Exploring alternative methods for evaluating real earnings management: an overview, comparison, and assessment

Liao, Wen-Ju, Lee, Cheng-Few, Sung, Hao-Chang
Review of Quantitative Finance and Accounting · Springer Nature · 2026-10-22 · DOI: 10.1007/s11156-026-01555-0

Abstract

How researchers measure real earnings management (REM) shapes inferences about opportunistic operating behavior. An abnormal operating decision may reflect reporting-motivated manipulation, a rational response to economic fundamentals, or model error, creating a fundamental construct-validity problem. We review major reduced-form REM approaches and replicate and re-estimate prior R&D- and SG&A-based evidence, extending it to compare alternative earnings-benchmark definitions and evaluate REM proxies across incentive alignment, intertemporal persistence, and future operating performance. Operating/core-earnings benchmarks show stronger incentive alignment primarily for SG&A-related REM, while R&D evidence is more sensitive to benchmark and proxy choice. Standard and total Tobin’s Q do not systematically improve expected R&D and SG&A models relative to classic Q. Abnormal R&D and SG&A cut-intensity measures exhibit relatively strong incentive alignment, particularly for SG&A, but greater persistence and mixed future-performance validity. We further develop an identification-oriented framework for structural REM research. Overall, no single REM proxy dominates across validity dimensions; measurement choice should reflect the research objective, construct-validity criterion, and available identifying variation.

Read at publisher →

Similar articles