Symmetric and asymmetric garch models for forecasting the prices of gold
Gold prices forecasts are of interest to many people. Gold prices however, change rapidly from period to period. In short, they are not constant. The change is not only in the mean, but also in the variability of the gold prices series. Daily gold prices per ounce, from January 3, 2000 to December 3...
Saved in:
| 主要作者: | |
|---|---|
| 格式: | Thesis |
| 語言: | English |
| 出版: |
2013
|
| 主題: | |
| 在線閱讀: | http://eprints.utm.my/id/eprint/47930/25/PungYeanPingMFS2013.pdf |
| 標簽: |
添加標簽
沒有標簽, 成為第一個標記此記錄!
|
| 總結: | Gold prices forecasts are of interest to many people. Gold prices however, change rapidly from period to period. In short, they are not constant. The change is not only in the mean, but also in the variability of the gold prices series. Daily gold prices per ounce, from January 3, 2000 to December 31, 2010 is used in this study with the Schwarz Information Criterion (SIC), Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) as the forecasting accuracy measures. For the purpose of this study, gold prices from ten major consumer countries are examined. The currencies are American dollar, Australian dollar, Canadian dollar, Swiss franc, Chinese renmimbi, Egyptian pound, Euro, Japanese yen, Turkish lira and South African rand. This study considers five models from the GARCH-family namely the Generalized Autoregressive Conditional Heteroscedasticity (GARCH (p, q)), GARCH-M, Power of GARCH (PGARCH), Threshold GARCH (TGARCH) and Exponential GARCH (EGARCH). These models are analyzed by using the E-Views 6.0 software. Several combinations of p and q values are considered to develop several GARCH (p, q) models. Using the maximum likelihood method to estimate the coefficients in the models, followed by model validation and model selection criteria, it is concluded that EGARCH (1, 1) and TGARCH (1, 1) are the best models for eight of the currencies understudied. |
|---|