This paper evaluates the extent of predictable measurement error induced by five alternative approaches to the estimation of discretionary accruals. The source and magnitude of the error is assessed by reference to the strength of the association between the discretionary accrual estimate and proxies for the non‐discretionary components of total accruals. Results indicate that discretionary accruals generated by the Healy model are associated with the highest level of predictable measurement error. While the remaining four models generate significantly lower error levels, the magnitude of error remains highly significant. Findings highlight the limitations of existing models and emphasise the need for further developments in relation to the measurement of earnings management activity.
No takes yet. Share an insight, caveat, or question.
A 1999 study studied this question.