Yesterday, at a very lively session, Jennifer Blouin presented her paper, Is U.S. Multinational Intra-firm Dividend Policy Influenced by Reporting Incentives?
Taking on a tricky empirical research question from the design standpoint, the paper finds evidence suggesting that publicly traded U.S. multinational companies (MNCs) with unrepatriated foreign source income respond not only to the possible tax advantages of keeping funds abroad, but to accounting considerations - namely, the managers' desire to report higher rather than lower accounting income.
The issue arises because U.S. accounting rules give U.S. MNCs essentially no recognized benefit on the income line of their financial statements if they avoid current U.S. tax by keeping money in lower-tax jurisdictions abroad, UNLESS they opt to designate the money as "permanently reinvested abroad" (PRE). Where a company chooses the PRE designation, the accounting treatment jumps from (a) assuming full, immediate, and certain U.S. taxation, albeit subject to the ability to claim foreign tax credits, to (b) assuming zero residual U.S. tax.
If reported earnings matter, this is obviously an appealing thing to be able to do. But investors can learn, from looking elsewhere on the companies' financial statements, how much they have by way of unrepatriated foreign earnings. Suppose the PRE designation is pure "cheap talk" that does not itself affect the subsequent likelihood of repatriation. In principle, choosing it would bring no stock price advantage if, with or without it, investors could and did assess for themselves the likelihood of repatriation and the U.S. tax hit that this would entail.
In the world we actually live in, however, where managers like high earnings and the numbers on earnings lines appear to matter more than they seemingly should, the PRE designation is obviously attractive to managers because it permits them to claim higher earnings, albeit at the price of having a negative adjustment later on if they end up bringing some of the money home. (Whereas, if they don't claim PRE, I believe that paying the tax one already said was going to pay has no financial reporting downside, at least on the income line.)
The paper finds that "reporting incentives have a negative effect on the amount of foreign earnings repatriated by MNCs .... [F]inancial reporting is an important factor in the repatriation decision of U.S. MNCs."
Every practicing tax lawyer I have ever spoken with about the role that accounting considerations plsy in companies' tax planning would say "OF COURSE this is true." They are quite aware of operating in a world where, as I once heard an investment banker say" Saving taxes is all very nice, but earnings per share make the world go round."
On the other hand, any economist with a strong University of Chicago-style orientation would likely be extremely reluctant to believe it could possibly be true. "That may work in practice, but it definitely doesn't work in theory" is probably about the strongest concession one could ever hope to get from such an individual.
Some takeaways from accepting the paper's finding (as I do) include the following:
1) As proponents of both worldwide and territorial appraoches to U.S. taxation of resident MNCs can agree, the deferred tax on repatriation of foreign earnings has undesirable incentive effects, inducing companies to engage in costly maneuvers to arrange their global cash as they prefer in the most "tax-efficient" manner. (There are theoretical conditions in which this actually might not be so - a point made by the so-called "new view" of cross-border dividend taxation, but the key precondition, a constant rate of repatriation tax that will at some point inevitably be paid, does not actually hold.) It's ironic, if that's the word for it, that a mere accounting rule, as opposed to the actual substantive rules of taxation, should also create lock-in.
2) There probably is no good accounting rationale for allowing the PRE designation, which is ridiculously discontinuous. Repatriation is treated as a one-or-zero decision rather than a probabilistic continuum as it more likely would be in practice. And managers get one more avenue for gameplaying. Other accounting rules in the neighborhood are also bad, however, such as not present-value-discounting expected future taxes, and the degree of secrecy companies are permitted regarding their U.S. and foreign tax positions and current payments is probably indefensible.
3) In principle, accounting design should take account of the real world consequences (outside accounting) that adopting a particular rule may have. Accountants generally resist this, however, and for a good structural reason. They are concerned that the long-run effects of opening the door to such arguments, which corporate insiders would then make to lobby for bad accounting rules that they hoped would reduce transparency, would be predominantly negative.