金融商品報酬率之類別動態相關係數探討

A Study on the Flexible Dynamic Conditional Correlation of Financial Products

張瑄(Hsuan Chang)

指導教授 ： 李孟峰

國立臺北大學/商學院/統計學系/碩士(2010年)

摘要

近年來國人投資理財的金融商品已逐漸多樣化，但其報酬率常讓人難以掌握，因此如何在金融市場中獲得高報酬率是許多投資人共同的目標。此外，在追求高報酬的同時，潛藏的風險亦隨之增加，因而懂得如何規避風險亦成為投資人不可缺少的工具。倘若投資人能瞭解各金融商品報酬率間的相關性，並利用其性質預測報酬率做定額投資的資產配置分散風險，方可助投資人於系統風險下進行適當的避險操作，進而獲得最佳利潤或減輕風險性。
早期的研究大多採用固定的相關係數探討金融商品或不同市場間的關聯程度，但這樣的方式與實際情況並不相符，因為不論商品或是市場間的相關性都常跟隨時間發生變動，因此本研究將採用Monica Billio、Massimiliano Caporin and Michele Gobbo (2006) 所發表的FDCC模型方法來探討台灣股市中同類型與不同類型股票報酬率的動態相關，並希望可將此研究結果運用至風險規避的策略上。而此模型主要是簡化 DCC模型的參數估計個數，並保留原有的參數限制條件。
研究結果發現：一、採用動態相關係數來衡量個股報酬率的相關程度比固定相關係數更能掌握彼此的變化情況。二、同類型個股間的動態相關係數相較於不同類股間的相關係數高。三、股市遭逢重大事件後，個股間的相關係數會發生明顯變化，相關性具逐步升高的現象，其中同類股間的衝擊比不同類股間的衝擊大。四、不同事件對個股間關聯性的影響時間長短不盡相同。

關鍵字

動態相關、相關性、FDCC、DCC

並列摘要

Recently there are variety of financial products available for people in Taiwan to manage their property. Naturally, pursuing high return definitely has been the mutual objective for all investors; however, we all know that the return is always difficult to be controlled well and high return usually accompany high risk as well. Therefore, risk hedging passes has gradually become an important issue and instrument investors would not miss. If investors could know the relation to the returns of different financial products and make use of its character to predict returns and do dollar-cost averaging assets allocation; it may help them to obtain optimal profits or reduce the possible risks under a suitable hedging operation.
The previous researches were adopted constant correlation coefficients to probe the relative level among the returns of financial products or different markets, but this method wasn’t consistent with the fact that the correlation coefficients of products or markets always change continuously with the time. Hence, this research applied the Flexible Dynamic Conditional Correlation (FDCC) model, proposed by Monica et al. (2006) and designed to reduce the number of parameters and reserve the limitation of parameters of the Dynamic Conditional Correlation (DCC) model, to investigate the dynamic correlation of stocks within the same sector or between different sectors of Taiwan stock market and expect the results of this research could be applied to the hedging strategy.
There are some findings of empirical study, including firstly evaluate the relevant change among individual stock return by adopting the dynamic conditional correlation model would be more suitable than the constant conditional correlation model. Secondly, the dynamic correlation coefficient of individual stocks in the same sector would be higher than that in different sectors. Thirdly, when stock market was shocked by serious incidents, the correlation coefficient among individual stocks would increase progressively and the impact particularly on individual stocks in the same sector would be greater than that in different sectors. Finally, the correlation coefficient among individual stocks was differently shocked by the dissimilar incidents on the basis of the different length of time.

並列關鍵字

Dynamic Correlation、Correlation、FDCC、DCC

參考文獻

6.王冠閔、黃柏農 (2004)，台灣股、匯市與美國股市關聯性探討，中央研究院經濟研究所臺灣經濟預測與政策34:2 (2004)，31-72。

1.李欣儒 (2009)，以ADCC模型探討衝擊事件對亞洲主要股市間動態相關的影響，國立臺北大學統計研究所碩士論文。

2.曾菀瑩 (2008)，台灣上市股票類股報酬率動態相關之探討，國立臺北大學統計研究所碩士論文。

5.黃小菁 (2005) ，DCC多變量GARCH模型之風險值計算-G7及臺灣等八國股市投資組合之實證研究，淡江大學財務金融研究所碩士論文。

1.Akaike, H. (1974). Information theory and an extension of the maximum likelihood principle. 2nd International Symposium on Information Theory, 267-281.