On tests for long memory in Pacific Basin stock returns

There has been growing interest in studying the behaviour of stock returns over long and short horizons. Previous studies showed that while autocorrelations in returns usually are positive or close to zero over short horizons, they become negative over long horizons. In this article, we examine the behaviour of stock returns among the four Pacific Basin markets. Three tests which are robust to short-term dependence and conditional heteroskdasticity are employed. They are the modified rescaled range test, the fractional differencing tests based on raw periodograms and on the Bartlett window-smoothed periodograms, respectively. The empirical findings in general provide little support for long memory in Pacific Basin stock returns.