We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Forecasting the Nikkei spot index with fractional cointegration.
- Authors
Lien, Donald; Tse, Yiu Kuen
- Abstract
The article presents a study of forecast performance of fractionally integrated error correction model against error correction and vector autoregression models for Nikkei Stock Average spot index. To examine the performance of the models in long-horizon prediction, researchers considered the forecasts of the Nikkei spot index up to 40 days ahead. Researchers incorporated the generalized autoregressive conditional heteroscedasticity structure into the models. Researchers found that, when heteroscedasticity, the fractional cointegration does not improve forecast. When conditional heteroscedasticity is correctly incorporated into the model together with fractional cointegration, researchers obtained the best forecasting performance for step sizes larger than twenty. The results are consistent with the notion that cointegration or fractional cointegration is important only for long-run predictions. Incorporating fractional cointegration in an error correction model improves the forecasting performance over conventional error correction models.
- Subjects
COINTEGRATION; HETEROSCEDASTICITY; NIKKEI 225; FORECASTING; TIME series analysis; STOCK prices
- Publication
Journal of Forecasting, 1999, Vol 18, Issue 4, p259
- ISSN
0277-6693
- Publication type
Article
- DOI
10.1002/(SICI)1099-131X(199907)18:4<259::AID-FOR723>3.0.CO;2-7