請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/102636| 標題: | XCCY Swap市場基差與理論基差共整合分析 Cointegration Analysis of Market and Implied XCCY Basis in XCCY Swap |
| 作者: | 朱志升 Chih-Sheng Chu |
| 指導教授: | 李賢源 Shyan-Yuan Le |
| 關鍵字: | XCCY Swap; 換匯換利基差交換; 共整合; 向量誤差修正模型; Granger因果關係 XCCY Swap; Cross-Currency Basis Swap; Cointegration; Vector Error Correction Model; Granger Causality |
| 出版年 : | 2026 |
| 學位: | 碩士 |
| 摘要: | 本研究探討市場觀察到的XCCY Swap基差與拋補利率平價(CIP)理論推導下的隱含XCCY Swap基差之間的動態關係。本文以EURUSD與USDJPY兩組5年期XCCY Swap為例,利用Bloomberg資料進行2012年至2024年的週資料實證分析。
首先檢定各變數定態性,結果顯示四組基差序列皆為I(1)過程,具備進行共整合分析的前提。接著透過Engle–Granger與Johansen方法檢定,確認Market與Implied XCCY Basis存在穩定的共整合關係。向量誤差修正模型(VECM)進一步顯示,Implied XCCY Basis會主動調整以收斂至Market XCCY Basis,而非反之,Granger因果檢定亦證實Market XCCY Basis具領先性。基於此結果,本文構建以誤差修正項(Error Correction Term, ECT)為核心的預測模型,並檢驗其對Market XCCY Basis未來變動方向之預測能力。雖然當期預測方向準確率約五成,與Naïve動能預測模型相當,但若擴大觀察至12期內預測正確情形,ECT模型的預測成功率可達83%以上,優於基準模型。在樣本外驗證下,ECT模型預測表現相較Naïve模型更加穩健。 本研究主要貢獻在於實證分析不同幣別下Market與Implied XCCY Basis間的共整合關係,補充既有文獻對CIP偏離現象的探討,過去文獻多聚焦於探討影響XCCY Basis的因子,較少深入探討理論隱含基差與市場交易基差之間的長期動態關係。另外,本研究嘗試找出預測Market XCCY Basis未來趨勢的領先指標,也是既有文獻鮮少探究的議題。 This study investigates the dynamic relationship between observed XCCY basis and the implied basis derived under the Covered Interest Rate Parity (CIP) framework. Focusing on two major currency pairs—EURUSD and USDJPY—this paper conducts an empirical analysis using weekly data from Bloomberg spanning the period from 2012 to 2024. Unit root tests confirm that all basis series are integrated of order one (I(1)), satisfying the conditions for cointegration analysis. The Engle–Granger and Johansen tests reveal a stable long-run cointegration relationship between the Market and Implied XCCY Basis. Vector Error Correction Model (VECM) further indicates that the Implied XCCY Basis adjusts to deviations from equilibrium, which is indicated by cointegration, while the Market XCCY Basis demonstrates leading behavior, as supported by Granger causality tests. Based on these findings, an Error Correction Term (ECT)-based predictive model is constructed to forecast the directional movement of the Market XCCY Basis. While the contemporaneous prediction accuracy is around 50%—comparable to a naïve momentum model—the ECT model achieves over 83% accuracy when extending the observation window to 12 periods, and outperforms the benchmark model in out-of-sample evaluations with greater robustness. The main contribution of this study lies in providing empirical evidence of cointegration between Market and Implied XCCY Basis across different currency pairs, complementing the existing literature on persistent deviations from CIP. Prior research has mainly focused on identifying determinants of XCCY Basis, while few have explored the long-run dynamics between theoretical implied and market-based swap spreads. In addition, this study proposes a forward-looking indicator based on ECT for predicting the future trend of Market XCCY Basis, addressing a gap rarely explored in the literature. |
| URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/102636 |
| DOI: | 10.6342/NTU202601432 |
| 全文授權: | 未授權 |
| 電子全文公開日期: | N/A |
| 顯示於系所單位: | 財務金融學系 |
文件中的檔案:
| 檔案 | 大小 | 格式 | |
|---|---|---|---|
| ntu-114-2.pdf 未授權公開取用 | 2.02 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。
