An Examination of Whether Firms Alter the Normal Evolution of Fourth Quarter Earnings to Avoid Losses

Abstract

Our paper provides a comprehensive examination of issues and concerns in the current unresolved debate over the existence of loss avoidance behavior. We rely on a methodology that avoids previous strong assumptions for approximating pre-managed earnings and provide, instead, new earnings management tests. The primary issue that we consider is whether firms alter the normal evolution of earnings around the zero profit line relative to other nearby locations in the earnings distribution. Using generalized logistic regressions, we find that firms starting in adjacent intervals of the cumulative earnings distribution after three quarters generally have the same odds of experiencing a given fourth quarter earnings except when such fourth quarter evolutions result in annual earnings in close proximity to the zero profit line. It is there that the odds of a given magnitude of fourth quarter earnings diverge in favor of firms experiencing smallest annual profits, in comparison to firms experiencing smallest annual losses. Our results are robust to the choice of earnings measure: earnings scaled by size, net income, and earnings-per-share.