- 學位論文
貴金屬指數型基金 (ETF) 評價: 以槓桿效應、外溢效果、動態波動和預測之檢測為例
An Evaluation of Precious Metal ETFs: Testing for Leverage Effect, Spillover Effect, Volatility Dynamics and Forecasting
摘要
此論文研究指數型基金，特別是針對貴金屬指數型基金為標的，檢測其槓桿效果、外溢效果、動態波動以及預測投資工具的干擾狀況。本論文透過三篇論文，針對自2005年後的熱門貴金屬ETFs交易進行研究。 首篇文章藉由ARFIMA-FIAPARCH的研究方法，提出在時間序列下，貴金屬ETFs交易波動的長期記憶特性。利用長期記憶特性的顯著證明分析應用到貴金屬及一般金屬ETFs的交易之中，比較正面與負面消息間的差異，正面資訊在非對稱性的波動間顯出較強且顯著的效果。此外，此篇文章應用GARCH-M-ARMA以及EGARCH-M-ARMA模型，以貴金屬(一般金屬)ETFs與貴金屬(一般金屬)價格指數間單向與雙向關係進行研究。研究結果指出貴金屬(一般金屬)ETFs報酬中，貴金屬(一般金屬)的價格報酬落後期存在顯著的正向效果，但反之亦然。 第二篇文章使用三個MGARCH模型之相關建構條件，並分析在貴金屬ETFs與期貨指數間的波動外溢效果及一致性分析。研究結果顯示BEKK模型是最佳配適資料的模型，並代表長期持續性的狀況;貴金屬ETFs 的波動衝擊可能因長時間範圍的期貨交易契約而受到衝擊。另研究結果指出跨產品的落後期以及共變數的落後期的衝擊是存在顯著性。因此，貴金屬ETFs報酬率的波動會影響到貴金屬期貨價格報酬。 第三篇文章應用渾沌效果、灰色關聯分析以及類神經網路方法，進行貴金屬報酬波動與貴金屬ETFs間的預測。使用BDS、R/S檢定法及相關維度分析，發現資料集合存在渾沌行為特性。此結果顯示所有的資料數列均呈現出有決定性的渾沌狀況。貴金屬及貴金屬ETFs表現出隨機過程及非線性的特徵。WTI對預測貴金屬及貴金屬ETFs有較佳的影響結果，且此投資商品不受到股票指數、匯率、CRB指數、波動指數、利率與期貨買賣比率影響。此外，BPN模型是四種類神經網路模型中(包括：BPN、RBP 、RNN、TDRNN)進行預測貴金屬及貴金屬ETFs最佳預測模型。在進行預測貴金屬時，所有變數組都比high-GRG 和low-GRG變數組更具有影響力。有鑑於此，藉由使用所有變數群及部分變數使用high-GRG組變數，在進行貴金屬ETFs的預測時，可以得到較佳的預測結果。因此，投資者及交易員在進行貴金屬及貴金屬ETFs操作時，可以透過前述七個決定因素的關連而獲利。
並列摘要
This dissertation investigates the leverage effect, spillover effect, volatility dynamic and forecasting for the innovation investment instrument, exchange-traded funds (ETFs), especially precious metal ETFs. The research is implemented based on actively precious metal ETFs through three essays with the inception date from 2005. The first essay provides evidence on long memory properties in volatilities of precious metal (base metal) ETFs by applying the Autoregressive Fractional Integrated Moving Average-Fractional Integrated Asymmetric Power Autoregressive Conditional Heteroskedasticity (ARFIMA-FIAPARCH). The strong evidence of long memory has been performed for both precious metal and base metal ETFs. The appearance of significant volatility asymmetry represents that positive news has stronger effect comparing to negative news. Moreover, this essay employs Generalized Autoregressive Conditional Heteroskedasticity-in-Mean-Autoregressive Moving Average (GARCH-M-ARMA) and the Exponentially Generalized Autoregressive Conditional Heteroskedasticity-in-Mean-Autoregressive Moving Average (EGARCH-M-ARMA) models to exploit the unilateral and bilateral positive relationship between precious metal (base metal) ETFs and precious metal (base metal) price indexes. Results indicate that the significant positive effects of lagged precious metal (base metal) price returns on current precious metal (base metal) ETF returns and vice versa. The second essay uses three Multivariate General Autogressive Conditional Heterokedasticity (MGARCH) models to model conditional correlations and analyzes the robust check the volatility spillovers between precious metal (base metal) ETFs and futures indexes. Baba, Engle, Kraft and Kroner (BEKK) model is recognized to fit data the best and represented the long-run persistence; the shocks on volatility of precious (base) metal ETFs might have impact on their futures contracts through range of a long time. The significance results exploit that the lagged covariances and lagged cross-products of the shocks are presented. Thus, the volatilities of precious metal (base metal) ETF returns have influenced on their futures price returns. The third essay applies chaos effect, grey relational analysis (GRA) and artificial neural network (ANN) to forecast the return volatility of precious metals and precious metal ETFs. The chaotic behavior is found in these data sets while using Brock Dechert Scheinkman (BDS) test, the rescaled range (R/S) analysis and correlation dimension analysis. The results showed that all the data series performed deterministic chaos. Precious metal and precious metal ETFs have represented random process and nonlinear properties. The West Texas Intermediate (WTI) index shows the greatest influence on forecasting precious metals and precious metal ETFs, followed by stock index, exchange rate, commodity research bureau (CRB) index, volatility index, interest rate and put-call (P/C) ratio. Moreover, the backpropagation network (BPN) model is the most powerful model among four ANN models like BPN, radial basic function (RBP), recurrent neural network (RNN) and time-delay recurrent neural network (TDRNN) in forecasting precious metals and precious metal ETFs. “All variables” group has stronger influence than high- or low-grey relational grade (GRG) variables for predicting precious metals. Whereas, some precious metal ETFs have better forecasting results by using “all variables” group and some others by utilizing high-GRG variables. Therefore, investors and traders can get profit through linkage seven determinants in forecasting precious metals and precious metal ETFs.