| 摘要: | 自1990年代以來，區域經濟整合已蔚為全球經濟發展的主流，為促進香港與中國經貿投資更緊密結合的「更緊密經濟夥伴關係安排」(Closer Economic Partnership Arrangement, CEPA)，於2003年6月29日簽訂且於2004年1月1日生效後，是否會造成香港與中國股市報酬率關聯性發生結構改變，一直是大家討論的議題。本文以條件copula方法探討CEPA對香港與中國股市報酬率關聯性結構改變時點與關聯性平均改變水準。
本文以三個面向探討CEPA對香港與中國股市報酬率關聯性結構改變的影響。首先，本文以CEPA生效日為關聯性結構改變時點，藉 為邊際模型，利用Gaussian、Gumbel、Clayton 及 copula探討兩股市報酬率的關聯性結構改變。發現僅Gumbel copula呈現顯著，意謂著兩股市在CEPA生效後，僅同漲時的關聯性有顯著的增加。第二面向假設關聯性結構改變時點為未知且 的邊際模型具有未知波動性結構改變，利用 值選定 copula為最適copula後，再以最大概似比檢定搜尋未知的關聯性結構改變時點。發現香港與中國股市關聯性結構改變時點為2005年2月2日。結構改變時點後關聯性平均水準增加27.21%。最後，考量邊際極端波動性對關聯性結構改變時點的影響，透過 邊際模型，將非尋常消息衝擊所引起的跳躍成份自全部報酬率衝擊中排除，僅以尋常消息衝擊配適所選定的 copula。發現在排除邊際的極端波動後，兩股市的關聯性結構改變時點為2005年7月19日。結構改變時點後關聯性平均水準增加28.23%。
由於關聯性的增加使得兩股市的風險趨向一致，不利於國際投資人的風險分散。本文用以探討關聯性結構改變的方法與結果，可供國際投資人在資產配置及政府部門在制定相關政策上的重要參考。
Since the 1990s, economic integration has become widespread in the world economy. To strengthen the economic relationship between Hong Kong and China, the two economies signed the Closer Economic Partnership Arrangement (CEPA) on June 29, 2003, and the arrangement became effective on January 1, 2004. Whether the CEPA changed the dependence structure between the Hong Kong and Chinese stock markets has become a popular issue in recent years. This dissertation employs conditional copula to investigate this issue, and searches for the dependence structure change point and dependence average change level between the Hong Kong and Chinese stock markets.
This dissertation explores the dependence structure change between the Hong Kong and Chinese stock markets in three aspects. First, it assumes the date in which the CEPA became effective as a known dependence structure change point. By using as a marginal distribution and using the Gaussian, Gumbel, Clayton and copulas as dependence structure models, this dissertation finds that only the Gumbel copula is significant, which means that only when the prices of the Hong Kong and Chinese stock markets are both rising, does the dependence significantly increase. Second, this dissertation assumes that the dependence structure change points are unknown, and furthermore, considers that there are unknown volatility structure change points in a marginal distribution. After choosing the copula as the best-fitting copula by criteria and using the supremum likelihood ratio test to search for unknown dependence structure change points, this dissertation finds that the dependence structure change point is February 2, 2005, and that the average dependence level increased by 27.21% after the structure change point. Finally, this dissertation further considers the influence of marginal extreme volatilities on the dependence structure change. To avoid the influences of marginal extreme volatilities, this dissertation employs as a marginal distribution. After excluding the jump components, which are caused by unusual news innovation, from the total innovation, this dissertation uses normal news innovation to fit the copula and finds that the dependence structure change point is July 19, 2005, and that the average dependence level increased by 28.23% after the structure change point.
The consistency of the dependence between the Hong Kong and Chinese stock markets undermines
international investors’ efforts in risk diversification. The results and methods in this dissertation can provide valuable implications for international investors in asset reallocation and for governments in decision making. |
