181 /304
181
臺大管理論叢
第
28
卷第
1
期
hypothesis is stated as follows.
Hypothesis: The positive relationship between dividends and the control divergence
of controlling shareholders is less pronounced in countries in which legal
investor protection is strong and the investment opportunities of firms
are good.
We employ three measures of legal investor protection and the governance of
controlling shareholder activities in accordance with the work of La Porta, Lopez-de-
Silanes, Shleifer, and Vishny (1998) and La Porta et al. (2000): law origin, the anti-
director rights index, and the rating of disclosure standards. Each variable measures a
specific aspect of legal protection. Law origin (
Legal
) measures the traditional quality of
legal protection, the anti-director rights index (
Right
) measures the existence of investor
protection rules, and the rating of disclosure standards (
Disclosure
) measures the existence
and quality of the information that is available to outside investors.
3. Sample and Variables
3.1 Sample
The sample in this study comprises firms from nine East Asian economies and
thirteen European countries based on the 2000 edition of the WorldScope database. In the
original database there are 77,893 firm-years from these twenty-two countries in the
period 1987 to 2000. However, there is ownership data for only 51,876 firm-years.
3
The
sample selection process is described in Panel A of Appendix A. We exclude financial
firms (those with a primary SIC code of between 6000-6999); firms with missing dividend
data (which is needed to compute the 3-year average dividend); firms with missing,
negative, or zero operating cash flow, sales, market value, or net income data (to compute
dividend payout ratios); firms with missing data that are needed for the construction of the
control variables that are used in the regression; and firms with dividends that are greater
than sales. To reduce outliers, we also exclude firms with industry-adjusted average
dividend payout ratios that are in the top and bottom 1%. Our final sample contains 6,093
3 The data for the ownership variables are obtained from the databases of Claessens et al. (2000) for
East Asia and Faccio and Lang (2002) for Western Europe, which refer to 1996 for East Asia and
France, Germany, Italy, Switzerland and the U.K. We do not extend the sample period beyond 2000 to
avoid the measure error due to the possible changes in ownership after 2000.