Abstract
In this article, it is argued that the estimation error in Value-at-Risk (VaR) predictors gives rise to underestimation of portfolio risk. We propose a simple correction and find in an empirical illustration that it is economically relevant.
Acknowledgements
The financial support from the Wallander–Hedelius Foundation is gratefully acknowledged. The author thanks Kurt Brännäs for many valuable comments and suggestions. The article has benefitted from discussions with David Granlund. The author would also like to thank the participants at the 28th Annual International Symposium on Forecasting, Nice 2008, and the 2nd International Workshop on Computational and Financial Econometrics, Neuchâtel 2008.
Notes
1In the simulation exercise in Section III, the covariance between the and the
series was close to zero for all four sample sizes.