191 /304
191
臺大管理論叢
第
28
卷第
1
期
divergence is less in countries with better investor protection when investment
opportunities are good, which is consistent with our hypothesis. In countries in which
minority shareholders have legal rights upon which they can rely and are thus less alert to
expropriation, controlling shareholders with a large control divergence have less need to
mask their expropriation activities by paying higher dividends. Our results suggest that
strong legal institutions play an effective monitoring role in the determination of the
dividend policies of controlling shareholders by mitigating the efforts of these
shareholders to mask their expropriation activities. We do not observe consistent patterns
for the coefficients of
Diverge*Legal
(
Right
or
Disclosure
), which indicates that
investment opportunities are an important determinant of dividend policy (Fama and
French, 2001).
The standalone control variables are reported in Panel A and are mostly significant,
as predicted in previous studies (Fama and French, 2001; La Porta et al., 2000; Faccio et
al., 2001). For instance, the coefficients of growth potential (
G
) are significantly negative
in all of the specifications, which implies that faster growing firms pay lower dividends.
The tax variable (
Dta
) is significantly positive in all of the specifications. The
(unreported) control variables in Panels B and C display almost the same patterns as those
in Panel A. In terms of explanatory power, the adjusted
R
2
is the highest with
IADvd/CFO
(23.52% in Panel A) and the lowest with
IADvd/E
(7.01% in Panel A), which may reflect
the pervasive earnings management and low earnings quality of firms with a large control
divergence (Fan and Wong, 2002; Haw et al., 2004).
We rerun regression equation (1) on the four dividend payout ratios before adjusting
for industry effects and add industry dummies (based on the 2-digit primary SIC code) to
the regression. Appendix B reports the coefficients and significant levels, which are quite
similar to those that are reported in Table 4 and from which it can be seen that the thrust of
our results remains the same.