Predictable errors in financial analysts' annual earnings forecasts and the evaluation of earnings forecast -based securities returns anomalies
This study addresses certain properties of the analysts' forecasts adjusted by predictable error patterns and re-examines the association between subsequent abnormal securities returns and two earnings forecast-based measures (Elgers, Lo, and Pfeiffer, 2001; Frankel and Lee, 1998).
The adjusted analysts' forecasts are found to be more accurate than the unadjusted analysts' forecasts, especially for firms with prior poor performance. Furthermore, the study shows that the adjusted analysts' forecasts are improved proxies for market expectations of earnings, compared to the unadjusted analysts' forecasts.
Elgers, Lo and Pfeiffer (2001) document a profitable hedge portfolio strategy based on the price-scaled analysts' forecasts. This study shows that the price-scaled adjusted analysts' forecasts (the more accurate forecasts) generate descriptively smaller amounts of hedge portfolio returns, though the decrease in hedge portfolio returns is not statistically significant. The lack of significant change is due to the minor impact of the adjustment on the composition of hedge portfolios. On the other hand, the directional decrease in hedge portfolio returns may suggest that analysts aim to forecast value-relevant earnings rather than actual earnings, especially for the early-in-the-year forecasts.
Frankel and Lee (1998) show that price-scaled implicit firm-values are reliable predictors of subsequent securities returns. The implicit firm values are generated by using analysts' forecasts as proxies for market expectations in a manner consistent with Ohlson's (1995) residual income model. The results show that the price-scaled implicit firm values generate very similar amounts of hedge portfolio returns after analysts' forecasts are adjusted. It is also because the adjustment of analysts' forecasts in this study is too minor to change the composition of hedge portfolios.
Rates of return;