The Denominator Problem: What CUSMA Is Actually Worth  ·  Part 2 of 4

The Accidental Experiment

Economists almost never get to run experiments on trade agreements. You cannot switch a treaty off for a year to see what it was doing. In March 2025 the United States did something functionally equivalent, by accident, and the results are now sitting in customs data.

Here is the mechanism. On March 4, 2025, tariffs of 25% took effect on goods from Canada and Mexico under IEEPA (the International Emergency Economic Powers Act, a statute allowing the president to regulate economic transactions after declaring a national emergency), with Canadian energy and potash at 10% (White & Case summary of the executive orders). Three days later, amendments exempted goods that qualify as originating under CUSMA (USMCA in US usage) (Federal Register, EO 14232; CBP IEEPA FAQ). On August 1, 2025, the rate on non-compliant Canadian goods rose to 35% (CRS IF12595).

Before March 2025, a Canadian or Mexican exporter comparing the CUSMA rate to the default MFN rate saw a gap averaging about 2 percentage points, since the US trade-weighted average MFN tariff is roughly 2.1% (WTO). After March 7, 2025, the same comparison showed 25 points, and for Canadian goods after August 1, 35. The preference margin, the price of not being certified, went up by an order of magnitude overnight. If the agreement’s paperwork was ever worth doing, it was worth doing now.

Firms responded exactly as a price signal predicts. In December 2024, 35.5% of the value of US imports from Canada and 49.5% from Mexico entered under claimed CUSMA preference. By July 2025: 78.7% and 76.1% (Kelly, Cañas and Torres, Brookings USMCA Forward 2026). The Tax Foundation’s combined series tells the same story at annual resolution: 44% compliant in 2024, 67% in 2025, peaking at 89% in October 2025, and above 80% through the first half of 2026 (Tax Foundation, June 23, 2026); the Penn Wharton Budget Model puts the aggregate at 83.8% as of May 2026 (PWBM, July 13, 2026). None of this trade changed what it was made of. It changed what it filed. The eligible-versus-claimed distinction from Part 1 is the whole explanation: the qualifying trade was there all along, uncertified because certification was not worth 2 points. At 25 points, it certified within months.

Now the arithmetic the series promised. Realized preference value is preference margin × utilization × trade value: the duties compliant goods would have paid, and did not. Every input below is stated so you can substitute your own.

Mexico. US goods imports from Mexico in 2025: $534.3 billion, up 6.2% from 2024 (US Census). Rate avoided by compliant goods: 25 points, the IEEPA rate that applied to non-compliant Mexican goods all year. Compliant share: take 80%, between the measured 76.1% of July 2025 and the October peak of 89%.

At 75% compliance the figure is $100 billion; at 85%, $114 billion. Call it roughly $100 to $115 billion annualized.

Canada. US goods imports from Canada in 2025: $381.9 billion, down 7.3% (US Census). Canada requires a split, because its export base is not uniform. Energy imports from Canada ran about $111 billion in 2025 (EIA, Today in Energy, July 29, 2026), and energy faced only the 10-point rate, with compliance effectively complete (the Bank of Canada’s assumption was 100% for energy, 95% for everything else; July 2025 MPR). Non-energy imports, about $271 billion, faced 25 points to July 2025 and 35 points from August 1; take compliance at 85%, consistent with EDC’s measured 86% as of September 2025 (EDC, January 12, 2026).

Total: roughly $69 billion annualized at the 25-point rate, roughly $92 billion at the 35-point rate that applied from August. A defensible single number is $70 to $90 billion. (A note for the record: this series’ own planning notes, built from July reporting, put Canada at $55 to $65 billion. Recomputing from Census trade values and measured compliance lands higher. The difference is mostly the compliance share, which kept rising, and the August rate step. This is what showing the assumptions is for.)

The pre-2025 comparison. Run the same formula on 2024: roughly $915 billion in combined imports, 44% claimed preference, and a margin averaging 2.1 points at most, less in practice because nearly half of US tariff lines are MFN duty-free, meaning many goods claiming preference avoided almost nothing. The realized value comes out in the mid-single-digit billions per year. The same insurance policy that, on this arithmetic, was paying out at a rate of $170 to $200 billion a year across the two partners in late 2025 had been paying out perhaps $5 billion a year in 2024.

Two honest caveats on all of this. First, this is a static calculation of border cash flows: it assumes the trade would have happened anyway at the non-compliant rate, which it would not have, so it measures the value of the exemption to the trade that existed, not the economic damage the tariffs would have caused without it. Second, avoided duty is not the same as welfare: who ultimately pockets the avoided 25 points (exporter margins, importer margins, consumer prices) is an incidence question, and Part 3 takes it up.

So the exemption was worth a fortune, and roughly a third more of it, in dollar terms, went to Mexico than to Canada. But the headline numbers hide the real asymmetry, which is about which goods each country needed to protect and whether the protection reached them.

Mexico’s exposed goods were originable. Its exports to the US are dominated by manufactures built inside North American supply chains, exactly the goods rules of origin were written for. When the margin jumped, Mexican shippers could certify their way out, and did: over 85% of Mexico’s exports entered duty-free through CUSMA compliance by late 2025 (Baker Institute, August 2025), and Mexico shipped a record $534 billion north, up 6.2%, becoming the United States’ largest goods supplier in the middle of a trade war (Census; CRS IF11175).

Canada held a different portfolio. Its single largest export category, energy, was barely threatened: 10 points at most, zero after certification, and certification was near-automatic. But the goods that were actually attacked in 2025, steel, aluminum, copper, autos, and lumber, were attacked under Section 232 (a national-security tariff authority under the Trade Expansion Act of 1962), and Section 232 tariffs apply regardless of CUSMA compliance. Steel and aluminum went to 25% in March 2025 and 50% in June; copper to 50% in August; autos to 25% in April, softened for CUSMA-qualifying vehicles only by excluding US content from the taxable value, with CUSMA-qualifying parts exempt; softwood lumber to 10% and wooden furniture and cabinets to 25% in October (CRS IF12595; Federal Register, auto procedures).

The results read like the portfolio, not like the compliance rate. Canadian exports to the US fell 5.8% in 2025 (Statistics Canada, February 19, 2026). Steel exports fell by half; aluminum fell by about half before partially recovering through redirection to Europe; lumber ran about 20% below 2024 (Bank of Canada, April 2026 MPR). Canadian manufacturing shed roughly 32,000 jobs between January 2025 and January 2026 (EDC, June 16, 2026). The US effective tariff rate on Canadian goods, duties actually collected over import value, went from a historical 0.1% to 2.4% for 2025, peaking at 3.9% in September (EDC; EDC, January 2026).

One metaphor, then the cash value of it. Canada held flood insurance, and the flood barely came: the IEEPA tariffs that CUSMA compliance could switch off landed mostly on goods Canada could certify or goods already at the 10-point rate. The fire, Section 232, was excluded from the policy, and the fire is what burned. Literally: the instrument Canada’s compliance rush defended against was suspended for compliant goods, while the instruments that did the measurable damage to steel, aluminum, autos, and lumber contained no CUSMA exemption at all, so no amount of certification could reach them.

A scope boundary, stated once and deliberately. This series’ arithmetic excludes subsidies, and that is an application of its own denominator discipline: the tariff calculation works because every input has a clean measurement chain (a published margin, an observable utilization rate, customs trade values), and subsidy support has no equivalent chain, only announced ceilings, contingent contracts, and disbursements reported with a lag. Subsidies enter this series structurally, in Parts 3 and 4, not numerically.

The puzzle the experiment leaves behind is the one worth sitting with. If the agreement’s realized cash value in 2024 was a rounding error, why did three governments spend 2018 to 2020 negotiating it, and why did firms keep building continental supply chains on top of it? Because the product was never the 2 points.

Next in the series: Part 3, The Insurance Policy and Who Pays the Premium, on what the agreement was actually selling, and the four groups who paid for it.

Running Glossary  ·  Terms Introduced in Part 2

IEEPA. International Emergency Economic Powers Act (1977). The emergency statute under which the February 2025 executive orders imposed the 25%/10% tariffs. The Supreme Court held in February 2026 that it does not authorize tariffs (Part 4).

Section 232. Trade Expansion Act of 1962 authority for tariffs on national-security grounds. Basis of the steel, aluminum, copper, auto, truck, lumber, and semiconductor tariffs. Carries no CUSMA exemption except a partial carve-out for CUSMA-qualifying vehicles and parts.

Effective tariff rate. Duties actually collected divided by import value. Distinct from statutory rates: a 25% tariff few goods actually pay produces a low effective rate.

Originating / originable. A good is originating if it meets CUSMA rules of origin; “originable” here means the good could meet them if the producer does the certification work.

(Part 1 defined: preference margin, utilization rate, MFN, rules of origin, TiVA, Section 122.)

A note on framing: Fully Briefed synthesizes publicly available government source material and translates it into financial terms. This is education, not legal, customs, or tax advice. The arithmetic in this series states its assumptions and sources so you can substitute your own inputs; for decisions that turn on your own tariff classification, origin documentation, or contract exposure, work with your customs broker and professional advisors.

Trevor Ryhorchuk, CPA, CIA, PMP

Fully Briefed — Canadian Trade Intelligence
tradeintel.fullybriefed.ca