آیا معیار غلبه مدیریت سود را تشخیص می دهد؟ شواهد ناشی از بی نظمی حسابداری Does benchmark-beating detect earnings management? Evidence from accounting irregularities

1. Introduction We examine whether meeting or slightly beating an earnings benchmark (hereafter, benchmark-beating) is (1) associated with accounting irregularities, an extreme and certain case of earnings management, (2) useful for detecting accounting irregularities both incremental and relative to discretionary accruals and to F-scores (Dechow, Ge, Larson, & Sloan, 2011), and (3) more useful for detecting opportunistic accounting irregularities, a more harmful form of earnings manipulation identified in Badertscher, Collins, and Lys (2012), than accounting irregularities in general. The literature documents three earnings benchmarks and measures benchmark-beating by identifying firms whose earnings slightly increase from last year’s earnings (the earnings change benchmark), whose earnings are slightly positive (the earnings level benchmark), and whose earnings are equal to or slightly above analyst earnings forecasts (the earnings forecast benchmark). Our research questions are important for several reasons. First, a large and growing volume of studies in the accounting literature use benchmark-beating as a proxy for earnings management while evidence that links benchmark-beating to actual earnings management is limited (see more detailed discussion in the next section). Dechow, Ge, and Schrand (2010), p.365) make the above point clear when they conclude, after reviewing the vast literature of earnings quality and earnings management, that “[t]the totality of the evidence indicates that the use of small profits as a proxy for earnings management more generally is unsubstantiated (emphasis added).” We seek to provide evidence on a link between benchmark-beating and earnings management in this paper. Second, benchmark-beating and discretionary accruals are arguably the two most widely used proxies for earnings management (Kothari, 2001) and F-scores are arguably the most powerful detector of earnings misstatements (Dechow et al., 2011). However, little is known about whether benchmark-beating, discretionary accruals, and F-scores capture the same or different aspects of earnings management, and how these three measures compare with one another in terms of detecting earnings management.1 If benchmark-beating captures the same aspects of earnings management as discretionary accruals and F-scores, the coefficient on benchmark-beating could become insignificant, in an earnings management detection model where benchmark-beating is an explanatory variable, after including discretionary accruals and Fscores as additional explanatory variables. On the other hand, if benchmark-beating captures aspects of earnings management different from discretionary accruals and F-scores, the coefficient on benchmarkbeating will remain significant after including discretionary accruals and F-scores as additional explanatory variables. In such a case, benchmark-beating has detective power for earnings management incremental to discretionary accruals and F-scores. Relatedly, it is also of interest to examine which measure, out of these three, is relatively superior in a one-on-one horse race to detect earnings management. Dechow et al. (2011, p. 23) highlight the need to compare different measures of earnings management by calling for future research “to analyze the role of governance, compensation, insider trading, short selling, incentives to meet and beat analyst forecasts, and so on and to determine the relative importance of these variables (emphases added) over financial statement information in detecting overstatements of earnings.” We answer this call in Dechow et al. (2011).