摘要
本研究運用38國自1981年第1季到2017年4季的追蹤資料(panel data)，分別建立OECD19國與非OECD19國之動態門檻追蹤資料模型(dynamic threshold panel data model, DTPDM)，以驗證不同特性之國家、通膨以及經濟成長是否會影響Fisher 假說與Fama代理假說的成立。由實證結果發現，雙體制DTPDM模型為最適模型，在OECD19國的Model1A 與2A，當門檻變數小(等)於門檻值時 (本文簡稱體制1)時，Fisher假說是成立的，但在另一體制內則是不成立的。而非OECD19國的結果恰好與OCDE 19國的結果顛倒。至於 Fama 代理假說；當門檻變數大於門檻值時(本文簡稱體制2)，在OECD 19國的Model 1B與 2B，Fama代理假說是成立的。相反地，在非OECD19國的Model 1B與 2B體制1之內，Fama代理假說是成立。最後，在非OECD19國Model 3B的體制1內，Fama代理假說無法成立。綜合兩群體國家的實證結果發現，變數之間存在不對稱的門檻效果，而Fisher假說與Fama代理假說成立與否，的確受到不同的經濟成長體制、不同特性的通膨以及不同國家特性的影響。
關鍵詞：Fisher假說、Fama代理假說、通貨膨脹、股票報酬、動態門檻追蹤資料模型Abstract
This study constructed a dynamic threshold panel data model (DTPDM) to examine whether country specific characteristics, inflation, and economic growth would affect the validation of Fisher hypothesis and that of Fame proxy hypothesis. This study explored the empirical data of 38 countries, including 19 OECD countries and 19 non-OECD countries, dating from the first quarter of 1981 to the fourth quarter of 2017. The empirical results showed that a DTPDM with two regimes is the optimal model. Fisher hypothesis has proved to be valid for OECD 19 countries in Model 1A and Model 2A when the threshold variable is smaller than (or equal to) the threshold value--that is regime 1 in this study. The same hypothesis does not hold true in the other regime. For non-OECD 19 countries, the hypothesis is just opposite to that of OECD. Moreover, Fama proxy hypothesis is valid for OECD 19 countries in Model 1B and Model 2B when the threshold variable is larger than the threshold value—that is regime 2 in this study. On the contrary, Fama hypothesis for non-OECD 19 countries is valid in regime 1. Finally, Fama hypothesis for non-OECD 19 countries in regime 1 of Model 3B is not valid. The total opposite results of these two samples imply that when there is asymmetric threshold effect among variables, the validation of Fisher hypothesis and that of Fame proxy hypothesis depend not only on economic growth regime and inflation, but also on country specific characteristics.
Keywords: Fisher Hypothesis, Fama Proxy Hypothesis, Stock Return, Inflation, Dynamic Threshold Panel Data Model