Skip to content
MONTCLARE
CAPITAL PARTNERS
CONTACT
Real Estate Asset ManagementTransfer PricingInternational Tax StructuringFinancingInternational DesksLeadershipPrivate PublicationsContact
North America

A Canadian ULC and the Treaty Denial Problem

Alfonso Martínez RuizFounder and Chief Executive Officer, Montclare Capital Partners · Published August 2026 · Reviewed September 2026

The unlimited liability company occupies a strange position in cross-border planning. For years it was the standard answer whenever a United States group needed a Canadian operating entity: a real corporation under Canadian law, resident and taxable in Canada, and at the same time capable of being disregarded or treated as a partnership for United States federal tax purposes. One entity, two characterizations, and losses usable where the group wanted them.

That arrangement was addressed directly by the protocol signed on 21 September 2007, which inserted a new paragraph 7 into Article IV of the convention between Canada and the United States. The provision is short, written in the abstract, and quoted incorrectly more often than almost any other clause in the treaty network. What it does is remove treaty benefits from payments that cross a characterization mismatch, and its effect on an ordinary distribution out of a ULC is not marginal.

What a ULC actually is

A ULC is incorporated under provincial legislation, in Nova Scotia, Alberta or British Columbia. Its shareholders lack the limited liability a shareholder of an ordinary corporation enjoys, which is the source of the name and the reason the vehicle is normally held through an intermediate entity.

For Canadian tax purposes a ULC is a corporation like any other. It is resident in Canada, it pays tax under the Income Tax Act, and what it distributes to a non-resident shareholder is a dividend under subsection 212(2).

For United States federal tax purposes it is an eligible entity, and the owner may elect to have it treated as disregarded or as a partnership. The attraction was that the Canadian operating results flowed into the United States return without an intermediate corporation while the entity remained a taxpaying corporation in Canada. That mismatch was the point. It is now the problem.

The mismatch, stated in one sentence

Canada sees a corporation paying a dividend to its shareholder. The United States sees nothing, because the payer does not exist as a separate entity for its purposes and a payment from a disregarded entity to its owner is not a payment.

Everything that follows is an elaboration of that sentence.

Article IV(7)(b), read as it is written

Paragraph 7 of Article IV provides that an amount of income, profit or gain shall be considered not to be paid to or derived by a person who is a resident of a Contracting State in two situations. Subparagraph (b) is the one that catches the ULC.

Under it, the amount is treated as not paid where the person is considered under the taxation law of the other Contracting State to have received the amount from an entity that is a resident of that other State, but, by reason of the entity being treated as fiscally transparent under the laws of the first State, the treatment of the amount under the taxation law of that State is not the same as its treatment would be if that entity were not treated as fiscally transparent under the laws of that State.

Three conditions, therefore. The source state must view the recipient as receiving the amount from an entity resident in the source state. The residence state must treat that entity as fiscally transparent. And the transparency must change the characterization of the amount in the residence state. Where all three hold, the amount is deemed not to be paid to or derived by that person, so for that person there is no dividend, no interest and no royalty to which any distributive article can attach. The fiction runs against the recipient, not the payment, which is why changing the recipient can change the answer.

The provision is not a limitation on benefits rule, and it is not an anti-abuse rule requiring a purpose. It operates mechanically on the mismatch, whatever the commercial reason for the structure.

The drafters’ own example

The Technical Explanation published by the United States Treasury Department alongside the protocol works the point through. It posits USCo, resident in the United States and sole owner of CanCo, an entity treated under Canadian law as a corporation resident in Canada and under United States law as disregarded, and assumes USCo is considered to have received a dividend from CanCo.

Canada, as source state, views USCo as receiving a dividend from a corporation resident in Canada; CanCo is fiscally transparent under United States law; and because CanCo is disregarded the payment is disregarded there, whereas if that law regarded CanCo as a corporation it would be a dividend. Subparagraph (b) therefore applies.

It then adds the sentence that decides most files: the same result obtains where USCo is considered under Canadian law to have received interest or a royalty instead of a dividend, the payment being equally disregarded in the United States while it would be interest or a royalty if CanCo were regarded. The denial is not confined to distributions on shares. It reaches the whole intragroup payment stream.

What it costs

Subsection 212(2) of the Income Tax Act imposes tax at 25 per cent on dividends paid or credited by a corporation resident in Canada to a non-resident. Article X(2) would reduce that to 5 per cent where the beneficial owner is a company owning at least 10 per cent of the voting stock of the payer, and to 15 per cent otherwise. Where Article IV(7)(b) applies there is no dividend for treaty purposes and no reduction. The rate is 25 per cent.

On interest the arithmetic is worse. Article XI(1) provides that interest arising in a Contracting State and beneficially owned by a resident of the other may be taxed only in that other State, so the treaty rate on interest from a Canadian payer to a United States lender is nil. Subsection 212(1) imposes 25 per cent. A non-arm's length loan from the United States parent to its ULC falls inside paragraph 212(1)(b), which reaches interest paid to a person with whom the payer is not dealing at arm’s length, so losing the treaty moves the withholding from nothing to a quarter of every payment. A management or administration fee moves the same distance, from nothing under Article VII absent a permanent establishment to the 25 per cent of paragraph 212(1)(a).

On royalties the swing is neither uniform nor twenty-five. It is the domestic rate less the treaty rate, and the steady side is the domestic one: subsection 212(1) charges at 25 per cent whenever it charges at all, and subsection 212(5) at the same rate. What moves is the treaty side, between nil and ten, so there are three answers, not two. Twenty-five where the Act charges and Article XII(3) sends the treaty rate to nil. Fifteen where it charges and Article XII(2) caps that rate at 10 per cent. And nothing where the Act does not charge, there being no Canadian tax for the treaty to relieve.

Which of the three applies turns on the structure of paragraph 212(1)(d), which is asymmetric. The paragraph charges a rent, royalty or similar payment, including, but not so as to restrict the generality of the foregoing, subparagraphs (i) to (v), then says but not including before subparagraphs (vi) to (xii). The first group illustrates a charge the general words already impose; the second is carved out of it, the Act repeating elsewhere that the phrase does not include any payment described in subparagraphs (1)(d)(vi) to (xii). Only an exclusion gives a zero.

Subparagraph (vi) is an exclusion, covering a royalty in respect of a copyright in respect of the production or reproduction of a literary, dramatic, musical or artistic work. A copyright royalty of that kind, and a software licence to reproduce, are outside the domestic charge, and Article XII(3) puts them at nil too: nothing moves, and nothing is lost.

Subparagraph (ii) is an inclusion, and this is the trap. It reaches information concerning industrial, commercial or scientific experience where the consideration depends on the use made of it, on production or sales, or on profits, but a fee that fails it stays in the paragraph, falling back on subparagraph (i), which reaches a payment for the use of or for the right to use in Canada any property, invention, trade-name, patent, trademark, design or model, plan, secret formula, process or other thing whatever, with no contingency required. Fixed-fee know-how therefore loses twenty-five, like the contingent kind, as do a patent royalty and a software licence to use rather than reproduce. The treaty nil here stops short of information provided in connection with a rental or franchise agreement.

Fifteen is the residue. Article XII(4) defines royalties by a closed enumeration reaching trade mark, design, model and plan, the use of tangible personal property, and copyright in a scientific work, while Article XII(3) exempts only four lettered cases, so whatever the definition reaches and the list does not stays at Article XII(2)’s 10 per cent against a domestic 25. Film and television are there, having left the paragraph by subparagraph (xii), another exclusion, and being charged instead by subsection 212(5), at 25 per cent on a right in or to the use of a motion picture film or a film or video tape for television, since Article XII(3)(a) repeats subparagraph (vi) but narrows it by excluding them. So is the know-how under a rental or franchise agreement just mentioned, and a copyright payment not for production or reproduction.

None of that depends on the amounts being large or the arrangement being aggressive. It follows from a characterization election often made years earlier for unrelated reasons. Paragraph 7 has applied since the date fixed by Article 27(3)(b) of the protocol, the first day of the third calendar year that ends after the protocol enters into force. The protocol entered into force on 15 December 2008, so the count begins with a calendar year with a fortnight left in it and paragraph 7 has had effect since 1 January 2010. The rule is not new, and a structure that has been paying at the reduced rate throughout is not protected by time.

The mirror case in subparagraph (a)

Subparagraph (a) runs in the opposite direction. It applies where the source state views the person as deriving the amount through an entity not resident in the residence state, and, the entity not being fiscally transparent there, the treatment differs from what it would be if derived directly.

The Technical Explanation illustrates it with a United States company that part owns a Canadian entity treated as fiscally transparent in Canada but as a corporation in the United States, receiving a Canadian dividend: Canada sees the company deriving the dividend through the entity, the United States sees the entity deriving it, and the mismatch denies the benefit. The Explanation adds that United States source income received by an entity organized in the United States that Canada treats as a corporation is not considered derived by its Canadian shareholders, even if United States law treats that entity as transparent.

That second limb speaks of shareholders resident in Canada, not of a European parent, and could not do otherwise: Article I applies the convention to residents of one or both Contracting States, so a Dutch holding is outside its personal scope. That is why a United States limited liability company inserted between a European parent and a North American operating entity so often produces results nobody intended.

The paid-up capital route, and what it does not solve

The practical response developed in the market turns on the mechanics of subparagraph (b), not on avoiding it. Subsection 84(1) of the Income Tax Act deems a corporation resident in Canada that increases the paid-up capital of a class of shares, outside the listed exceptions, to have paid a dividend equal to the amount of the increase, and each shareholder to have received a proportionate part of it. A subsequent reduction of paid-up capital returns cash to the shareholder without a further deemed dividend.

The argument is that a deemed dividend on a paid-up capital increase is invisible in the United States whether or not the ULC is regarded, so the third condition in subparagraph (b) is not met and the treaty rate survives. The Canada Revenue Agency has agreed: provided the deemed dividend is disregarded in the United States, and would be disregarded equally if the ULC were not fiscally transparent, Article IV(7)(b) does not apply.

Two cautions. The technique addresses distributions and does nothing for interest, royalties or service fees, where the larger recurring exposure normally sits; that the denial reaches those is the Technical Explanation’s proposition, not the Agency’s, and on debt the Agency pointed the other way: subparagraph (b) does not apply, it said, where the loan is owed instead to the United States grandparent and the interest is treated in its hands as it would be if the ULC were not fiscally transparent, though making no categorical statement about anti-avoidance. And it rests on an administrative statement made in 2011 and since archived, not on the words of the convention, so confirm it against the Agency’s current view before the steps, not after.

Where a ULC still earns its place

None of this makes the vehicle useless. It makes it a vehicle with a known cost, and the question is whether that cost buys what the group needs. Where the Canadian operation is expected to generate losses in its early years and the shareholder can use them, the ULC still earns its keep, though a Canadian branch of the United States company reaches the same losses, as below. Where the Canadian entity retains and reinvests profit rather than distributing it, the withholding question is deferred rather than incurred.

Where the requirement is a Canadian operating company that distributes profit upward, an ordinary corporation is the better answer, and a branch is worth comparing. Section 219 taxes a non-resident corporation’s taxable income earned in Canada at 25 per cent, and Article X(6) caps the additional tax at 5 per cent of earnings not previously subjected to it, with five hundred thousand Canadian dollars deducted in computing them. Those are known numbers with no characterization risk attached.

For groups approaching Canada from Europe rather than through the United States, the point is narrower than it looks. The convention between Canada and the Netherlands, as consolidated by the Department of Finance, contains no provision equivalent to Article IV(7); its residence article addresses dual residence and stops there. Nor does the Multilateral Convention to Implement Tax Treaty Related Measures to Prevent Base Erosion and Profit Shifting supply one, because Canada reserved the entirety of its Article 3, on transparent entities, under Article 3(5)(a). A Dutch parent holding a Canadian subsidiary directly does not meet this rule. It meets it when a United States entity is placed in the chain, precisely the moment when nobody re-reads the Canadian treaty.

This article is informational and does not constitute tax or legal advice. The treatment of any structure depends on its facts and on the law of each jurisdiction involved. Each engagement is subject to scope and applicable regulation.

SPEAK TO US

Thirty minutes, no obligation

If something here applies to your group, the useful next step is usually a conversation rather than more reading. Leave your address and we will come back to you.

We use your address only to reply. Nothing else. See our privacy notice.
← ALL PUBLICATIONS
BEGIN A CONFIDENTIAL CONVERSATION Prefer to talk? Book a 30-minute call