Nonparametric Forward-Looking Value-at-Risk

Abstract

This paper proposes a new model for computing value-at-risk forecasts. The model is fully nonparametric and easy to implement. Further, it incorporates information about the market’s perceived uncertainty about the future. The forward-looking information is obtained from the option market via the Chicago Board Options Exchange’s implied volatility index (VIX). Using S&P 500 data from 1990 to 2010 we find that the use of option implied volatility compares favorably with generalized autoregressive conditional heteroscedasticity (GARCH)-type models in terms of forecast performance. By comparing the model primarily used in the banking sector to our new model, we find that a financial institution using our model has on average a lower market induced capital requirement (MCR). However, during the time period leading up to the financial crisis our model gives a 40% higher MCR.