From NAFTA to USMCA

The phone call came in late January 2017, one week into the Trump presidency. Mexican President Enrique Peña Nieto was told that the U.S. intended to renegotiate NAFTA — or withdraw entirely. The threat was credible: Article 2205 of NAFTA allowed any party to withdraw with six months’ notice, and the new president had won the Rust Belt by promising to tear up the deal. “We’re going to renegotiate NAFTA,” Trump told cheering crowds in Ohio and Michigan. “And if they don’t want to renegotiate, we’ll terminate it.”

Negotiations began in August 2017 and ran for fourteen months. Robert Lighthizer, the U.S. Trade Representative, led the American side with a clear agenda: tighten rules of origin to prevent third-country goods (primarily from China and Japan) from entering the U.S. tariff-free through Mexican and Canadian back doors. Lighthizer, a veteran of Reagan-era trade battles with Japan, saw NAFTA not as a free-trade agreement but as a loophole — a way for foreign companies to build assembly plants in Mexico using Chinese components and ship the finished product to America duty-free.

The deal that emerged — signed November 30, 2018, by Trump, Peña Nieto, and Canadian Prime Minister Justin Trudeau — carried a new name (USMCA, the United States-Mexico-Canada Agreement) and several substantive changes. Whether those changes achieved their stated goals is a question the data answers definitively.

The auto content rules were the centerpiece. Under NAFTA, a car needed 62.5% North American content to qualify for duty-free treatment. Under USMCA, that threshold rose to 75%. More importantly, the new agreement added a “labor value content” requirement: 40% of a car’s value (45% for pickup trucks) had to be produced by workers earning at least $16 per hour. Since the average Mexican auto worker earned roughly $3.50 per hour in 2018, this provision was aimed squarely at Mexico. The idea was simple: if Mexican plants couldn’t meet the wage threshold, automakers would shift production to the U.S. or Canada, where workers already earned above $16.

The dairy fight was Canada’s headline. Canada’s supply management system — a regime of production quotas and import tariffs reaching 270% on some dairy products — had been a flashpoint since NAFTA’s original negotiation. Trump, at one point, accused Canada of being “very unfair” to Wisconsin dairy farmers. Under USMCA, Canada agreed to open 3.59% of its dairy market to American producers — a small but politically significant concession that affected roughly $600 million in trade. Quebec dairy farmers protested. Wisconsin dairy farmers celebrated. The net impact on the $770 billion trade relationship was negligible.

A digital trade chapter was entirely new. NAFTA, negotiated before the internet became commercial, said nothing about e-commerce. USMCA banned customs duties on digital products (software, e-books, music), prohibited data localization requirements (governments couldn’t force companies to store data locally), and protected platforms from liability for user-generated content. This was arguably the most forward-looking provision in the agreement — and the one with the least measurable impact on the trade balance, since digital services were already flowing freely.

The sunset clause was Trump’s insurance policy. USMCA has a 16-year term with a formal review at year six (2026 — this year). If the parties don’t agree to extend it, the agreement terminates at year 16. This gave future presidents leverage to renegotiate again — and it gave markets permanent uncertainty about North America’s trade architecture. The first mandatory review is scheduled for July 2026, and it’s expected to reopen the auto content rules that have become the agreement’s most contentious provision.

The BEA data tells a story the negotiators did not intend. Under NAFTA (1999–2016), the U.S. goods deficit with Mexico grew from $24 billion to $69 billion — an increase of $45 billion over seventeen years, or about $2.6 billion per year. Under the USMCA era (2016–2024), the deficit grew from $69 billion to $181 billion — an increase of $112 billion in eight years, or $14 billion per year. The new agreement didn’t slow deficit growth. It coincided with a fivefold acceleration.

The auto content rules, specifically, appear to have backfired. The 75% North American content requirement and the $16/hour labor value provision were supposed to discourage production in Mexico. In practice, automakers adapted. Some paid the 2.5% tariff on finished vehicles rather than restructure their supply chains — a cost easily absorbed on a $40,000 truck. Others shifted their supplier mix within North America to hit the 75% threshold while keeping final assembly in Mexico. Still others invested in automation at Mexican plants, boosting productivity per worker without raising wages to $16. The Mexican auto deficit grew from $75 billion in 2016 to $139 billion in 2024 — nearly doubling under the very rules designed to shrink it.

General Motors provides the case study. In 2019, GM closed its Lordstown, Ohio assembly plant (1,700 workers, on the payroll since 1966) and its Detroit-Hamtramck plant (1,300 workers). That same year, GM announced a $1 billion expansion of its Silao complex in Guanajuato, Mexico, where 6,000 workers build the Chevrolet Silverado and GMC Sierra. The math was straightforward: even accounting for USMCA’s content rules and potential tariffs, Mexican production was more cost-effective. The Silverado, America’s second-best-selling vehicle, is substantially built in Mexico.

Since auto content rules were the USMCA’s signature innovation, the auto trade data serves as the purest test of whether the agreement worked as designed. The table below tracks every year since the renegotiation began.

The deficit in automotive vehicles, parts, and engines nearly doubled from $75 billion to $139 billion. U.S. auto exports to Mexico grew by 30% ($33B to $43B) — these are largely components shipped south for assembly. But Mexican auto imports grew by 69% ($108B to $182B) — finished vehicles and subassemblies shipped north. As explored in Episode 2, the auto deficit alone now accounts for 77% of the entire goods deficit with Mexico. The USMCA’s content rules didn’t break this pattern. They may have reinforced it, by creating compliance requirements that favored large incumbents with established Mexican operations over new entrants who might have built in the United States.

No analysis of the USMCA era is complete without acknowledging the elephant in the room: the deficit with Mexico didn’t grow in isolation. It grew as the deficit with China shrank. In 2018, the U.S. goods deficit with China peaked at $417 billion. By 2024, it had fallen to $295 billion — a $122 billion improvement. Over the same period, the Mexico deficit grew by $97 billion (from $84 billion in 2018 to $181 billion). The timing is not coincidental.

Trump’s tariffs on Chinese goods — 25% on $250 billion of imports, imposed in waves during 2018 and 2019 — made Mexico relatively more attractive overnight. A Chinese-made television facing a 25% U.S. tariff could be assembled in Tijuana using Chinese components and shipped to America duty-free under USMCA. The arbitrage was irresistible. As we’ll explore in detail in Episode 9, Chinese companies poured billions into Mexican factories — not to serve the Mexican market, but to serve the American one.

This is the central irony of the USMCA era. The agreement was designed to reduce trade diversion — to stop third-country goods from entering the U.S. through NAFTA back doors. But the simultaneous China tariffs created the strongest incentive for trade diversion in history. The USMCA’s 75% content rule applied to autos, but not to televisions, washing machines, or server racks. For electronics and consumer goods, Mexico became China’s front door.

While Mexico grabbed the headlines, Canada may have benefited most from the USMCA era — though not from the agreement itself. Canada’s goods deficit with the U.S. has been remarkably stable: $69 billion in 2024, essentially the same as $70 billion in 2004. It swings with oil prices (down to $16 billion in 2016 when crude crashed, up to $87 billion in 2022 when it surged) but shows no structural trend. The USMCA didn’t meaningfully change the U.S.-Canada relationship because there was less to change: Canadian wages were already comparable to American ones, and the auto content rules were already largely met by existing production.

What benefited Canada was the services trade explored in Episode 7. The $33 billion services surplus — driven by consulting, finance, and technology — absorbs nearly half the goods deficit and occasionally turns the total balance positive. Canada’s dairy concession (3.59% of the market, roughly $600 million) was a political sacrifice that barely registered in the $770 billion goods relationship. In exchange, Canada secured continued tariff-free access to the world’s largest consumer market for its auto, energy, and commodity exports.

The dairy fight, for all its political theater, illustrates a broader truth about trade negotiations: politicians fight over symbolic goods while the structural flows go unaddressed. Trump spent more public energy on Canadian dairy tariffs ($600 million at stake) than on Mexican auto imports ($182 billion at stake). The USMCA’s dairy provisions received more press coverage than its auto provisions. But the auto flows determined whether the deal would achieve its stated objectives — and on that measure, it fell short.

Under Article 34.7, the USMCA requires a “joint review” by the three parties no later than six years after entry into force — which means July 2026. The review will determine whether to extend the agreement for another 16 years or let it begin winding down. All three parties are expected to seek modifications.

The likely flashpoints are already forming. Auto content rules will be revisited, with American unions pushing for stricter enforcement of the $16/hour labor value provision and automakers lobbying for exemptions. Electric vehicles weren’t contemplated in the original USMCA; battery minerals (lithium, nickel, cobalt) sourced from outside North America create content-rule complications for EV manufacturers. Tesla’s planned Gigafactory in Monterrey, announced in 2023, would assemble vehicles using batteries with Chinese-sourced cathode materials — potentially failing the 75% content test. Chinese investment in Mexico will be the most politically explosive topic: whether to add provisions blocking USMCA benefits for goods produced in Chinese-owned Mexican factories.

The data presented in this series will be central to those negotiations. The $181 billion goods deficit, the $139 billion auto gap, the China-to-Mexico trade diversion — these numbers frame the political argument. Whether the next iteration of North America’s trade architecture addresses the structural forces behind these numbers, or simply rearranges the political furniture, will determine whether the NAFTA experiment’s next chapter looks different from its first thirty years.