8. Conclusion
Motivated by two major behavioral principles, we investigate the short-term dynamics of manager’s forecasts and
examine whether managers become overconfident in their ability to predict future earnings after a series of good
predictions. In contrast to other studies (e.g., Hribar and Yang, 2010), we treat overconfidence as endogenous, with an
intensity that varies with the length of success. Overconfidence in this setting implies that managers weight their own
estimates too heavily and rely too little on public signals. If this is the case, then the subsequent forecasts of these
managers would be more likely to display a greater prediction error. To test this hypothesis, we regress forecast error on
the number of accurate predictions in the previous four quarters, and find that forecast error in the current period is
positively correlated with past success. The effect is both statistically and economically significant. In addition, these
overconfident managers display greater divergence from the analyst consensus and are more precise. Finally, we find that
market participants downplay the forecasts issued by overconfident managers. More specifically, after controlling for
manager fixed effects, we find that both investors and financial analysts place less weight on the forecasts issued by
managers who have recently made a series of accurate predictions.