The Treaty Didn't Expire. Its Ceiling Did.

On July 1, the Office of the United States Trade Representative said something that sounds like the end of an era and reads, in the fine print, like a change of clocks. 'The United States did not agree to renew the USMCAUSMCAThe United States-Mexico-Canada Agreement, the North American free-trade pact that replaced NAFTA and took effect on July 1, 2020. in its current form. As a result, the USMCA is not renewed,' Ambassador Jamieson Greer stated, in the readout of the mandated first joint reviewjoint reviewA scheduled meeting where the three countries decide whether to extend the agreement's guaranteed lifespan..

Then almost nothing happened. The Mexican stock index closed essentially flat, down about two hundredths of a percent. The peso barely twitched, easing less than a tenth of a percent on the day. No tariff line changed; the pact that governs roughly $873 billion in US-Mexico goods trade stays fully in force. If you were watching prices, you would have missed it entirely.

That gap, between a shrug in the markets and a structural change in the treaty's design, is the story. Because what Washington declined was not the agreement itself. It was one specific thing: the chance to reset the agreement's guaranteed lifespan and push it back out into the future in a single stroke.

The tempting question is 'Will USMCA survive?' That is the wrong one; the answer is unknowable and years away. The narrow, answerable question is this: when Washington declined the six-year renewal, what exactly got shorter, and would a cost land in prices, or in the duration of capital commitments?

Note what the claim does not say. It does not say the deal could vanish next year. That is the most common misreading of 'annual reviews,' and it is wrong: what became annual is the review meeting, not the deadline. USMCA cannot expire before July 1, 2036, and the three governments can still reset the full 16-year term at any point by putting an extension in writing. The claim also does not say a cost has appeared. It says where to look for one, and where not to.

USMCA was built with a deliberate piece of psychology inside it. During the 2017-18 renegotiation, the first Trump administration pushed for a hard five-year sunset, a clause that would kill the deal automatically unless it was actively renewed. Canada and Mexico refused, arguing that a five-year fuse would scare off exactly the long-horizon factory investment the deal was meant to attract. The compromise: a 16-year life, with a joint review at year six to decide whether to reset the clock for another long stretch.

That year-six review was the moment the runway could be extended in a single decision. On July 1, 2026, it arrived, and the US let it pass without agreeing to reset. Under Article 34.7Article 34.7The clause in USMCA that schedules a joint review every six years; if all three parties agree, the treaty's 16-year term is extended, and if they do not, it switches to yearly reviews counting down to a possible expiry in 2036., that does not end anything. It flips the review from once every six years to once a year, each meeting another chance to agree on an extension, all of it counting down toward July 1, 2036 if no agreement is ever reached.

The tariffs are unchanged. What changed is the shape of the promise: a runway that could be reset in one move is now a countdown that has to be actively stopped.

So the honest way to describe the change is not 'the deal is dying' and not 'nothing happened.' It is that a company weighing a 20-year plant can no longer point to a single renewal that locks in a long, guaranteed horizon. Instead it faces a shorter guaranteed floor and an annual negotiation whose outcome nobody can promise. The legal floor is still a decade out. The certainty is what got thinner.

If the change is about the horizon rather than the price, then the place to look for a cost is not the peso or the stock market, assets you can sell in seconds, but the slowest money in the economy: brand-new factories that only pay back over decades. That is the logic of duration. A shorter guaranteed horizon should, in theory, weigh most on the investments whose payback runs past the horizon's edge.

Here the evidence is genuinely mixed, and it is important to say so. Mexico had just posted a record. Foreign direct investment hit US$23.591 billion in the first quarter of 2026, the highest first quarter in the data going back to 1999. On its face, that looks like the opposite of investors fleeing a shortening horizon.

But open the box and the record looks thinner. Of that $23.591 billion, about $22.2 billion, roughly 94%, was reinvested earnings: profits that companies already operating in Mexico simply plowed back in rather than fresh money arriving from abroad. Genuinely new, from-scratch investment was just $1.7 billion. (The pieces do not add to the headline in the obvious way because a third component, intercompany accountsintercompany accountsLoans and balances moved between a foreign parent and its local affiliate; this component can be negative in a quarter if affiliates send money back to the parent, which is why the FDI parts need not add up neatly to the headline total., was slightly negative, affiliates sent about $0.3 billion back to their parents.) The headline is a record; the new-commitment engine inside it was running quietly.

It would be easy to read the flat market as investors missing the point. It is more likely they are pricing it correctly. More than 80% of Mexican exports still cross into the US tariff-free under USMCA, and none of that changed on July 1. For an asset you can exit tomorrow, a risk that first bites in 2036, if it ever bites, is close to irrelevant today. The peso and the Mexbol are doing exactly what short-duration instruments should do with a long-dated, low-probability risk: almost nothing.

The mismatch is the whole point. A cost that shows up in the duration of capital commitments would not register in a daily price print at all. It would show up as a factory decision deferred, a site selection that quietly picks somewhere else, a board that waits one more annual review before committing. Those do not move a ticker. They move the investment data, months later.

This is where intellectual honesty matters most, because the same soft new-investment number has at least three innocent explanations that have nothing to do with the review switch. Auto tariffs already in place give exporters their own reason to pause. Global interest rates make every long-horizon project harder to justify. And Mexican FDI has leaned on reinvested earningsreinvested earningsProfits a foreign company already earns inside a country and plows back in, rather than fresh money brought in from abroad. for years, so a low share of fresh money is closer to the norm than to an alarm. Any of these could produce a freeze in new commitments with no help at all from the certainty horizon.

So the claim has to be falsifiable, and it is. The pre-event baseline already shows new commitments as the weak component. The test is what happens next: whether new-investment commitments keep freezing in the Q2 and Q3 2026 FDI data, released later this year, the first readings that fall after July 1. If greenfield investmentgreenfield investmentBrand-new investment that builds something from scratch, such as a new plant, as opposed to expanding an existing one. holds up as calmly as the peso did, the duration story fails, and the shrug was the right call all the way down. If new commitments keep stalling while reinvested earnings carry the headline, the cost of a shorter horizon will have shown up exactly where the theory said it would: not in the price of anything, but in the things that quietly never got built.

The treaty didn't expire. Its ceiling did, and the bill, if there is one, arrives in the investment data, not the ticker.