Abstract

This study applies the methodology of Guan and Ederington (1998) to Canadian data. Historical volatility is estimated then forecast to examine the comparative performance of various time series models in forecasting volatility. Statistically significant regressions were provided that showed that forecasts of volatility produced low forecast errors, comparable to those found in Guan and Ederington. These errors, reported as the Root Mean Squared Forecast Error (RMSFE), were calculated using in-sample and out-of-sample data. Both static and dynamic forecasts were used. Static forecasts of volatility consistently produced lower forecast errors than dynamic forecasts.