Discontinuity in Earnings Distributions: A Theory and Evidence

Abstract

This paper presents a model of financial reporting in which investors infer both pre-managed earnings and the precision of earnings from reported earnings. Over-reporting earnings has two opposing pricing effects: investors infer higher pre-managed earnings from an inflated positive earnings surprise (i.e., positive effect); however, investors also infer a lower earnings precision, leading to a lower pricing weight placed on the higher surprise (i.e., negative effect). For firms with strongly positively autocorrelated earnings, the trade-off between the two opposing effects creates a pooled report right above the prior mean of the earnings distribution and a no-reporting "hole" right below the prior mean in equilibrium (i.e., an earnings discontinuity consistent with empirical findings). The model also predicts: (1) no earnings discontinuity exists for firms with negatively or weakly positively autocorrelated earnings, and (2) the earnings discontinuity is more pronounced for firms with more positively autocorrelated earnings. The empirical evidence supports the two predictions.

Keywords: earnings discontinuity, earnings precision

JEL Classification: M41, M43