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

165 NTU Management Review Vol. 30 No. 2 Aug. 2020 The above Eq. (11) can be restated as follows: (12) where δ M is equivalent to β M in Eq. (3); and δ S is equivalent to β S in Eq. (3). H1 predicts negative δ M and δ S . H3 further predicts negative δ M and δ S , if agency costs, as measured by whether the executive served as the chairperson during the fiscal year, worsen the problem of asymmetric OCI adjustments. Panels A and B of Table 9 provide descriptive information for the variables used in Eq. (12). Panel C reports results that are consistent with H3. Although δ M has an insignificant coefficient, significantly negative δ S indicates that asymmetric OCI adjustments are more pronounced for firms with higher agency costs, as indicated by CEO duality, Dual i,t .