A General Examination of Earnings Management Around Financial Benchmarks

Abstract

Previous literature on earnings management has examined the maintained hypothesis that firms barely beating earnings benchmarks are earnings manipulators with earnings before accounting manipulation otherwise slightly below their benchmarks and has implemented research designs that treat all suspect earnings-managing firms as engaging in this behavior. Little support has been found for this hypothesis. In this paper, we develop a two-way, reported earnings conditional on non-discretionary earnings, research design to examine this conjecture and find a more complex story: earnings management to barely meet benchmarks consisting of positive, negative, large, and small discretionary accruals – earnings smoothing. This result holds for all three financial benchmarks (i.e., zero earnings levels, earnings changes, and analyst forecasts) across a number of conservative and robust research designs.