As promised in my prior post, here is the text of my talk at NYU today regarding Reuven Avi-Yonah's co-authored article, The Case for Dividend Deduction. (See prior post for my link to Reuven's article.)
The text refers to a chart I had distributed to attendees, demonstrating that allowing corporations to deduct dividends paid is potentially identical to the seemingly very different system where they are taxed, but dividend distributions to shareholders get the benefit of "imputation" (i.e., corporate-level tax is in effect treated as an advance payment by the shareholders of their taxes on the corporate income). Imputation is or at least was a very common corporate integration system around the world, whereas dividend deductibility is not. I argue that the paper over-distinguishes between the two and that they can be made identical. Anyway, the chart, which offers a simple illustration of the potential equivalence between the two methods, is available here.
Perhaps the most novel point in my talk is one that I under-developed in the text because it would have taken too long to explain it. So here goes. Reuven argues that dividend deduction would be more effective than imputation in encouraging the managers to pay dividends, on the ground they often don't especially care about shareholder taxes but love to get company-level deductions. Thus, even if the two methods of dividend deduction and imputation are in fact economically equivalent, the former will in fact induce greater payouts.
One could certainly challenge this view on multiple grounds, but the one I emphasized, in particular because I thought it was more of a new point, reflects accepting the basic premise (at least arguendo) but then asking what the managers really care about. A common answer, with considerable real world empirical support, would be that they appear to care more about financial accounting income than about income tax liability. So the entity-level tax benefit of dividend deductibility won't affect their behavior as strongly as Reuven anticipates unless there is an accounting benefit. But would there be?
I am not an accountant, but I've played in or near their waters often enough to realize that this is a trickier question than it may initially seem from a lawyer or economist standpoint.
Presumably the financial accounting rules would NOT be revised to allow dividend deductions against financial accounting income, even if newly made deductible against taxable income. But wouldn't financial accounting income reflect the benefit of reducing federal income tax liability through dividend payouts?
Not necessarily. An initial point to keep in mind is that financial accounting often ignores the mere deferral of federal income tax liability. In principle, under dividend deduction, the ultimate corporate tax is zero as all earnings get paid out (or at least an amount equal to taxable income, which is often less than the tax measure of earnings for dividend purposes). So it would seem that the accounting rules in the dividend deduction scenario should either (a) ignore federal taxes on the ground that they're merely temporary, at least until they escape the possibility of being reversed through net operating losses created by dividend deductions, or at least (b) ignore the difference between paying out deductible dividends this year or next year. I'm not in fact sure how it would all end up playing out, but we should recognize that (a) there would be a tricky issue for the accountants to work out and (b) there wouldn't necessarily be a straight accounting benefit for the tax liability effect.
By analogy, consider the accounting rules for the foreign earnings of U.S. companies' foreign subsidiaries. These get the U.S. tax benefit of deferral - that is, they aren't subject to U.S. tax until they actually are repatriated for tax purposes. But companies get no accounting benefit from deferral - they are treated as if the U.S. repatriation taxes were being fully paid on a current basis - unless they solemnly declare to their accountants that the funds are being "permanently" reinvested abroad. Once this happens, the potential future U.S. repatriation taxes are discounted by 100% (i.e., to zero) rather than by zero percent.
Anyway, if the timing of repatriation is effectively ignored in financial accounting on the view that it doesn't matter exactly WHEN it happens (despite the potential effect on the present value of U.S. tax liability, then the same idea might apply, albeit in a somewhat different and (at least to me) unpredictable fashion, with regard to the effect of current versus future dividend payouts on entity-level, purely domestic U.S. income tax liability.
Any accountants out there, your thoughts on this admittedly esoteric issue (either in the comments page here or by e-mail to me) would be of potential interest.
UPDATE: The following is hoisted from the first comment on this blog entry (by Elijah):
"As you suggest (directionally at least), corporations would presumably book a deferred tax asset for undistributed dividends, much in the same manner as they currently book a deferred tax liability for the (non-permanently reinvested) unrepatriated earnings of their foreign subsidiaries. In this manner, they would "ignore" the actual timing of the (tax) deduction and the tax rate (on U.S. earnings) would move towards zero.
"I say 'move towards' because capital needs will prevent corporations from ever truly distributing everything prior to liquidation, which itself may be too remote a possibility for the corporation to consider for financial accounting purposes. Rather, corporations would undoubtedly get into the usual arguments (with their auditors) about how much of the the deferred tax asset could really be realized, and whether some amount of offsetting valuation allowance would be appropriate. This would ultimately be reflected in the (book) tax rate.
"The more interesting question, I think, is whether any of this would influence managers to pay dividends. My initial thought is that it would not. Managers would, theoretically at least, not be able to influence the book tax rate (year to year) by paying dividends in year 1 versus year 2 (or 3, or 4, etc.)."