ABSTRACT
I empirically assess the extent to which real earnings management metrics capture opportunistic behavior versus firms' fundamental factors such as performance. For the traditional proxies proposed by Roychowdhury (2006), I find (1) the economic magnitude of the proxies to be high relative to two relevant benchmarks; (2) they exhibit persistence; and (3) they vary predictably with performance. These findings suggest that the traditional proxies likely not only capture opportunistic behavior, but also likely reflect fundamental factors. I also examine several adjusted proxies based on refinements proposed by subsequent studies. I find that those proposed by Vorst (2016) and those based on Kothari, Mizik, and Roychowdhury (2016) seem to be the most effective at attenuating correlation with underlying fundamentals. Additional simulation tests on bias and power reveal that, between the two adjusted proxies, those based on Kothari et al.'s (2016) are generally more preferable.
JEL Classifications: M41; M1.