亞洲匯率指數與指數型基金之預測分析-以ARFIMA-FIAPARCH模型為例

The Forecast of Asian Currency Exchange Index and Exchange-Traded Funds: The Analysis of ARFIMA-FIAPARCH Models

高秀貞(Hsiu-Chen Kao)

指導教授 ： 陳若暉

中原大學/商學院/企業管理研究所/碩士(2016年)

https://doi.org/10.6840/cycu201600445

摘要

透過全球性重大金融事件之影響性，匯率產生大幅度波動，使整體金融環境動盪不安。尤其1997年受到亞洲金融風暴襲捲，亞洲各國皆受到重挫，匯率的不穩定性漸而得到重視。本研究參考歐洲貨幣籃，歐元建構整合方式及其中心匯率成功案例，將台灣、中國、香港之淨外匯存底、每人生產淨額及出口貿易值等三種變數進行加權，建構出1992年3月至2016年2月之大中華單一貨幣化(CCU)的中心匯率。再將建構出之中心匯率、彭博-摩根大通以亞洲十種貨幣建構出亞洲貨幣指數（ADXY index）以及iPath GEMS亞洲八大國匯率ETN(AYT)、WisdomTree人民幣策略主動型ETF(CYB)及台灣、香港、中國實質有效匯率等相關變數，探討亞洲及大中華匯率變數之間的波動關聯性。
本研究利用RiskMetrics、ARFIMA- FIAPARCH、ARFIMA- FIGARCH模型實證，結果顯示大中華單一貨幣(CCU)經由SDR修正過後，其模型穩定性及預測能力優於未修正的CCU。CCU(SDR)修正後匯率波動穩定且預測力佳，相當適用於未來兩岸三地制定貨幣政策之金融工具。另發現台灣、中國實質有效匯率與CYB ETF確實存在非對稱性且對於市場波動性具有負面影響。重大金融事件的衝擊對波動性產生動盪，且前期報酬率的結果也會對當期產生負面影響。ADXY Index則發現具有緩長記憶特性，且穩定性高，對市場波動性呈現正向影響。藉由匯率波動觀察，在未來發生類似全球金融危機時，穩定亞洲及大中華市場匯率變動，則政策制定者可參考各變數在模型上的實證分析及匯率前後期之間的關連性，及早提出適當對應政策方可為國內金融市場降低傷害並迅速止血。
在放空及做多方面，本研究取用ADXY指數及AYT ETN，並採用RiskMetrics、ARFIMA-FIGARCH、FIAPARCH等模型，針對風險值配適度進行分析，分配設定為常態分配、學生t分配、偏態t分配作比較分析。結果顯示模型配適度與預測精確性最佳為FIGARCH，次佳為FIAPARCH模型。整體分配設定下，以偏態t分配設定在模型迴歸分析下優於常態分配。除了CCU與香港實質匯率之外，將變數投入常態分配設定下，則FIGARCH、FIAPARCH兩模型的結果類似，而RiskMetrics模型表現較為優異。

關鍵字

大中華單一化貨幣 ； ARFIMA ； RiskMetrics ； ARFIMA-FIGARCH ； ARFIMA- FIAPARCH

並列摘要

The fluctuating exchange rate is violent through the influence at financial crises. This had been impelled the finacial environment during the 1997 Asia financial crisis. As Asia economies particularly hard hit by the crisis, the unstable exchange rate has been much attention. This study refers to the structure of EURO currency basket, and constructs the Chinese Currency Unit（CCU）using the weights based on GDP per capita, export and net reserves of China, Taiwan and Hong Kong to measure Chinese Currency Unification’s central rate from 1992/3 to 2016/2. This paper valuates the exchange rate fluctuation related to Asia and Greater China region among variables including CCU, Bloomberg JPMorgan Asia Dollar Index, iPath GEMS Asia 8 ETN(AYT) , WisdomTree Chinese Yuan Strategy Fund (CYB ETF), China REER, Taiwan REER, and Hong Kong REER.
This study uses RiskMetrics、ARFIMA- FIAPARCH、ARFIMA- FIGARCH Models for empirical analysis, and the results show that the modified CCU(SDR) has CCU stability and better forecasting than regular CCU. Thus, this is suitable for Greater China region to use CCU(SDR) making monetary policy in future. The results also reveal that the real effective exchange rate of Taiwan and China and the CYB ETF appear the asymmetry of market volatility for having a negative impact on the volatility. The major financial turmoils impact on changing volatility, and the previous returns will have a negative impact on current return. ADXY Index is found to have a long memory characteristic and high stability, and it has positive effect on market volatility. As long as using the historical exchange rate fluctuations properly, policy makers can observe through the influences of various variables and exchange rate of earlier and later periods to stable the changes of currencies in Asia and the Greater China region. The adoption of suitable polices can be reached in advance in response of the shock of the global financial crisis.
This study uses ADXY index and AYT ETN based on the measurements of RiskMetrics, ARFIMA-FIGARCH, FIAPARCH models. From the prospectives of short position and long position to analyze the goodness-of-fit model for Value at Risk (VaR), the comparison of the distribution analysis includes Normal distribution, Student-t distribution, and Skewed- Student- t distribution. That results show that the best goodness-of-fit model and the accuracy of forecasting performance is FIGARCH model and next-best is FIAPARCH model. Among overall distribution, Skewed- Student-t distribution in regression analysis is better than Normal distribution. In addition to CCU and Hong Kong’s real exchange rate variables setting into the Normal distribution, the results of FIGARCH and FIAPARCH models are similar, while the performance of RiskMetrics model is excellent.

並列關鍵字

China Currency Unit ； ARFIMA ； RiskMetrics ； ARFIMA-FIGARCH ； ARFIMA- FIAPARCH

參考文獻

17. 許士峰(2009)，中華單一貨幣與金融危機之關連分析，中原大學企業管理學系碩士論文。

15. 楊仁吉(2008)，亞洲單一貨幣化與短期資本流動之關聯性研究，中原大學企業管理學系研究所碩士學位論文。

13. 夏南新(2007)，「基於 ARFIMA-APARCH 模型的中國黑市匯率－非對稱性和槓桿效應研究」，管理科學與統計決策，第4卷，第一期，18-25頁。

4. 杜玉振、魏郁珣(2009)，「多空交易部位之最適風險值模型研究 － 以股票、商品及外匯市場為例」，創新與管理，第七卷第一期，1-28頁。

1. 王啟明、陳怡君、蕭倩琳(2009)，「歐洲與亞洲共同貨幣政策之發展: 從比較區域研究角度分析」，全球政治評論，第二十五期，117-150頁。