Analysts' Forecasts in 'Good-News' and 'Bad-News' Environments: Evidence of Economic Significance of the Differential Variance

Abstract

There is compelling empirical evidence to suggest that good news about corporate performance comes out early but bad news comes out late. Regardless of the underlying reason for such differential timing, existence of such systematic behavior leads to three related questions that we seek to answer in this paper. First, would the existence of this differential timing fundamentally alter the information environment surrounding the main event (earnings announcement) in a way that the forecasts of earnings under a good-news and a bad-news environment would be associated with different levels of uncertainty exhibited by different forecast variances? Second, if earnings forecasts do have different variances under good-news and bad news environment, can these variances be exploited to construct an improved consensus earnings forecast? Finally, can we make ex-ante arbitrage profit based on portfolios constructed on variance of analysts' forecasts?

JEL Classification: M41, M45, G29, G14