The out-of-sample forecasting performance of traditional stock return models (dividend yield, t-bill rate, etc.) is compared with the forecasting performance of the Livingston survey. The results suggest that the survey forecasts are much like a “too large” forecasting model: poor performance...