55 / 322
臺大管理論叢
第
26
卷第
3
期
55
2010; Wang and Guo, 2011). However, researchers who have argued that tax rate changes are
unrelated to corporate dividend payments have based their arguments not on the Tax
Capitalization View, but on dividend stability (Alli et al., 1993; Brav et al., 2008). Both the
United States and China are under classical tax systems that differ from Taiwan’s integrated
income tax system. Such results are unsuitable for Taiwan’s condition.
Previous research on countries that use an integrated income tax system has supported
the argument that tax reforms affect dividend payments (Alstadsater and Fjærli, 2009;
Korkeamaki et al., 2010; Kaserer et al., 2011). Similarly, those studies have documented the
impact of changes in personal tax rates or the introduction of partial double taxation, which
also differs from the reduction of corporate income tax rate in Taiwan.
This study was conducted to contribute to the dividend literature by analyzing whether
corporations have retained their earnings rather than distributing them to shareholders in
order to reduce investor tax burdens since the tax reform act reduced Taiwan’s corporate
income tax rate from 25% to 17%. In addition, we also discuss the moderating effects of a
stable payout policy on dividend payout ratios.
2. Design/Methodology/Approach
The research sample comprised of financial statement data from 2007-2012 of
companies listed on the Taiwan Stock Exchange and GreTai Securities Market. Empirical
data were retrieved from the Taiwan Economic Journal Database. Following prior studies,
we excluded financial institutions and insurance firms due to their regulated characteristics.
In addition, we omitted financially distressed firms and excluded firms with missing data. To
reduce the effects of extreme observations, we omitted any observations where the variables
used in the regression were more than three standard deviations from their mean value. These
selection procedures yielded a final sample of 5,698 firm-year observations; 2,809
observations were from the period before 2009, and 2,889 observations were from the period
after 2010.
The dependent variables used for the regression analysis are dividend payout ratios
based on total, stock, and cash dividend payouts. The independent variables comprise an
indicator variable with a value of one for observations following the reduction of corporate
income tax and zero if otherwise, dividend stability, growth opportunity, insider ownership,
institutional ownership, quick assets to total assets, firm size, debt ratio, profitability, gross
domestic product growth rate, consumer price index growth rate, and industry-specific
dummy variables.