Mean Reversion of Abnormal Stock Returns

Abstract

This study tests for mean reversion in abnormal stock returns that divert more than one standard deviation from the mean. Biases due to a small sample, the January effect, and unique events are avoided by using large samples generated by a block bootstrap procedure starting in random months and studying two different periods. The results show stronger mean reversion in abnormal returns than in all returns for large- and small-company stocks during both periods, thus supporting the rationale for time diversification. Both large and small-company stocks exhibited the strongest mean reversion for four-year returns from 1926–1966 and for five-year returns from 1967–2007.