Properties of the Variance of Analysts' Forecast of Earnings in Good-News and Bad-News Environments: Theory, Evidence and Usefulness

Abstract

This study documents that the variance of analysts' forecast of earnings is smaller when the expected or actual news of earnings is smaller when the expected or actual news about earnings is good (relative to when it is bad). An examination over the entire fiscal year provides evidence of earlier consensus building amongst analysts, as measured by reduction in variance over time in the good-news environment. Forecasts of earnings can be improved when variance of analysts' forecasts are used along with its mean. A trading strategy based on the variance of analysts' forecasts earns positive abnormal returns, when conditioned upon the nature of the news.

Keywords: analyst forecasts, analyst forecast dispersion, earnings prediction

JEL Classification: G14, G29, M41