Why some
firms distribute generous cash dividends while others are reluctant to do so remains an
unanswered question despite decades of scholarly examination. Although the extant literature on
dividend policy has explored the effects of determinants at the country, industry,
firm, and
firm-year
levels, it remains unclear whether and how much each level of analysis matters to dividend policy.
Consequently, this study seeks to move the literature forward by decomposing the variance at each level
associated with dividend policies in a global sample of 8903
firms over an 11-year time period. We
employ hierarchical linear modeling and
find that all four levels of analysis help to explain dividend
policy, but the
firm and
firm-year effects account for the majority of variance. Furthermore, decomposing
the variance within each year reveals that the
firm level has the strongest effect on dividend policy.
Finally, while the variance in dividend policy explained by each level varies according to the dividend
policy measure used, it is largely stable over our study period. We discuss implications of these
findings
for future research on dividend policy and for the
field of comparative corporate governance.