應用多變量的動態變幅波動模型於兩岸三地期貨避險

Apply Range-Based Multivariate Volatility Models in the Hedge of TAIEX Futures, HSI Futures and CSI 300 index Futures

許綵羚(Tsai-Ling Hsu)

指導教授 ： 李沃牆

淡江大學/商管學院/財務金融學系碩士班/碩士(2015年)

https://doi.org/10.6846/TKU.2015.00079

摘要

隨著國際金融的發展及期貨避險被廣泛運用，使得投資人和金融機構可透過交易指數期貨，進而規避系統性風險。再者，由於近年來中國資本市場的開放，金融交易活動漸趨國際化與自由化，例如2014年的滬港通，顯示兩岸三地之間的金融活動也更加緊密。
從以往的文獻中可看出大多數的研究皆集中在GARCH族模型之比較，而且較少文獻是以兩岸三地為研究對象，因此，本文探討兩岸三地指數期貨之避險績效，採用以報酬率為基礎的模型如CCC-GJR-GARCH、DCC-GJR-GARCH及Copula-based GJR-GARCH模型，以波動變幅為基礎的模型有DCC-CARR及Copula-based CARR模型。研究CARR族模型是否比GARCH族模型更能捕捉資料的波動，以及加入Copula函數後，是否更能配適資料間的相關性。
實證結果顯示，不論在樣本內外，台灣和中國皆以Copula-based CARR模型表現較優，台灣市場的結果甚至發現加入動態Copula函數的CARR模型表現最佳。香港的樣本內避險雖以Copula-based CARR模型績效最好，但樣本外卻以Copula-based GARCH模型表現較優，同樣也是加入動態Copula函數後的GARCH模型表現較靜態和傳統模型好。

關鍵字

避險績效 ； GJR-GARCH ； CARR ； Copula ； 最小變異避險比率

並列摘要

With the development of the international finance and a wide use on hedging strategies using futures, investors and financial institutions can avoid systemic risks by index futures trading. Moreover, due to the opening up of Chinese capital markets in recent years, financial transactions become more international and liberalized. For instance, the Shanghai-Hong Kong Stock Connect scheme in 2014 demonstrates a closer financing activities among Taiwan, Hong Kong and China become more closely.
In early studies, most researchers concentrated on the comparison in GARCH family models, but few placed their interests in the markets, including Taiwan, Hong Kong and China. Therefore, this study aimed to investigate the hedge performance of TAIEX futures, HSI futures and CSI 300 index futures. Apart from investigating the hedge fund performance in Greater China area, this research tried to discover whether CARR family models have better ability to catch price volatility than of GARCH family models, and to figure out whether the correlation between spots and futures become fitter after adding Copula functions. In this study, some return-based GARCH family models, namely CCC-GJR-GARCH, DCC-GJR-GARCH and Copula-based GARCH models, and some ranged-based CARR family models, including DCC-CARR and Copula-based CARR models.
The empirical application of the models proposed in this study shows that both Taiwan and China perform better, no matter in-sample or out-of-sample in Copula-based CARR models. Furthermore, compared to other markets, Taiwan performed the best in the Copula-based CARR models with dynamic copula functions. Also, the result shows Hong Kong performs the best in in-sample tests in Copula-based CARR models. However, it has a better performance in Copula-based GARCH models on out-of-sample tests. In the same way, Hong Kong performs better than both static and traditional models after adding the dynamic copula functions.

並列關鍵字

Hedge performance ； GJR-GARCH ； CARR ； Copula ； Minimum variance hedging ratio

參考文獻

6.李亦屏 (2004)，黃金期貨之避險分析，中原大學企業管理學系碩士論文。

9.柯星妤 (2013)，金磚五國之期貨避險績效─應用Copula-based GJR-GARCH模型，淡江大學財務金融學系碩士論文。

19.黃薇之 (2010)，重新評估DCC-GARCH及DCC-CARR模型之避險績效，淡江大學財務金融學系碩士在職專班碩士論文。

5.巫春洲、劉炳麟、楊奕農 (2009)，農產品期貨動態避險策略的評價，農業與經濟，第42卷，頁39-62。

3.吳俞蓉 (2012)，個股期貨契約避險績效之研究，國立中正大學財務金融學系碩士論文。