Are Stock Returns Long Term Dependent? Some Empirical Evidence
Posted: 28 Jun 1998
We show that investigation of the return series of indices of five European countries, the United States and Japan rejects the conclusion of long term dependence in these series. The statistic used to establish this result is the 'modified' rescaled range, suggested by Lo (1991). This statistic adjusts the 'classical' rescaled range (introduced by Hurst (1951)) for short term dependence. We also report additional results of a Monte Carlo simulation to determine the empirical size and finite sample distribution of these statistics when the data exhibit so-called volatility clustering. Our simulation evidence indicates that in the presence of short term dependence and volatility clustering the modified and classical rescaled range reject the null hypothesis of no long term dependence too frequently.
JEL Classification: C20, G10
Suggested Citation: Suggested Citation