Why §909 Was Enacted (Whether We Like It or Not)

By Kimberly S. Blanchard, Esq.

Weil, Gotshal & Manges LLP, New York, NY 

In the almost 15 years that have elapsed since the check-the-box (CTB) rules were promulgated, there has been a great deal of confusion over the proper application of the so-called "technical taxpayer" rule in Regs. §1.901-2(f). But the rule was confusing long before adoption of CTB. It grew out of the Biddle1  case, decided in 1938, and was a direct reaction to the Abbot2  case, decided in 1958. Taken together with some even more confusing rulings and decisions involving Brazilian withholding taxes (which, you will be glad to know, I do not intend to cover here), the whole area of law was very difficult to comprehend – and still is.

The thesis of this short piece is that §909 was enacted because the IRS has for too long failed to distinguish two completely distinct classes of foreign tax credit (FTC) problems.  The first is the case where a tax is technically paid by one person, but both countries agree who the economic owner of the income is.  The second is the case where the two countries disagree over who the owner of the income is.

Biddle was the first type of case, and involved the U.K. ACT system. Under that system, corporate profits are taxed only once. The essential question, for U.S. FTC purposes, is "To whom are such profits taxed?" As proof that this is not an easy question to answer, the Supreme Court had to step in to resolve a split in the circuits. The Court held that it was the corporation, and not its shareholders, that paid the tax. The reasoning of the Court was, essentially, that it was the corporation, and not its shareholders, that had legal liability to pay the tax. The Court did not care who bore the economic burden of the foreign tax; perhaps it realized (as the IRS certainly did) that it would be too difficult to prove that fact in many cases, and opted for a more administrable rule. Biddle and the technical taxpayer rule stand for the proposition that we give effect to the foreign country's choice as to which person to tax, at least as long as it doesn't offend our notions of who earned the related income.

The important point for our purposes here is that in Biddle no one in the United Kingdom or in the United States was confused about who earned the income in question – the corporation earned income and declared a dividend to its shareholders out of its after-tax income. The United Kingdom simply taxed a different person than we would have taxed.

Abbot was the second type of case.  It involved an Argentine business entity, owned by U.S. persons, that Argentina treated in a manner similar to a partnership but that was treated for U.S. tax purposes as an opaque corporation – a "natural reverse hybrid." The U.S. owners sought to claim an FTC for the Argentine tax they paid, but the courts denied the claim, holding that the tax should be treated as having been paid by the entity that the United States viewed as the taxpayer and earner of the related income.

In hindsight, the most surprising aspect of the Abbot case was that, after winning in court, the IRS reversed its position and decided that the tax should have been treated as paid by the U.S. owners after all. The current technical taxpayer regulations, issued in 1983, represent an explicit overruling of the IRS's own victory in Abbot. The IRS apparently believed that the result in Abbot was inconsistent with the holding of Biddle. As this piece attempts to demonstrate, it was not. The Abbot case was a case in which the two countries, Argentina and the United States, disagreed as to who earned the income. That is an entirely distinct problem.

Into this mess stepped the Federal Circuit Court of Appeals with its decision in Guardian,3  a post-CTB case. In Guardian, a Luxembourg company, as the common parent of a Luxembourg consolidated tax group, was liable to pay the tax for the group as a whole. The Luxembourg company, which was owned by a U.S. person, was treated as a disregarded entity for U.S. tax purposes. Thus, the tax it paid was treated as paid by its U.S. shareholder, giving rise, under a literal reading of the technical taxpayer rule, to a direct §901 FTC. The court held in favor of the taxpayer and allowed the FTC.

Guardian, like Biddle, is a true technical taxpayer case. The fact that the Luxembourg parent was a regular hybrid entity had little to do with the issue presented in the case. Recall that in Guardian, all that Luxembourg was doing was asking the common parent of a consolidated group to pay the taxes of all the group members. There is nothing odd or surprising about this – we have the same rule for U.S. groups. In Guardian, no one from any country thought that what Luxembourg was doing was assigning taxable income to a person who didn't earn it. Everyone agreed as to who earned the income. Luxembourg knew, as well as we did, that the income that related to the tax was earned by other members of the group.

During the period of time it was litigating Guardian, the IRS proposed its new regulations under Regs. §1.901-2(f) that would reverse the result reached in Guardian and re-instate the result in Abbot.4  But fixing the Guardian problem was (and remains) easy, fixing the Abbot problem is hard, and the two problems are entirely unrelated. The IRS's principal mistake in proposing these regulations was to believe that one problem had anything to do with the other.

Because no one disagrees who the economic owner of the income is, all that is needed to fix the Guardian problem is to treat the tax paid by the common parent as allocable to all the members in accordance with the income each earned. This is, of course, exactly what the proposed regulations did. But you can't "fix" the reverse hybrid problem in any analogous way. In a reverse hybrid structure, one country thinks that the income is earned by the "partners," and imposes its tax on them. But we in the United States think the income is earned by the entity, because it is a corporation for our tax purposes. So the two countries have a different view of who the economic owner of the income is. It's an on-off switch, not an allocation problem.

The 2006 proposed regulations proposed to "solve" the reverse hybrid problem by pushing the tax down into the hybrid, where the IRS thought the income resided. But the proposed regulation's result-oriented rule worked for the IRS only where the reverse hybrid was owned by a single U.S. shareholder. Where the reverse hybrid was a partnership with a foreign partner, the construct adopted by the proposed regulations offered an attractive nuisance to taxpayers.  It effectively shifted tax from the hands of the U.S. partner, where the foreign country believes the tax should reside, to the entity owned in part by a non-U.S. partner. Because the entity is now entitled to claim a U.S. FTC, the non-U.S. partner has, in effect, a "double dip" that can be exploited in negotiating the economic deal between the partners.

I think the drafters of the proposed regulations saw this, which is why the regulations were never finalized. I think they realized they had to get a different kind of authority to deal with the reverse hybrid problem. What they came up with was §909.  Instead of pushing taxes or income around, §909 merely suspends play until the income catches up with the tax.5 

The proposed regulations made a similar mistake in the now-infamous Example 3, the repo case. A repo is a classic case of two countries seeing two different persons as the owner of income. No one is asking someone to pay someone else's tax, as was the case in Guardian and Biddle, or in any withholding tax case. The countries simply have different views of who earns the income, and obviously that affects who they think should be paying tax on that income. The proposed regulations accorded meaning to the source country's view of the transaction, which was that the borrower (in our terms) actually owned the repo'd assets and was liable for the tax. I think the IRS believed, incorrectly, that it was compelled by Biddle to do this. But again - this is not what Biddle stands for. Biddle did not involve a case in which the two countries disagreed about who the economic owner of the income was.6 

So, in summary, the problem with the proposed regulations was their failure to distinguish the case where a tax is technically paid by one person (e.g., the common parent in Guardian) from the case where a foreign country actually thought that person was the economic owner of the related income (Abbot and Example 3). The first problem is a technical taxpayer problem and therefore can be fixed using technical taxpayer rules. The second problem is not a technical taxpayer problem at all. It is an arbitrage problem.

It will be interesting to see whether the IRS comes to the realization that they didn't need §909 to deal with Guardian and the technical taxpayer issue it presents, and that they can simply finalize the proposed regulations to deal with that case. If they do so, §909 could become a special rule for reverse hybrids and other arbitrage issues. The question that should then be asked is whether it's worth the candle.

This commentary also will appear in the May 2011 issue of BNA's Tax Management International Journal. For more information, in BNA's Tax Management Portfolios, see Liebman, 901 T.M., The Creditability of Foreign Taxes — General Issues, and in Tax Practice Series, see ¶7150, U.S. Persons — Worldwide Taxation.


 1 Biddle v. Comr., 302 U.S. 573, 58 S.Ct. 379 (1938).  

 2 Abbot Lab. International Co., 160 F. Supp. 321 (N.D. Ill. 1958), aff'd, 267 F.2d 940 (7th Cir. 1959).  

 3 Guardian Industries Corp. v. U.S., 477 F.3d 1368 (Fed. Cir. 2007).  

 4 Prop. Regs. §1.901-2, REG-124152-06, Aug. 4, 2006. 

 5 Much ink has been spilled, and much more will be spilled, over how to apply a tracing principle that seeks to match foreign tax expense to "related" income. As the drafters of the §864(e) interest allocation rules are fond of reminding us, money is fungible. Outside of the simplest diagrams, tracing is not possible.  

 6 Unlike reverse hybrids, where at least in theory the related income can "catch up" to the foreign taxes upon a distribution out of the reverse hybrid, in a repo case there will never be any "catch up." All that will happen will be that the "loan" will be repaid (or, in the other country's parlance, the owner will sell the bond back to the original seller). That will not create any match, because the underlying income has already been earned and taxed, and doesn't reverse itself on the repayment. This is why a repo can't be a splitter, and shouldn't be confused with one.