Testing the "new" and traditional views on dividend taxation in an integrated tax setting

The dividend puzzle continues to challenge researchers. Recent accounting research claims support for the "new" (or capitalization) view and suggests that stock prices fully impound investor level dividend taxes. This dissertation examines both the "new" and traditional views on dividend taxation using the Australian and New Zealand settings. These countries adopted integrated tax systems in the late 1980s, effectively eliminating one layer of the traditional double taxation of corporate income. While prior researchers have suggested the results in the US setting reflect capitalization, I find at most limited support for capitalization in the pre-integration period in Australia which parallels the US system. In the post-integration period in Australia, I find that the effect attributed to tax capitalization by prior research persists in both the credit sample (where firms pay dividends with tax credits that offset personal taxes) and the non-credit sample. For the post-integration period in New Zealand, the effect attributed to tax capitalization persists despite the elimination of the second layer of tax on corporate income. As with the Australian setting, the effect does not differ between the credit and non-credit samples. This contradicts the capitalization hypothesis In contrast to the tax capitalization view, the traditional view suggests that firms treat taxes as a cost of dividends and thus increase payout when the tax cost falls. I test this view using a sample of firms which survive and are profitable for the periods 1975-1985 and 1990-2000. The analysis offers moderate support for the traditional view on dividend taxation with some support for a sustained increase in payout ratios following the introduction of integration. The findings from both the analysis of the capitalization view and the traditional view need to be interpreted in conjunction with one another. The results tend to refute capitalization while offering some support for the traditional view. While it would be unrealistic to claim a solution to the dividend puzzle, these results contribute an additional piece of evidence to the overall stock of knowledge. In this respect, they should be interpreted as support for the traditional view relative to the "new" or capitalization view.