Long memory in international equity markets: revisited

Abstract

This study provides empirical evidence of the long-range behaviour in international equity markets. We test for long memory in the daily returns using the modified rescaled range statistic R/S proposed by Lo (1991) and the rescaled variance V/S statistic developed by Giraitis et al. (2003). Long memory is found to be weak in the return series when using R/S but some evidence of long memory is found in USA and Germany based on V/S analysis. Our results confirm those reported by Lo (1991) using only the rescaled range analysis and should be useful to regulators, practitioners and derivative market participants, whose success depends on the ability to forecast stock price movements.

Notes

¹The most interesting finding of the cited literature is that of Lo's (1991) study that did not detect long memory in USA market using his modified R/S analysis. We reexamine his finding using a new test statistic called the (V/S).

²Unconditional volatility is measured through absolute, squared and modified log squared returns. We applied the new testing methodology to these measures and all of them show evidence of long memory. We provide the results for squared returns but other results are available from the author upon request.

³Summary statistics and tests for stationarity are available from the author upon request.

⁴In other words, the 95% confidence interval with equal probabilities in both tails is [0.809, 1.862]. The fractiles are given in Lo (1991).

Table 1. R/S analysis applied to international equity market returns and squared returns

⁵Higher lag orders were used but the same results were obtained.

Table 2. V/S analysis applied to international equity market returns and squared returns