Does the Stock Market Overreact?

De Bondt, Werner F. M. and Richard Thaler. “Does the Stock Market Overreact?” The Journal of Finance Vol. 40, No. 3 (1985): 793-805. (link).

The authors perform empirical tests to determine whether there is aggregate evidence in the stock market which is consistent with the overreaction phenomenon observed in experiments of individual decision making in psychology. The idea of this paper is that, if agents react individually in an irrational manner to good or bad fundamental news, the aggregate effect of the irrational individuals will be to push stock prices too high or too low relative to their fundamental value. Those irrational valuations should, over time, be corrected and the evidence of overreaction will be observed through return predictability based on past returns. The main assumption here is that stocks with the highest past returns represent those which are most likely to have had an overreaction to positive fundamental news, and vice versa. Given this, the authors state two hypotheses:

1. “Extreme movements in stock prices will be followed by subsequent price movements in the opposite direction.”
2. “The more extreme the initial price movement, the greater will be the subsequent adjustment.”

Main Empirical Test:

The authors first sort NYSE stocks at each time, t, based on past excess returns and group them into “winner” and “loser” portfolios with the former being made up of stocks with the highest excess returns and the latter made up of stocks with the lowest excess returns. They measure excess returns in three ways; 1) market-adjusted excess returns, 2) residuals from a market model, and 3) excess returns from the Sharpe-Lintner CAPM. The results are similar regardless of the measure of excess returns, so they choose to report only the results from the first. That is:

where R_jt is the return on stock j at time t and R_mt is the return on the market at time t. They then look at how each of these portfolios, formed based on return information prior to time t, performs, on average, after time t. The authors use 16 test periods where they average the excess return across all stocks in each portfolio each month for 36 months after portfolio formation. Those returns are then cumulated across the 36 months and then the cumulative returns are averaged across the 16 test periods for each month.  Figure 1 from their paper provides the main result.

As we can see, the portfolios made up of previous “loser” stocks significantly outperform both the market and the “winner” portfolios, on average, on a cumulative basis over the following 36 months. The difference between “loser” and “winner” average cumulative returns at month 36 is 24.6%. They also run other tests using alternative number of months for portfolio formation, different average number of stocks in each portfolio, and alternative number of months of post formation cumulative returns and show similar results

Conclusion:

The evidence the authors present is consistent with their hypotheses of overreaction in stocks and poses a significant challenge to an efficient market.