenarios and typically ignore the diversification benefits in a trading portfolio. However, stress tests based on simple historical simulation fail to affect VaR or the level of stress capital. Therefore, using data on USD-INR and GBP-INR between January 2000 and February 2001, we estimate the impact of stress scenarios on VaR and Expected Shortfall using volatility-weighted historical simulation. Our analysis takes care of fat tails and volatility clustering in a simple framework.

Keywords: Extreme Losses, Volatility updation, diversification, Fat tails, nonparametric methods

JEL Classification: G21, G32