Evidence that Prices Do Not Fully Reflect the Implications of Current Earnings for Future Earnings : An Experimental Approach

Analysts have been found to underweight the innovation in most recent quarter earnings when forecasting next quarter earnings, and these expectations have been posited as an explanation for post-earnings announcement drift. This study uses an experimental asset market to examine whether similar errors are made by student subjects in forecasting quarterly earnings. We examine two aspects of behavior: (1) Do subjects ignore the autocorrelation in quarterly earnings when forming earnings expectations? and (2) Are asset prices consistent with subjects failing to account for the autocorrelation in quarterly earnings? We observe errors in forecasts by subjects which underweight, but do not ignore, extreme innovations in the most recent quarter earnings. The prices in the experimental markets also fail to reflect fully the most recent innovation in quarterly earnings. We are able to predict the sign of the mispricing in 74 percent of the 135 markets from the mean initial earnings predictions of the subjects. The forecast errors observed in this study are consistent with errors observed for analysts, and this consistency suggests that errors in analysts' forecasts may be at least partially attributable to the use of judgmental heuristics