Change in Systematic Bias and News as Drivers of Earnings Forecast Revision - A New Forecast Revision Model

Abstract

We predict and find that revisions are driven by the same determinants as forecast errors. In addition to the intuitive impact of news on revisions, we show that a second major driver of revisions is the change in analyst incentives to systematically bias their earnings forecasts. Taken together, both drivers explain about 42% of the variability in earnings forecast revisions. Price reaction to revision depends crucially on the underlying revision driver. Therefore, when allocating a revision according to the revision drivers, the explanatory power of a regression of price changes on revision more than doubles. Moreover, we find that revisions can be at least partly predicted by using prior information. This information refers to the downward revision to compensate for the initial induced forecast optimism.

Keywords: Analyst earnings forecast revision, forecast error, systematic bias

JEL Classification: G14, G17