Seasonality in Stock Price Mean Reversion: Evidence from the U.S. and the U.K.

TLDR: The evidence of slowly mean-reverting components in stock prices has been controversial. as mentioned in this paper used a regression model that yields the highest asymptotic power among a class of regression tests.

Abstract: The evidence of slowly mean-reverting components in stock prices has been controversial. The hypothesis of stock price mean-reversion is tested using a regression model that yields the highest asymptotic power among a class of regression tests. Although the evidence that the equally weighted index of stocks exhibits meanreversion is significant in the period 1926-1988, this phenomenon is entirely concentrated in January. In the post-war period both the equally weighted and the value-weighted indices exhibit seasonal mean-reversion in January. A similar phenomenon is also observed for the equally weighted index of stocks traded on the London Stock Exchange. IN A RECENT PAPER Fama and French (1988) report that "25-45 percent of the variation of 3- to 5-year stock returns is predictable from past returns" and pose a serious challenge to the long-held view that stock prices follow a random walk. In a subsequent paper, Poterba and Summers (1988) use variance ratio tests on the stock price data from the U.S. and 17 other countries and conclude that "The results consistently suggest the presence of transitory components in stock prices." However, Kim, Nelson, and Startz (1988) and Richardson (1989) criticize the earlier papers on statistical grounds and suggest on the basis of simulation evidence that the results of Fama and French and Poterba and Summers do not violate the random walk model. The conflicting opinions expressed in these papers clearly highlight the controversy surrounding the interpretation of the seemingly large point estimates of long-term serial correlation in stock returns. The economic interpretation of the estimates in these tests is difficult because they have low precision, and hence more efficient tests are required.