Random Walks and Predictability in Foreign Exchange Rates

何佳(Jia He) ； 劉憶如(Christina Y. Liu)

《中國財務學刊》 1卷1期 (1993/07) Pp. 79-92

https://doi.org/10.6545/JoFS.1993.1(1).4

摘要

本文檢驗匯率之random work假說並測試每星期匯率變動之可預期性。這種實證研究可加強對匯率預測及匯率決定模型理論之了解。本文首先採用一個導自於He and Liu（1992）一文中之新的random walk之檢驗方法，joint variance-ratio test，對五種不同之世界主要外匯匯率進行測試。其次，採用一些與外匯風險相關之變數進行對過去20年來每星期匯率變動之預測模型的檢驗分析。相異於過去20年來大部份文獻中對匯率random walk假說之支持，本文之實證結果提出對匯率random walk假說之反證。這個反證加強了Liu and He（1991 a）一文中對匯率random walk假說所提出之新發現。另外，本文亦發現匯率之每週變動中包含著可預期的成份。整體而言，本文之實證結果不但認為匯率之變動並非純粹random而且建議其不為random之原因極可能源於外匯風險補償性之存在。

並列摘要

This paper reexamines the random walk hypothesis in foreign exchange rates and tests the predictability of weekly increment s in foreign currencies, yielding implications on the model of exchange rate forecasting and the theory of exchange rate determination. A joint variance-ratio test developed in He and Liu (1992) is first applied to five pairs of weekly foreign exchange rate series for testing the random walk. Then, the weekly increment in currencies is examined by adopting several risk-related financial instrumental variables as predictive variables. The test results in this paper reject the random walk in foreign exchange rates, which contrast with previous literature and yield additional support for the findings in Liu and He (1991a). Moreover, it is shown that the increment in foreign exchange rates contains a forecastable component. Overall, the results in the paper suggest that while increment s in foreign currencies are not IID (independently and identically distributed), the rejection of the IID is mostly due to the existence of risk-premia.