Trade and the Comovement of Stock Returns: Evidence from Japan

In April 2000, in one day, 30 stocks were replaced in the Nikkei 225 index in Japan. We analyze the change in comovement of returns of stocks added to and deleted from the index with the returns of stocks remaining in the index. A simple model shows that upon inclusion into (deletion from) a stock index, stocks should begin to comove more (less) with the index, due to a change in their trading pattern. The empirical findings provide sound support for these predictions: In the sample, daily index betas of the added stocks rose by an average of 0.60, while the average beta of the deleted stocks fell by 0.71. Our results confirm additional predictions of the model for changes in R2, turnover, and the autocorrelation of returns upon index inclusion and deletion, and hold at daily, weekly and bi-weekly return horizons. Fundamentals based explanations fail to account for these findings. We conclude that correlated trading of index stocks causes excess comovement of stock returns. We argue that the distinct trading mechanism on the Tokyo Stock Exchange contributes to the significance and magnitude of our results.