An empirical comparison of methods for incorporating fat tails into value-at-risk models

Vijay Pant and Weita Chang

Abstract

ABSTRACT

The authors compare three techniques commonly used to incorporate fat-tailed asset returns in value-at-risk (VaR) models - modeling the asset returns as a Bernoulli mixture of normal distributions, as a Student t-distribution, or using a stochastic volatility model. In order to perform this comparison, the VaR is computed for three sample portfolios using each of these models over a period of time. These VaR estimates are then compared with each other in order to understand the magnitude of the possible variance in VaR due to the choice of model.