A Re-Examination of Analysts’ Superiority over Time-Series Forecasts of Annual Earnings

Abstract

We re-examine the widely held belief that analysts’ earnings per share (EPS) forecasts are superior to random walk (RW) time-series forecasts. We investigate whether analysts’ annual EPS forecasts are superior, and if so, under what conditions. Simple RW EPS forecasts are more accurate than analysts’ forecasts over longer horizons, for smaller or younger firms, and when analysts forecast negative or large changes in EPS. We also compare the accuracy of a third forecast of longer-term earnings based on a naïve extrapolation of analysts’ one-year-ahead forecasts. Surprisingly, this naïve extrapolation provides the most accurate estimate of long-term (two- and three-year-ahead) earnings. These findings recharacterize prior generalizations about the superiority of analysts’ forecasts and suggest that they are incomplete, misleading, or both. Moreover, in certain settings, researchers can rely on forecasts other than these explicit forecasts.

Keywords: analyst forecasts, earnings per share, time-series forecasts, earnings forecasts, earnings expectations

JEL Classification: M43, M41