A New Measure of Earnings News and Post Earnings Announcement Drift

Abstract

This paper introduces a new proxy for expected value-relevant earnings: the most optimistic (pessimistic) forecast of earnings that is higher (lower) than the median of all analysts’ earnings forecasts over the 90 days prior to the earnings announcement when the median falls short of (exceeds) actual earnings. While our measure is ex post in the sense that it occurs after the distribution of forecasts over the 90 days prior to the earnings announcement is known, it is ex ante in the sense that investors could use our measure to compute earnings surprise at the time of the earnings announcement and take a trading position that potentially generates post-earnings announcement drift. We expect our ex post measure of earnings surprise to have a moderating effect on the traditional earnings news measure based on the consensus of analysts’ forecasts. For example, if the consensus is less than actual earnings and the most optimistic forecast is close to (much greater than) the consensus, then we expect the traditional analyst forecast error to contain more (less) information about future earnings, and we expect a larger (smaller) contemporaneous and delayed market reaction to earnings surprise based on the consensus forecast. Our results confirm these expectations. Furthermore, we find that our new measure of earnings surprise contains unique value-relevant information about future earnings, some of which generates a statistically significant immediate contemporaneous market response and some of which generates a statistically and economically significant amount of post-earnings announcement drift. We appreciate Henk Berkman’s helpful comments.