本文以Chan and Maheu (2002)所提出ARJI (Autoregressive Conditional Jump Intensity)模型探討台積電、聯電與日月光等三檔美國存託憑證(ADR)及其標的物之波動性,同時考量報酬率呈現非常態之特性,因此採用偏態t分配 (Skewed student t distribution) 進行實證分析。再者,本文進一步納入Hansen (1996)所提出之門檻自我迴歸模型(Threshold Autoregression model, TAR),估計ADR報酬率之門檻值,將樣本區分為高報酬與低報酬區間,並將其與波動性相對應,分析在高報酬區間之波動行為與低報酬區間之波動行為有何差異。最後,選取樣本期間內之重大事件,探討事件期間ADR與其現貨之特性。由ARJI模型估計結果顯示,不連續跳躍過程是影響報酬率不可忽視的重要現象,而當ADR與現貨報酬落入低報酬區間時,其平均跳躍頻率及機率都相對較大,表示異常負面消息的衝擊反應較為強烈,存在不對稱效果。在重大事件的衝擊影響方面,美國911事件所造成之衝擊最為強烈,其次為國內319 槍擊案,其餘事件下之反應則因各檔股票之差異而有不同。
The paper investigates the return and volatility in ADR and their underlying stocks using the ARJI model (Chan and Maheu, 2002). Three ADRs are examined in this study, including TSMC, UMC and ASX. Considering the non-normality features of these assets returns, this paper use skewed student t distribution in the empirical model. Further, the TAR model (Hansen, 1996) is used to distinguish the variance and jump intensity into high and low intervals by the threshold of ADR returns, for examining the difference between these two intervals. Moreover, we also analyze the characteristics of volatility of ADR and the underlying stocks during the influential events. The empirical results in ARJI model show that the jump cannot be ignored in estimating returns. The jump intensity is higher in low return intervals, indicating that the stronger responses of the returns of ADR and those of underlying stocks returns to abnormal the bad news. This suggests that the asymmetric effects exist. Finally, the greatest impact on the assets is the 319 shooting event, the next is the 911 event. However, no consistent pattern of the impacts of other events on the assets can be observed.