Detecting earnings management: a comparison of accrual and real earnings manipulation models

Purpose The use of models for detecting earnings management in the academic literature, using accrual and real manipulation, is commonplace. purpose current study to compare power these a United Kingdom (UK) sample 19,424 firm-year observations during period 1991–2018. authors include artificially-induced manipulation revenues expenses between zero ten percent total assets random samples 500 within full sample. two alternative samples, one with no reversal (sample 1) following year 2). Design/methodology/approach artificially induced Findings find that have lower than models, when manipulating discretionary revenues. Furthermore, model detect overproduction has high misspecification, resulting inflating model. examine an expense generates higher Roychowdhury (2006) Modified (Srivastava, 2019) are used as robustness be more misspecified some cases but less others. extend analysis setting which known occur, i.e. around benchmark-beating consistent evidence forms this all examined. Research limitations/implications This contributes literature by providing misspecification currently accounts manipulation. Practical implications Based on findings, recommend caution interpreting any findings future research. Originality/value address guided agency theory.