Measuring the Prevalence of Earnings Manipulations: A Novel Approach

Abstract

We provide prevalence estimates for five forms of earnings manipulation based on executives’ reports about their firms’ actual reporting practices. After preregistering our methods and analyses via the Journal of Accounting Research’s registration-based editorial process, we recruit nearly a thousand executives from firms listed in the Russell 3000 Index to participate in either a survey or a list experiment; the hallmark of the latter being additional privacy protections designed to promote honest disclosure about self-incriminating information. In our survey, 26.8% of executives disclose at least one form of earnings manipulation at their firm in the 2018–2023 period: 18.0% report changing an operational activity to meet a near-term earnings target at the expense of long-term value (i.e., real earnings management), 8.8% report intentionally obfuscating unfavorable information, 6.6% report manipulating accruals, 3.9% report withholding material information, and 0.0% report accounting fraud. Our list experiment produces an economically higher result in two cases, estimating that 29.9% of firms engaged in real earnings management and 12.4% committed accounting fraud over the same time period. We conclude that while a traditional survey can provide credible lower-bound estimates for the prevalence of many forms of earnings manipulation, list experiments encourage more honest disclosure in some cases.