This is a lengthy post so let me preface with the executive summary, after which I will provide more detail:
- US and Canada have an existing tax treaty that imposes specific rates for investment income earned by Canadian residents from US sources.
- FATCA places a new condition on receiving those rates.
- This restricts the benefits of the treaty, which is treated as a treaty override by the terms of the treaty itself.
- The treaty provides that the remedy for such an override is a change to the terms of the treaty.
- the IGAs do not change the terms of the treaty but purport to interpret it to allow a different new condition (which condition is itself an override of the domestic FATCA statute) to take effect immediately.
And now for the detail.
Canada and the US have a tax treaty in force in which each government agrees to impose specified tax rates on domestic-income received by investors in the other country. For example, a Canadian person (individual or entity) that invests in the stock of a US corporation and receives dividends on that stock would be subject to a maximum rate of 15% US withholding tax on that dividend under the treaty (see Art. 10); for royalties, the maximum rate would be 10%, and for interest and most capital gains, no tax would be withheld by the US (see Art. 11 and 13(4)). In most cases, a tax treaty overrides domestic statutory law that would impose a higher source-based tax rate on payments made to foreign persons. Accordingly, the statutory US rate on a Canadian resident receiving passive income from US sources would be 30% (with several exceptions, see s. 871). The agreement undertaken in tax treaties is that the US will not impose that statutory rate on payments to Canadian residents, but will restrict its tax to 15% (dividends), 10% (royalties), or even 0 (most interest and capital gains).
Every tax treaty also includes information exchange provisions under which each country agrees to "exchange such information as may be relevant for carrying out the provisions of this Convention or of the domestic laws of the Contracting States concerning taxes to which this Convention applies insofar as the taxation thereunder is not contrary to this Convention." In the Canada-US treaty that is Article 27.
Although this is not critical to the override argument, it is worth noting that the various provisions of the treaty are not conditional on each other; that is, Canadian residents are entitled to the specified tax rate even if, for example, the countries get into a tangle over their information exchange efforts.
FATCA's effect is to impose a new condition on the treaty-based withholding tax rate. That is, under FATCA, the only way for resident Canadian institutions to continue to get the treaty rate (of 0, 10, or 15%, depending on the type of income in question) is to fulfill FATCA information gathering and reporting requirements. If they do not fulfill these requirements, they will not be eligible for the treaty rate, but rather they will be subject to a 30% withholding rate on all "withholdable payments"--an expansive concept of US-source income items which you can read in the statute.
FATCA is thus a new condition on the treaty rate, a condition that is not by any stretch included or even contemplated in the treaty. Indeed, how could FATCA be contemplated by any treaty that came into force prior to 2010, as FATCA did not exist as law before then. This is not to say that no conditions can be placed on the access of Canadian residents to treaty rates. There are many existing conditions for treaty benefits--see particularly the limitation on benefits clause (Art 29A), which are quite expansive and form a major part of any treaty negotiation with the US. But it is to say that FATCA's particular condition is not in the treaty.
Therefore FATCA overrides the existing treaty by unilaterally denying the treaty rate to Canadian residents who would otherwise qualify therefore under the existing, duly negotiated, treaty provisions currently in force.
Now let us look at Art 29(7), which tells us how the countries are supposed to deal with potential tax treaty overrides that arise when one country enacts a domestic law that conflicts with the treaty in effect:
"Where domestic legislation enacted by a Contracting State unilaterally removes or significantly limits any material benefit otherwise provided by the Convention, the appropriate authorities shall promptly consult for the purpose of considering an appropriate change to the Convention."Thus the remedy to a law that would restrict or remove a material benefit--namely, a specified tax rate on a payment of US-source income to a Canadian investor--is a change to the convention.
A change to an existing convention is undertaken in a protocol. A protocol is in legal terms nothing less than a new treaty that overrides specific provisions of the existing treaty to reflect the parties' later agreement. That is, to change a treaty, each government must agree to the change via a new treaty, which each government must ratify under its internal treaty-making processes.
This therefore suggests that the proposed intergovernmental agreements (IGAs) are not a valid means to get FATCA to work as a matter of law. This is because the US is not treating IGAs as treaties at all; it is treating them as interpretations of the existing treaty, specifically, the information exchange provision. You can read this setup in the preamble to these agreements. This position seems plainly incorrect, but the subject of the legal status of the IGAs is its own complicated analysis, and I will post more on that subject very soon.
I will note, however, that the IGAs further muddy the interpretive waters since what they do is in fact override the terms of the FATCA statute, by switching the reporting relationship from Canadian resident institutions to a government-to-government relationship. Some have argued that they do not override but merely use Treasury's mandate to define exemptions to FATCA. Perhaps, but Treasury was not given any mandate by Congress to encase those exemptions in international agreements. Moreover, it seems a real stretch to assert that the IGAs simply interpret existing information exchange provisions, especially when it is clear that many or most countries will have to enact domestic legislation to fulfill the new reporting requirements. The valid way forward for Treasury would have been to create straightforward conditional exemptions: exempt countries from FATCA provided those countries enacted laws according to Treasury specifications. That would still be an extra-territorial reach, perhaps, but there are precedents for the mechanism (such as what was done to shut down bearer bonds--thank you to Michael Schler for reminding me of that example, and I know that there are others as well). But, importantly, this established way forward would not solve the tax treaty override problem. Therein may lie a main reason for going the IGA route, even though it is not a clearly valid resolution.
What is clear at this stage is that FATCA overrides the existing tax treaty by significantly limiting a material benefit thereunder, and the only valid way to fix that override is to change the treaty itself, by entering into a new protocol. We can see that the US and Canada have a lot of experience with protocols: protocols to the current tax treaty were signed on June 14, 1983, March 28, 1984, March 17, 1995, July 29, 1997 and September 21, 2007. One of those was to fix another infamous US tax treaty override: the branch profits tax. In short, it doesn't seem to have been that big deal in the past to agree to changes to the treaty terms by the normal treaty-making process, even when the issue was unilateral override, which is typically seen as a bit of bad faith in international relations. That raises the question why this particular change is not being marshaled through the same process. I will leave it to the reader to speculate for now, and will have more on this later as well.