The Random Walk Hypothesis of the Emerging Stock Markets Revisited : A Comparison of Test Power of the Variance Ratio and Rescaled Range Models

Abstract

In this paper, we compare the result from the modified rescaled range model and the variance ratio model for the emerging markets. In addition, four different data generating processes are utilized to critically evaluate the robustness of the two models via Monte Carlo simulations. It is found that twelve of the fourteen markets exhibit nonrandom behaviors under the variance ratio model. On the other hand, the null hypothesis is rejected for only 3 emerging markets using the modified rescaled range model. Moreover, when the data follow a MA(1) process even with a small coefficient, the VR model tends to reject the null hypothesis while the R/S model does not. Key words: Modified Rescaled Range, Variance Ratio, Random Walk Hypothesis, Monte Carlo Simulation