| 摘要: | 本論文利用EBS交易系統中，交易最熱絡的歐元兌美元和美元對日元兩種貨幣之報價和成交資料研究外匯市場不同時區價格發現的過程與其影響因子。我們將24小時交易切割成四個子市場，亞洲、歐洲、倫敦紐約重疊交易時段和美國，進而比較四個子市場的價格發現能力。
本論文的第一個研究主題探討總體經濟宣告如何影響價格發現過程。我們發現倫敦紐約重疊交易時段的價格發現能力只有在當天有美國的經濟數據宣告時，才會顯著最大。但是，歐洲或日本當地的總體宣告資料卻沒有大幅提升歐洲或亞洲對於歐元兌美元或是美元兌日圓各別的價格發現能力。
第二個主題我們透過修正第一個研究主題的方法更進一步去探討總體經濟宣告與價格波動和價格發現的關係。我們將總和的價格波動度拆解成連續和不連續兩個部分。在拆解後，本論文利用連續與非連續的兩個部分各別去估算四個子市場的價格發現能力。就連續的波動度所估算的訊息份額，我們發現倫敦紐約重疊交易時段都會有顯著較大的價格發現能力。但是就非連續波動度所估計出來的結果與第一個研究主題結果一致，倫敦紐約重疊時段的價格發現能力只會在美國有宣告的時候才顯著強過其他三個子市場。但是我們更進一步發現，當歐洲和日本有宣告時，亞洲的價格發現能力會大幅上升，並且顯著強過其他三個市場。所以拆解的訊息份額更能幫助我們了解價格傳遞的過程。並且在我們的模型分析中發現，非連續的部分對於總體宣告相對敏感，更能幫助我們理解價格波動與總體宣告間的關係。
第三個主題利用資訊優勢交易的比例來探討價格發現與總體宣告的關係。本研究利用Hascrouck (1991) 提出的方法計算訊息份額。我們發現不管有無宣告，倫敦紐約重疊時段對於整天的資訊傳遞貢獻都是最大的。此結果反應出在倫敦紐約重疊交易時段，會有較多的資訊優勢交易者。此一發現，提供我們一個針對為什麼倫敦紐約重疊交易時段會價格發現相對強的解釋。
;This dissertation studies the price discovery function across different trading periods in the foreign exchange markets and determinants for price discovery ability by quoted price and trades. We divide the 24 hours into four trading periods, including the Asian, European, European–U.S. overlapping, and U.S. markets. The data this dissertation used are from the Electronic Broking Services (EBS), and we employ the most traded currencies, EUR/USD and USD/JPY, to examine our hypothesis.
In the first essay, we focus on how macroeconomic announcements affect daily price discovery patterns. We find that the trading in the overlapping trading hours of London and New York only dominates price discovery in currency trading on days with U.S. announcements releases. But the local macro news announcements in Europe and Tokyo do not make Asian and European periods lead the price discovery in EUR/USD and USD/JPY markets, respectively.
In the second essay, we employ a new concept of volatility in pricing to discuss the intra-day price discovery patterns. Since the first essay find that European and Japanese macro-news does not significantly increase the information shares of European and Asian markets and recent literature suggests that discontinuous volatility component contains more or different information compared with continuous component, this paper therefore decomposes the price volatility into continuous and discontinuous components and separately employs both components to calculate information shares. Based on discontinuous volatility, this chapter presents that Asian market dominates price discovery in the USD/JPY and EUR/USD on days with Japanese and European news announcements, respectively. Moreover, the results further confirm that discontinuous volatility is more sensitive to macroeconomic announcements and unscheduled surprises. Hence, this study suggests that it would be better to separately discuss the news incorporation with continuous and jump components in price volatility.
Finally, to further study how the price discovery process is related with informed trading, we analyze changes in information content of order flow among four trading periods. We use the methodology developed by Hasbrouck (1991) to calculate the information content and show trades taken place in the overlapping trading hours of London and New York are the most informative than those in the other three trading periods. However, the releases of macro announcement actually do not alter the intraday pattern of trading informativeness across different trading periods, and this finding is even hold on days with European and Japanese announcements for the EUR/USD and USD/JPY, respectively. Overall, the findings in this essay are consistent with information shares computed with continuous price volatilities. |
