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

157 NTU Management Review Vol. 30 No. 2 Aug. 2020 asymmetric OCI adjustment is less serious for firms with higher ethics as measured by CSR performance. Model (3) of Table 4 still supports H3 after including both market-based and accounting-based information inputs as regression independent variables. As we predicted in H3, both γ M and γ S are positive. However, although both are 0.004 as shown in the table, γ M is statistically significant at the 5% level ( t -statistic = 2.56) while γ S is insignificant ( t -statistic = 1.29). This difference in significance suggests that ethics matter for market-based, rather than accounting-based, information inputs. It is also consistent with the idea that market-based information inputs contain more diversified information other than changes in fair values (see, e.g., Barth et al., 1998; Eccher et al., 1996; Nelson, 1996). Table 4 Agency Costs and Asymmetric OCI Adjustments Expected sign (1) (2) (3) β 0 0.003** -0.012*** -0.026*** (1.96) (-8.73) (-15.49) β M 1 + 0.134*** 0.101*** (31.42) (25.33) γ M 0 – -0.091*** -0.064*** (-13.98) (-10.63) γ M KLD + 0.008*** 0.004** (4.71) (2.56) β S 1 + 0.357*** 0.326*** (56.24) (51.91) γ S 0 – -0.18*** -0.162*** (-20.87) (-19.01) γ S KLD + 0.009*** 0.004 (2.78) (1.29) Adj. R 2 0.0761 0.1726 0.2163 Hypothesis test of regression β M 1 + γ M 0 = 0.043*** 0.037*** F-value (117.3) (98.26) β M 1 + γ M 0 + γ M KLD = 0.051*** 0.041*** F-value (123.22) (89.82) β S 1 + γ S 0 = 0.177*** 0.164*** F-value (583.62) (521.58) β S 1 + γ S 0 + γ S KLD = 0.186*** 0.168*** F-value (450.69) (376.09)

RkJQdWJsaXNoZXIy MTYzMDc=