The implications of piecewise linear process of normal accruals

The present study investigates whether the basic assumption in the Jones model, which normal accruals are a linear function of change in sales, is empirically valid. It also discusses and addresses the implications of the assumption violation in the earnings management detection tests. The research employs a sample of 2832 observations of the annual information of firms listed in Tehran Stock Exchange and Iran Fara Bource for the period of 2002 to 2017. The findings suggest that normal accruals follow a piecewise linear function with respect to sales changes. The results also suggest that the Jones model generates predictable systematic measurement error, which is consistent with its failure in capturing the accruals nonlinear behavior. The results of a Monte Carlo simulation of the type I error show that the level of abnormal accruals are understated for extreme sales growth and overstated for medium sales growth. The research highlights the importance of controlling the asymmetric behavior of normal accruals, especially in studies where earnings management incentives are correlated with sales growth.