Article ID: | iaor200969438 |
Country: | United Kingdom |
Volume: | 27 |
Issue: | 8 |
Start Page Number: | 670 |
End Page Number: | 689 |
Publication Date: | Dec 2008 |
Journal: | Journal of Forecasting |
Authors: | Choudhry Taufiq, Wu Hao |
Keywords: | Kalman filter, GARCH |
This paper investigates the forecasting ability of four different GARCH models and the Kalman filter method. The four GARCH models applied are the bivariate GARCH, BEKK GARCH, (the acronym BEKK comes from Baba, Engle, Kraft, and Kroner) GARCH-GJR (Glosten-Jagannathan-Runkle) and the GARCH-X (the cross-sectional volatility) model. The paper also compares the forecasting ability of the non-GARCH model: the Kalman method. Forecast errors based on 20 UK company daily stock return (based on estimated time-varying beta) forecasts are employed to evaluate out-of-sample forecasting ability of both GARCH models and Kalman method. Measures of forecast errors overwhelmingly support the Kalman filter approach. Among the GARCH models the GJR model appears to provide somewhat more accurate forecasts than the other bivariate GARCH models.