Is there a consensus trap in earnings forecasts? : an empirical study
Abstract
This paper revisits Grüner (2009) and seeks to establish whether there is a consensus trap in
earnings forecasts. There is a consensus trap if analysts’ forecasts are more likely to be wrong
when forecasts are homogeneous, than heterogeneous, all else equal. This hypothesis is tested
by using a standardized measure of the forecast distribution to explain forecast errors. The
empirical research is based upon earnings forecasts recorded on Thomson Reuters I/B/E/S
Summary-Level Historical Earnings Estimates Database. Our results rejects that there is a
consensus trap in earnings forecasts. The empirical research de facto shows evidence of a
significant positive relationship between forecast errors and heterogeneity. Idiosyncratic risk
in earnings is then offered as a mechanism explaining the findings, by showing that
heterogeneity proxy idiosyncratic risk.
The paper contributes to the literature on forecast dispersion and systematic forecast errors. It
also offers an empirical founded mechanism explaining why there should be a positive
association between forecast errors and heterogeneity. As this research is based upon
summary-level data, we would recommend subsequent researchers to examine detailed data.