USMCA Review Enters Rolling Annual Phase as July 1 Deadline Passes; IEEPA Refund Chaos Deepens Ahead of June 9 Hearing

The Analysis

Strictly speaking, nothing expired. That is the first thing to get right about the USMCA review, because the coverage reads like a near miss and it is not one. With the 1 July mandated review deadline passed without a formal renewal, the agreement enters a rolling annual review. Existing terms remain in force. No preference you claimed last quarter became invalid the day after.

What changed is the shape of the risk, and it changed in one specific way. A renewal date is a date you can plan to. A rolling review is a standing option held by any of the three parties to reopen any provision at any time. The cliff is gone. The trapdoor is not.

The second file in the same story is about money already paid. Of USD 166 billion in potential IEEPA tariff refunds, only USD 85 billion has entered processing, and a 9 June hearing will decide whether full retroactive refunds must be issued. Two halves of one balance sheet: what you may owe, and what you are owed.

Four elements make up the review outcome, and each one matters separately. There is no formal renewal. Existing terms remain in force, which is the part that keeps your current qualification valid. Any of the three parties can reopen negotiations on a specific provision at any time, which is the part that removes your horizon. And the friction is already named: rules of origin enforcement on Chinese inputs inside Mexican supply chains, plus U.S. tariff levels on autos, steel, aluminum and copper. The refund file adds two numbers, USD 85 billion in processing against USD 166 billion of potential refunds, and one date, 9 June.

Why a rolling mechanism rather than a renewal or a lapse. Renewal needs all three parties to sign a text they currently disagree on. Lapse would be disruptive for all three and legally untidy for the two that have built supply chains around the preference. A rolling review is the only outcome that requires no agreement on substance, because it defers every disputed item into a process rather than resolving it. That is not a failure of the agreement. It is the agreement doing the one thing all three can live with, and it tells you that the disagreement is real rather than theatrical.

The distribution of pain is lopsided, and it does not follow size. Autos, steel, aluminum and copper sit on capital cycles longer than any review cycle, so a rolling review forces every investment case to assume the applicable tariff level can move mid-payback. Nothing has changed for them today, and that is exactly why the planning problem is worse: there is no event to point at when the assumption is challenged. Producers assembling in Mexico with Chinese inputs face a different exposure, and it is evidentiary rather than rate-based. A plant that can trace a component to its source answers a rules-of-origin inquiry in days. A plant that buys through traders answers it in months, and the answer may not be the one it wants. Most importers, meanwhile, feel nothing at all this quarter. The terms are unchanged, the preference holds, and the correct description of their situation is stable but conditional.

Timing behaves differently for the two files. The review has no transmission date, and that is the point of it: nothing is scheduled, so nothing can be prepared against a schedule. What you can watch is the agenda, because a reopening starts as a request rather than an announcement. The refund file has a real date, 9 June, but a ruling in your favour does not produce cash on 10 June. Processing is the constraint, and processing does not accelerate because a judge ruled.

Here is the part the headline misses, and it is structural rather than tactical. Under a rolling regime you certify at entry against a rule that can be reopened afterward. The claim was valid when you made it and it can still be challenged later, which means the exposure is retrospective by construction. A normal tariff change is prospective: you learn the new rate and you price it forward. This is the opposite. You price it now and find out later whether the basis survived. Anyone building a compliance file to satisfy today's rule, and stopping there, has built for the wrong question.

The second missed point is arithmetic. Eighty-five against one hundred and sixty-six is roughly fifty-one per cent. So the refund story is not mainly about entitlement. Half the pool is not disputed, it is unprocessed, which makes administrative throughput the binding constraint. That matters for two reasons. It tells you the timeline is set by capacity rather than by law. And it tells you that the same capacity is what a rolling review will consume, because every reopened provision generates documentation work. Assume a 6% cost of funds, a stated assumption, and USD 81 billion of stalled refunds carries about USD 4.9 billion a year in financing cost across the pool. That is the price of a queue.

An illustrative sum for a single company, assumptions stated. Assume your IEEPA claim is USD 4 million and that your claim is representative of the pool, so 51% is in processing and 49% is stalled. That is USD 2.04 million moving and USD 1.96 million sitting still. At a 6% cost of funds the carry on the stalled half is about USD 118,000 a year, close to USD 10,000 a month, which is the number to put in front of a CFO rather than the headline claim. Now the USMCA side.

Assume USD 50 million a year entering the U.S. under preference, and assume 20% of that volume depends on inputs whose origin you cannot document within five days. That is USD 10 million of volume with a duty basis that can be re-tested. I am not going to put a rate on it, because the rate difference for autos, steel, aluminum and copper is sector-specific and I would be guessing. The point is the base: ten million dollars of exposure that is currently invisible because nothing has been asked yet.

The counterfactual is what separates a plan from a posture. If a party tables a rules-of-origin change within the next twelve months, the planning assumption flips from preference is stable to preference is conditional, and every SKU with unverified inputs needs a re-test. If nobody tables anything, the rolling review is a standstill in all but name, and the correct response was documentation and nothing more. On the refund side, a ruling for full retroactive refunds ends the entitlement question and turns it into a scheduling question. A ruling against keeps USD 81 billion contested and makes documentation the only asset that matters. Note the asymmetry between the two risks: the review risk has no ceiling and no date, while the refund risk is capped by the size of your claim. They need different hedges, and treating them as one story is how a treasury ends up hedged against the wrong thing.

The mistake to avoid is reading the absence of a renewal as an absence of change. Nothing was signed, nothing expired, and the terms are intact, which is precisely the combination that lets a company stop paying attention for a year. The rolling review does not announce itself. It arrives as a request to reopen one provision, and by then the qualification file is either current or it is not.

↑ Back to the key points

— By Dr. Ingrid Voss