Running a Canadian Company and a US Company: When the Two-Entity Structure Is Worth It, How Money Moves Between Them, and the Annual File That Keeps It Legal

The two-company structure — Canadian corporation plus US corporation under common ownership — is the natural adult form of a genuinely two-country business, and the mistake is building it too early or running it too casually. When it's warranted: the expansion playbook's triggers (US employees, leases, enterprise customers wanting a domestic counterparty) plus the liability logic (each country's operations walled from the other's lawsuits), payroll cleanliness (each entity employs its own country's staff natively), financing and exit fit (US lenders and acquirers dealing with a US entity), and — for Canadian-owned groups — the repatriation architecture, where the US subsidiary's active earnings come home to a Canadian corporate parent as exempt-surplus dividends (no further Canadian corporate tax) after the treaty's 5% US withholding at the corporate ownership threshold. When it isn't: the pre-triggers business for whom the treaty-plus-protective-filing posture and the nexus program already answer everything — the second entity founded on optics alone buys 5472s and trapped losses for nothing. Ownership design is decided at formation because it's expensive to change: the Canadian parent holding the US subsidiary is the default for Canadian owners (the exempt-surplus and 5% pipeline runs through it; consolidated exits sell cleanly); sister companies under a common individual owner arise where history or US-owner facts drove it, and forego the corporate-dividend pipeline (money moves between siblings only through priced transactions or the shareholder level); and US-citizen owners layer the CFC/GILTI analysis over whichever shape — the Canadian entity in an American's structure carrying the 5471-and-GILTI file whatever the US entity does. Running the pair is the intercompany discipline from the transfer-pricing playbook operating as a system: the agreements (distribution, services, cost-sharing, loans) signed before the flows; arm's length pricing with the annual memo; monthly invoicing and settlement so balances don't silt into undocumented loans; the twin disclosures (T106 north, 5472 south) filed from one workpaper; and the governance hygiene — separate boards and minutes, separate banking, no commingling — that keeps the liability wall real when it's finally needed. Money then moves home through a short menu, each channel with its character: intercompany service and product payments (deductible where paid, income where received — the operating channel, transfer-priced); interest on documented loans (deductible against US income within the US interest-limitation rules, treaty-rate withholding, income to the Canadian lender); royalties where genuine IP sits north (the scrutiny-heavy channel, implemented only with substance behind it); and dividends — the subsidiary's after-tax profit to the Canadian parent at 5% withholding into exempt surplus, the channel that makes the parent-sub shape pay for itself. The annual rhythm binds it: the pricing memo and true-up, the twin filings, the intercompany balance review, the dividend decision against both countries' positions, and — every few years — the structure question re-asked, because businesses outgrow their architecture in both directions and the pair that was right at five employees may want a third entity, or one fewer, at fifty.

Key takeaways

• Build on triggers, not optics: employees, premises, enterprise contracts, financing, and liability walls justify the pair; ambition alone buys compliance without benefits. The pre-trigger business already has a complete answer in the treaty posture.
• Ownership shape is a formation decision: Canadian parent → US sub unlocks the 5%-withholding exempt-surplus dividend pipeline and clean exits; sister companies forego it; American owners carry the GILTI file over either — model the shapes before the first filing, because reorganizing later is a project.
• The intercompany system is non-optional: agreements before flows, arm's length prices with the annual memo, monthly settlement, T106/5472 from one workpaper — the transfer-pricing starter file operating as the group's circulatory system.
• Repatriation is a designed menu: operating payments, interest, royalties-with-substance, and dividends each carry withholding, deductibility, and scrutiny profiles — the annual dividend decision is made against the surplus accounts and both countries' rates, not against the bank balance.
• The liability wall needs governance to exist: separate minutes, banking, and contracting — the corporate-separateness habits that cost an hour a quarter and decide whether the wall holds when a US plaintiff tests it.
• Re-ask the structure question on schedule: growth adds entities (the IP company, the third country) and contraction removes them; the two-company pair is a stage, and the annual review names the triggers for the next one in both directions.

The group's annual file

One binder, five tabs, every year: the intercompany agreements as amended, with this year's pricing memo and true-up; the twin information returns (T106, 5472) and the workpaper reconciling them to both sets of financials; the loan and balance schedule with interest evidenced; the dividend/repatriation memo — surplus computation, withholding applied, the Canadian-side exempt-surplus treatment documented; and the governance minute book entries proving two companies actually behaved like two companies. The binder is what turns every future event — the CRA screening letter, the state audit, the buyer's diligence, the lawsuit testing the wall — into a retrieval exercise. Groups that keep it spend a few thousand a year on the discipline; groups that don't spend multiples reconstructing it for whichever examiner asked first.

Worked example

An Ottawa cybersecurity firm, five years in: the Canadian corporation holds the IP and the engineering team; a Delaware subsidiary (formed at the first US enterprise deal, per the triggers) employs eight US staff and holds the US customer contracts. The system: a distribution agreement prices the US entity's license revenue share; a services agreement charges northbound engineering support at cost-plus; a documented loan funded the US build-out at a defensible rate; monthly settlement keeps the balance inside terms. The annual file: pricing memo trued up against actuals; T106 and 5472 filed from the shared workpaper; the dividend decision — the US sub's US$900,000 of after-tax earnings — pays US$600,000 home at 5% withholding into the parent's exempt surplus (no further Canadian corporate tax), with US$300,000 retained for US working capital, the memo documenting both; board minutes for each entity record their own approvals. Year six brings the tests the file was built for: a CRA T106-profile query closes in three weeks by correspondence; a US customer's lawsuit against the subsidiary never reaches the Canadian parent's assets — the governance tab is the wall's evidence; and a strategic acquirer's diligence on the group takes the binder as its tax workstream's starting point, which is how a discipline that cost four figures a year quietly added a multiple's worth of credibility to the exit conversation.

Official sources

"The business profits of a resident of a Contracting State shall be taxable only in that State unless the resident carries on business in the other Contracting State through a permanent establishment situated therein." — Canada-United States Tax Convention, Article VII, https://www.canada.ca/en/department-finance/programs/tax-policy/tax-treaties/country/united-states-america-convention-consolidated-1980-1983-1984-1995-1997.html

"A 25% foreign-owned U.S. corporation (including a foreign-owned U.S. disregarded entity (DE)), or a foreign corporation engaged in a trade or business within the United States" files Form 5472; a foreign-owned U.S. DE is "classified as a corporation" for this purpose. "A penalty of $25,000 will be assessed on any reporting corporation that fails to file Form 5472 when due." — Internal Revenue Service, Instructions for Form 5472, https://www.irs.gov/instructions/i5472

Practitioner note

Two-company groups succeed as systems and fail as improvisations: the entities are easy to form and the discipline — agreements, pricing, settlement, twin filings, governance — is what makes them worth having. Our group clients run the five-tab annual binder, and the pattern is consistent enough to state as law: every examiner, lender, and buyer who ever looks at a cross-border pair asks for the same binder, and the only variable is whether it exists before they ask.

See also: For the best US entity for a Canadian owner, see the best US entity for a Canadian owner; and browse every cross-border tax topic guide, organized by situation.

Next step

Fairlight prepares the two-entity group program — structure and ownership design against the repatriation pipeline, the intercompany system build, the annual binder with twin filings and the dividend memo, and the scheduled structure reviews as the business grows. See cross-border pricing or book a call.