The Accuracy and Complementarity of Analyst and Earnings-to-Price-Ratio-Based Forecasts of Annual Earnings

We compare the accuracy of analyst (I/B/E/S consensus) and earnings-to-price ratio (E/P)-based forecasts of annual earnings across firms. We find that generalizations of Beaver Lambert and Morse's (BLM 1980) E/P-based forecasting model are more accurate than analyst forecasts both for most firms and on average though analyst forecasts are more accurate for the two lowest and the highest E/P decile firms. This result reflects prior research's finding that analyst forecasts are biased and do not fully incorporate the implications of the current annual price and earnings changes for future earnings. Only when the errors in analyst forecasts due to these sources comprise a relatively unimportant portion of the subsequent earnings change (i.e. in the extreme E/P deciles) do analysts beat E/P-based models