匯率非線性均值回復之實證研究

The Empirical Study of Non-linear Mean Reversion–Evidence from Exchange Rates

殷君豪(Chun-Hao Yin)

指導教授 ： 楊奕農

中原大學/商學院/國際貿易研究所/碩士(2007年)

https://doi.org/10.6840/cycu200700783

摘要

匯率不論是在國際財務市場或國際貿易，皆扮演著舉足輕重的地位。隨著固定匯率制度的瓦解，匯率是否符合隨機漫步假說，成為長久以來研究的爭論議題之一。過去文獻利用平滑轉換自我迴歸模型，拒絕匯率隨機漫步假說的成立，認為匯率調整行為應呈現非線性均值回復，可能原因為交易成本的存在。但文獻中發現匯率常存在異質變異的情況，卻忽略或假設其無異質變異。理論上，有異質變異的情況，卻忽略會造成錯誤的模型估計與推論。此外，一般而言使用長時間的資料，除了均數方程式有結構轉變的現象，變異數方程式亦可能有結構轉變現象。因此，本文先用單根檢定、變異數比率檢定、非線性單根檢定，來檢定匯率資料是否服從隨機漫步假說。資料以美元為本國選澳幣、加幣、歐元、日圓、新台幣與英鎊等六種貨幣，後分別利用：(1)均數與變異數方程式皆線性、(2)只考慮均數方程式且存在結構轉變(與先前文獻模型相同)、(3)均數方程式為線性且變異數方程式存在結構轉變、(4)均數方程式存在結構轉變與變異數方程式為線性、(5)均數與變異數方程式均存在結構轉變等五種模型，來估計匯率資料。其結果發現，假若只考慮均數方程式存在結構轉變，不加入異質變異的情況，則匯率資料呈現非線性均值回復之現象；若同時考慮均數與變異數方程式皆存在結構轉變，則匯率調整行為會拒絕非線性均值回復之假設，服從隨機漫步假說。

關鍵字

均值回復 ； 隨機漫步假說 ； 非線性GARCH模型

並列摘要

A continuous debate has been conducted in many empirical results concerning stabilities of the exchange rates. In addition, the exchange rates play one of the most prominent roles not only in international financial markets, but also in international trade. The literature rejects random walk hypothesis in the exchange rates by non-linear mean reversion models. That also offers some reasonable explanations (ex: transition cost). However, the previous papers do not consider that the variance equation may be time-varying, not constant. Even there may be structural changes in the variance equation as well. If we ignore previous situation, the empirical results will be probably misleading. It suggests that the time-varying variance and regime-switching should be considered in the models.
In this paper, first we apply the tests, as variance ratio test and non-linear unit test, to test random walk hypothesis of the exchange rates are hold or not. We separately estimate five kinds of these models, such as AR-GARCH (1,1), ANST-GARCH, STARD, STARD-GARCH (1,1) and STARD-ANST-GARCH. However, there are inconsistent estimated results with the past empirical results which are supported non-linear mean reversion. The random walk hypothesis can not be rejected while thinking non-linear patterns in the variance equation model as well.

並列關鍵字

Mean reversion ； Non-linear GARCH model ； Random walk hypothesis

參考文獻

Foreign Exchange Market Efficient?”Journal of Banking and Finance,20,687-

Ajayi, R. and Karemera, D. (1996) “A Variance Ratio Test of Random Walks in

Exchange Rates: Evidence from Pacific Basin Economies,” Pacific-Basin

Finance Journal, 4, 77-91.

Series Analysis of Economic and Finance Data. Kluwer Academic