It seems that the random-walk property of stock prices is well establi
shed. However, some studies argue that the mean reversion of the stock
prices has its theoretical and empirical support and the conventional
unit-root tests have weak power against stationary alternatives. This
paper uses unit-root tests in panel data to re-examine the time-serie
s properties of the stock prices as it is claimed that the method can
increase the power of the tests substantially even with a small number
of cross sections. The test result suggests that we cannot reject the
random-walk hypothesis for G-7 country stock-price indices.