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

153 NTU Management Review Vol. 30 No. 2 Aug. 2020 By contrast, the two independent dummy variables, D M and D S , are either insignificant or merely significant at the 10% level. Table 2 Asymmetric OCI Adjustments Expected Model sign (1) (2) (3) (4) β 0 0.019*** 0.013*** -0.006*** -0.008*** (15.9) (13.58) (-4.82) (-4.66) β M 1 + 0.129*** 0.093*** 0.096*** (52.63) (40.05) (34.66) β M 2 – -0.127*** -0.09*** -0.09*** (-37.53) (-28.03) (-27.91) β S 1 + 0.283*** 0.262*** 0.263*** (97.12) (89.83) (78.7) β S 2 – -0.125*** -0.112*** -0.111*** (-34.17) (-30.5) (-30.26) Coef. on D M 0.004* (1.82) Coef. on D S 0.001 (0.66) Adj. R 2 0.0497 0.1524 0.1777 0.1777 Hypothesis test of regression β M 1 + β M 2 = 0 0.002 0.003 0.006 F-value (0.77) (2.63) (5.90) β S 1 + β S 2 = 0 0.158*** 0.15*** 0.152 F-value (2,466.84) (2,276.37) (1,659.20) Note: This table presents results from a pooled cross sectional OLS regression of the following model: where NA i,t is the book value of net assets, measured as total assets (Compustat item “AT”) minus total debt (“DT”) of firm i at the end of fiscal year t , and excluding net income (“NI”) and shares sold (“SSTK”), plus dividends paid (“DVC”) over year t ; and MVE i,t is the market value of equity, measured as the absolute value of price (“PRCC_F”) times shares outstanding (“CSHO”) of firm i at the end of year t . In model (4), we allow a different intercept by adding D M i,t , which is a dummy variable that takes the value of 1 when MVE i,t is less than MVE i,t –1 , and 0 otherwise. Sale i,t is net sales (“SALE”) of firm i in year t . In Model (4), we allow a different intercept by adding D S i,t , which is a dummy variable that takes the value of 1 when Sale i,t is less than Sale i,t –1 , and 0 otherwise. The sample period ranges from 2000 to 2014 (15 years), covering 55,541 observations for 9,610 firms. *, **, and *** indicate statistical significance at the 0.10, 0.05, and 0.01 levels, respectively, based on two-tailed t -statistics in parenthesis.