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

111 NTU Management Review Vol. 30 No. 2 Aug. 2020 countries (i.e., Bermuda, Channel Islands, Marshall Islands, and British Virgin Islands) the OECD has designated as tax havens. For robustness check, we also repeat the analysis after eliminating firms that domicile in counties (i.e., Argentina, Channel Islands, China, Ireland, Korea, and Marshall Islands) with only one individual firm. We find that parameters and significance levels on all of the variables are generally unchanged from those in Table 5. 7.6 Pre-Adoption Differences Pre-existing differences and/or changes around the relaxation of reconciliation requirement could drive the documented differences between firms voluntarily switching to U.S. GAAP and those to IFRS. To examine that possibility, we first conduct similar tests for our U.S. GAAP and IFRS test samples using firm-year observations during the U.S. GAAP reconciliation sample period 2003-2006. 26 Untabulated results suggest that discretionary accruals and real activities manipulation are partial substitutes for earnings management and that their magnitudes are determined simultaneously for both samples of U.S. GAAP and IFRS firms in the U.S. GAAP reconciliation period (all significant at p = 1%). However, with respect to the reporting incentives, only the proxies for “income smoothing” reporting (i.e., SM_SUM it , SM_PROD it , and SM_PROD it ) are significantly negative as predicted at p = 1% significance level for both samples. The proxies for the existence and tightness of debt covenants (i.e., DEBT it and ST_DEBT it ) are significantly positive as predicted at better than p = 10% only for the U.S. GAAP firms. The coefficient on the “big bath” variable ( BBATH it ) either is insignificant or does not have the predicted sign for the two samples. The results suggest that the “big bath” incentive does not play an essential role in determining the magnitudes of accrual-based earnings management and real activities manipulation in the U.S. GAAP reconciliation period, which are inconsistent with the findings from our primary analysis, namely the IFRS reporting period. The combined evidence suggests that differences in the earnings management strategies between these two categories of firms before the SEC eliminated the reconciliation requirement generally do not explain their differences after the elimination. 26 The sample to examine the pre-existing differences consists of 83 (132) firm-year observations during the reconciliation sample period for firms changing their filing status to U.S. GAAP (IFRS) voluntarily after the elimination of reconciliation requirement.