The efficient market hypothesis (EMH) has recently been under attack as a result of the methodology employed in the early tests of the theory. More recent evidence has shown that the traditional tests of random walks are susceptible to errors because of spurious autocorrelation induced by non-synchronous trading which is characteristic of stock markets in developing countries. This is the first study to apply the recently introduced Lo-MacKinlay variance ratio test to the Korean Stock Exchange (KSE). The KSE is chosen because it represents a typical emerging stock market. The results show that under the assumption of homoscedastic error term, the random walk hypothesis is rejected. However, with heteroscedastic stochastic disturbance term, we could not reject the hypothesis of random walk for daily data. When the test is applied to longer horizons such as weekly, monthly, 60-day and 90-day interval data, we could not reject.the random walk hypothesis. There is also evidence that suggests the presence of spurious autocorrelation and heteroscedasticity in the market. These could be due to official intervention and nonsynchronous trading in the market.
