臺大管理論叢 NTU Management Review VOL.30 NO.2

The Effects of Relaxing the Reconciliation Requirement in Foreign Private Issuers’ SEC Filings on Earnings Management Strategies: IFRS Adopters versus U.S. GAAPAdopters 108 7 Further Checks and Additional Analyses 7.1 Alternative Significance Tests A potential problem with statistical inferences in small samples is the validity of the normal distribution assumption (Hung and Subramanyam, 2007). To address this issue, we apply the bootstrapping methods, which are resampling techniques for assessing uncertainties, to the estimations of the equations in Table 5 (see Efron and Tibshirani, 1993). Specifically, we use the bootstrapping procedures for the significance tests by constructing 1,000 random samples and assessing the 5% and 95% confidence limits based on 1,000 random parameter estimates. The results (not reported) are qualitatively similar and the inferences are unchanged. 7.2 Validation of “Income Smoothing” Proxies Given the inherent difficulty in knowing managers’ true motives for their behaviors, one criticism of the literature is that earnings management may be due to behavior other than managerial opportunism. A large number of studies (e.g., Dierynck, Landsman, and Renders, 2012; Gunny, 2010; Yin and Cheng, 2004; Zang, 2012) provide evidence that firms manage earnings to avoid breaching debt covenants, to take a “big bath,” or to show a smooth earnings pattern (mostly to meet internally or externally imposed earnings benchmarks such as prior year’s earnings, zero earnings, and analyst consensus forecast). However, few studies (e.g., Roychowdhury, 2006; Zang, 2012) indicate that firms just meeting/beating benchmarks might not be the only firms that manage earnings. Other firms might manage earnings and still miss the aforementioned benchmarks, or they may manage earnings for other incentives or unobservable targets. Accordingly, we choose more general proxies for income smoothing in our primary analysis. Nonetheless, we attempt to specify SMOOTH it better by refining the three measures for income smoothing to provide additional assurance that they are capturing the intended reporting incentives. Specifically, we redefine SMOOTH it (i.e., SM_PROD it , SM_DISX it , and SM_SUM it ) to equal the difference between firm i ’s current earnings (measured before discretionary accruals and real activities manipulation such as abnormal production costs, abnormal discretionary expenditures, and their aggregate income effects, respectively) and prior-year earnings, divided by beginning total assets, when this change is either above the median of nonzero positive values or below the median of nonzero negative values of this variable and when firm i just meets either prior year’s earnings or zero earnings