Variance Ratio Test of Random Walk in Indian Stock Market

Abstract

We conduct tests of the null hypothesis of a random walk at the aggregate level of market indices and disaggregate level of individual shares to the Indian stock market over various data periods and a comparison of two sub-periods namely the pre liberalization and the post liberalization period. For this, we use the Lo-MacKinlay (1988) variance ratio test. Although the oldest test i.e. the serial correlation coefficient test is also applied to the same data to establish the relationship between the two tests but its results are not elaborated in this paper. The strength of this paper lies in the voluminous data base and a powerful testing tool that it makes use of. It is observed that the market is highly inefficient at daily returns level, thus imbibing high degree of predictability in stock returns, and even the weekly returns show the existence of trend. Monthly returns, however, support the random walk hypothesis across all periods. Thus it is concluded that further refinement of reform measures is required.