Long-term memory effect in stock prices: An empirical study from Nairobi stocks market.

Abstract

This study demonstrated that Nairobi stock market asset return do not behave as any type of white noise processes using the Lo and MacKinlay variance ratio test. This was done by considering the Nairobi All Share Index (NASI) and testing for long memory using Classical Rescaled range analysis, Detrended Fluctuation Analysis and the semi-parametric approach Geweke and Porter-Hudak tests. Data sets consisted of daily return index of NASI for a consecutive period of 8 years, i.e. from when the index was launched in 2008 to 2013 and long memory tests for the returns series. Al three tests suggested presence of long memory, while those of randomness test using variance ratio tests rejected the random walk hypothesis. The test for random walk model has a lot of implication in both theoretical as well empirical researches. Rejection of random walk model implies that the market is inefficient in processing information and one can predict future prices using past prices. Results show evidence of long memory in the Kenyan stock returns, which is inconsistent with weak-form market efficiency, implying that Kenyan stock index consists of impact of news and shocks in recent past. Speculative earnings could be gained via predicting stock prices. These findings will help investors, financial managers and regulators dealing with this market.