On the validity of the random walk hypothesis applied to the Dhaka stock exchange.

This paper employs a battery of statistical tests to examine the random walk variant of the weak-form efficient market hypothesis (EMH) using the daily data of the Dhaka Stock Exchange, the major equity market of Bangladesh, over a period of January 1990 to December 2000. The test results, however, are at variance across testing procedures and sub-periods. Results based on the random walk model and unit root tests show that the null hypothesis of randomness cannot be rejected and stock prices have a significant random walk or permanent component. Our analysis of autocorrelation functions indicates mean-reversion behavior of stock returns in most cases albeit with stock returns exhibiting some memory and predictable components during the bubble and post-speculation periods. The evaluation of the EGARCH-M model suggests significant asymmetric and leverage effects during the sub-period of speculative bubbles of 1996–1997. The BDS test indicates evidence of nonlinear long-term dependence during the pre-speculation period, while during the speculation and post-speculation periods the null hypothesis of nonlinear independence was not rejected. Overall, based on this evidence we do not categorically claim that the Dhaka Stock Exchange is weak-form efficient. However, these findings underscore the predictive significance and relevance of the random walk hypothesis as a generalized theory in explaining movements of share prices.