EVALUATING UNEXPECTED ACCRUALS TO DETECT ACCOUNTING MANIPULATIONS: EVIDENCE FROM EMERGING MARKETS

Abstract

This paper estimates and tests the power of unexpected accruals models as indicators of earnings misstatement. The models are implemented on a comprehensive sample of listed firms in Africa, with available information from 2006 to 2020. The paper aims to determine the specification-correctness as well as to confirm the most powerful of the models. The findings suggest that the models are well-specified when used on the considered firm-years. All the models are found to be powerful, although the Jones, modified Jones and adapted models are identified as most powerful on the basis of the induced expense (revenue) manipulations. At 5% level, the expense (revenue) manipulation of 6–10% accommodate at least 97% (98%) nulls’ rejections for Jones, modified (Jones) and adapted models, but slightly reduced rejections ranging 95%–98%, at 1% test level. This study offers vital assessment of earnings manipulations in trying to exploit future earnings to grow stock prices.