Earnings Forecasts and the Predictability of Stock Returns:Evidence from Trading the S&P

Abstract

We develop a simple error-correction model, based on a well known theory espoused by Benjamin Graham and David Dodd, and others, which presumes stock returns tend to restore an equilibrium relationship between the forecasted earnings yield on common stocks and the yield on bonds. The estimation uses I/B/E/S analysts forecasts of S&P earnings. To evaluate the model, we use rolling regressions to obtain out-of-sample forecasts of excess returns. Tests of association show the implicit timing signals to be statistically significant. Further, a strategy of investing in cash when the excess return is forecasted to be negative and in the S&P otherwise outperforms the S&P, yielding higher returns with smaller volatility. Using the bootstrap methodology we demonstrate that the findings are statistically significant.

JEL Classification: G11, G14