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

85 NTU Management Review Vol. 30 No. 2 Aug. 2020 IFRS after the elimination using a probit model. We match each sample firm, based on industry, size (i.e., market value of equity), and year, with a comparison firm from the same country that elects not to adopt U.S. GAAP/IFRS after the elimination and continues to report using local standards. Following prior studies such as Harris and Muller (1999) and Hung and Subramanyam (2007), we predict that the decision to adopt an internationally recognized accounting standard (i.e., U.S. GAAP or IFRS) is a function of economic factors, such as financial performance, leverage, firm sizes, and financial needs. Specifically, we conduct split-sample tests by estimating the following probit model separately for the two samples (i.e., U.S. GAAP and IFRS) and their corresponding comparison observations: SELECT it = α 0 + α 1 ROA it + α 2 LEV it + α 3 MV it + α 4 NMKT it + α 5 CS it + α 7 LT_DEBT it + ε it (5) where SELECT it is an indicator variable equal to 1 for the sample (U.S. GAAP/IFRS) firms and 0 for the comparison firms. ROA it is return on assets, which equals firm i ’s net income divided by total assets at the end of t . LEV it is proxy for leverage, which is firm i ’s total liabilities divided by total asset at the end of t -1. MV it is firm size, which equals the natural logarithm of firm i ’s market value of equity at the end of t . NMKT it is the number of foreign markets, excluding the U.S., in which firm i ’s shares are traded during year t . CS it is an indicator variable equal to 1 if there is an increase in firm i ’s par value of common stock during year t . LT_DEBT is an indicator variable equal to 1 if there is an increase in firm i ’s long-term debt during year t . We obtain z it , the fitted value of the probit regression index function, and calculate inverse Mills ratios ( MILL it ) as φ( z it )/Φ( z it ), where φ is the standard normal density and Φ is the normal cumulative probability. In the second stage of the Heckman (1979) procedure, we include MILL it in the tests of the relation between discretionary accruals and real earnings manipulation as an additional control variable to correct for biased estimates that result from a nonrandom treatment effect. 4.2 Empirical Tests: The Relation between Accrual-Based Earnings Management and Real Earnings Manipulation To test whether managers determine the levels of accrual management and real activities manipulation simultaneously, we use the following system of equations to