This study employs a new modified rescaled range (WS) test to examine the long memory behavior of stock returns in four foreign markets. Unlike the conventional FUS analysis, the modified WS analysis is robust to short-term dependence and conditional heteroskedasticity, that are shown to exist in stock return data. Empirical results show that the conventional and new modified RIS analyses can lead to entirely different conclusions about the presence of long cycles in foreign stock returns. While the conventional WS analysis seems to indicate the presence of long cycles, no significant evidence of long cycles can be found using the modified R/S analysis once short-term dependence and conditional heteroskedasticity in the data are accounted for. Implications of the findings are discussed.
Journal of International Financial Markets Institutions & Money, 3(1), 33-47