Abstract:
For predicting future volatility, empirical studies find mixed results regarding two
issues: (1) whether model free implied volatility has more information content than
Black-Scholes model-based implied volatility; (2) whether implied volatility outperforms
historical volatilities. In this thesis, we address these two issues using the Canadian
financial data. First, we examine the information content and forecasting power between
VIXC - a model free implied volatility, and MVX - a model-based implied volatility.
The GARCH in-sample test indicates that VIXC subsumes all information that is
reflected in MVX. The out-of-sample examination indicates that VIXC is superior to MVX
for predicting the next 1-, 5-, 10-, and 22-trading days' realized volatility. Second, we
investigate the predictive power between VIXC and alternative volatility forecasts
derived from historical index prices. We find that for time horizons lesser than 10-trading
days, VIXC provides more accurate forecasts. However, for longer time horizons, the
historical volatilities, particularly the random walk, provide better forecasts. We conclude
that VIXC cannot incorporate all information contained in historical index prices for
predicting future volatility.
