Long Memory in Stock Returns: An Analysis Using a Wavelet Based Semi-Parametric Estimator
The Empirical Economics Letters, 17(2): (February 2018), ISSN 1681 8997
10 Pages Posted: 1 May 2018
Date Written: February 1, 2018
The estimation and the analysis of long memory parameters have mainly focused on the analysis of long-range dependence in stock return volatility using traditional time and spectral domain estimators of long memory. The definitive ubiquity and existence of long memory in the volatility of stock returns is an established stylized fact. The presence of long memory requires major revisions in the standard estimation procedures without which the estimated results can be seriously biased. Therefore, a wavelet based semi-parametric estimator of long range dependence is applied to test for the presence of long memory in the Indian stock returns and returns volatility. We find the presence of long memory in the volatility of the stock returns as well as the returns themselves, when the analysis is performed using rolling windows. The presence of long-memory implies that distant observations in each of the volatility series are related to each other. This implication leads to the rejection of efficient markets as long range dependence in returns volatility seems to be incompatible with market efficiency.
Keywords: Wavelets, Hurst Estimator, Evolutionary Efficiency, Time-Varying Dependence
JEL Classification: C13, C14, C22, C32, G15
Suggested Citation: Suggested Citation