摘要
本研究以三種不同類別的金融市場（股價指數、匯率及個股股價）、六種金融商品為對象，應用六種時間序列的波動性估計模型（歷史標準差、指數加權移動平均、GARCH(1,1)-G、GARCH(1,1)-N、限制最小平方法RLS及一般化限制最小平方法G-RLS）作預測能力的比較，以探討不同市場是否適用不同模型、過去報酬衝擊權重設定的適當性及參數估計值與估計程序的關聯性，並以MAE、RMSE、HMAE及HRMSE四種方法來評估模型的預測績效。實證結果發現：歷史標準差與指數加權移動平均的預測能力並沒有明顯的差別；而GARCH(1,1)-G普遍較GARCH(1,1)-N模型的預測能力好。整體而言，在此六個金融商品中，波動性預測能力較佳的三個模型分別為G-RLS、RLS以及GARCH(1,1)-G。
並列摘要
In this paper, there are six financial commercial products in the three kinds of financial markets such as stock index, exchange rate and individual stock markets. We estimate the volatility of the three financial products by using the six time-series models, STD, EWMA, GARCH(1,1)-N, GARCH(1,1)-G, RLS, G-RLS and evaluate the forecasting ability of these six time-series models by using the four performance-evaluating criteria, MAE, RMSE, HMAE, and HRMSE in order to explore three main issues: First, whether the different models are applied to the different financial commercial products or not. Second, whether the weighting of the past returns are assigned correctly or not. Third, we explore the relationship between the values of the estimated parameters and the estimating procedure. From the empirical results, we can find that in the forecasting ability of the volatility it is not different for the STD and EWMA model, the GARCH(1,1)-G model is better than the GARCH(1,1)-N model, and as a whole the three better models are G-RLS, RLS and GARCH(1,1)-G models for the six financial commercial products.
參考文獻
延伸閱讀
- 鄭力維（2008）。波動的預測與評估〔碩士論文，國立暨南國際大學〕。華藝線上圖書館。https://doi.org/10.6837/NCNU.2008.00045
- 許能凱（2006）。金融商品波動性預測〔碩士論文，淡江大學〕。華藝線上圖書館。https://doi.org/10.6846/TKU.2006.00354
- Kao, C. W. (2008). 外匯市場價格發現、波動與中央銀行干預之分析 [doctoral dissertation, Tamkang University]. Airiti Library. https://doi.org/10.6846/TKU.2008.00412
- 張碩航（2015）。選擇權資訊加總對期貨市場波動率預測能力之影響〔碩士論文，淡江大學〕。華藝線上圖書館。https://doi.org/10.6846/TKU.2015.01054
- 張傳章、張森林、許博翔（2000）。Pricing Barrier Options Under Stochastic Volatility。中國財務學刊，8(3)，41-77。https://doi.org/10.6545/JoFS.2000.8(3).2