NAFTA and the USMCA

North America built the deepest economic integration outside Europe while building almost none of Europe’s institutions. There is no common external tariff, no parliament, no court with primacy over national law, no cohesion fund and no free movement of workers — and yet automotive components cross the Rio Grande and the 49th parallel several times before a finished vehicle is sold. The syllabus names NAFTA, which no longer exists: it was replaced on 1 July 2020 by the United States–Mexico–Canada Agreement, and that replacement is not a deeper version of the same thing.

North America before the treaty: integration built by industry, not by institutions

PSIR Optional Notes {PDF}

Complete Notes Package (Paper I & II) | 1500+ Pages | Instant Download

Discounted Price | Buy Now @ ₹1000 →

• The common shorthand that North American integration began in 1994 is wrong in both directions. Canadian–American industrial integration was already three decades old by then, and Mexican integration into American manufacturing was already thirty years old on the border.
• What NAFTA did was extend an existing bilateral arrangement southward and give a middle-income country the same terms of market access as a rich one — which is why it was, at the time, the first free trade agreement between developed and developing economies of comparable ambition.

The Auto Pact and the Canada–US Free Trade Agreement

• The Canada–United States Automotive Products Agreement, signed by Lester Pearson and Lyndon Johnson in January 1965 and universally called the Auto Pact, is the true origin of continental supply chains.
  • It removed tariffs on vehicles and original parts moving between the two countries, but only for manufacturers who accepted production safeguards: assemblers undertook that Canadian production would not fall below 1964 levels and that they would maintain a production-to-sales ratio in Canada.
  • This was managed trade, not free trade, and the USMCA’s automotive rules work on the same logic half a century later.
  • The effect was dramatic. Exports of Canadian-built vehicles to the United States rose from roughly 7% of Canadian production in 1964 to about 60% by 1968, and automotive manufacture overtook pulp and paper as Canada’s largest industry.
  • Its costs were visible early: employment was concentrated in southern Ontario and overwhelmingly blue-collar, while design and corporate functions stayed in Detroit. A branch-plant economy is integrated without being equal — a Canadian criticism of every agreement since.
  • The Auto Pact was abolished in 2001 after a WTO panel found its conditional tariff exemption incompatible with multilateral rules, by which time NAFTA had superseded it in practice.
• The Canada–United States Free Trade Agreement (CUSFTA) is NAFTA’s actual predecessor, and forgetting it makes the 1994 agreement look more novel than it was.
  • Negotiations opened in May 1986 under Brian Mulroney and Ronald Reagan, led by Simon Reisman for Canada and Peter Murphy for the United States; the text was concluded on 4 October 1987, signed 2 January 1988 and entered into force 1 January 1989.
  • Tariffs were already low — Canadian and American average applied rates were close to 1% by the 1980s — so the agreement was less about tariffs than about secure access: locking out the American trade-remedy actions that periodically closed the border to Canadian lumber, steel and agricultural goods.
  • Canada’s price for signing was a binational panel review of anti-dumping and countervailing duty determinations, allowing a panel of nationals from both countries to review whether an agency had applied its own domestic law correctly. Canada carried this into NAFTA as Chapter 19 and into the USMCA as Chapter 10, and has refused to give it up in every renegotiation since.
  • The 1988 Canadian federal election was fought as a referendum on it, with the Liberals and New Democrats warning of becoming a “fifty-first state”. Mulroney won a majority on a minority of the vote as the only pro-agreement party — North American trade agreements are ratified against domestic majorities, not with them.

Mexico’s turn: from import substitution to lock-in

• Mexico’s arrival was the genuinely new thing about NAFTA, and it followed from a national economic collapse rather than from any enthusiasm for regionalism.
• The 1982 debt crisis — Mexico’s August 1982 announcement that it could not service its external debt — ended five decades of import-substituting industrialisation (ISI), the strategy of building domestic industry behind high tariffs, import licences and state ownership.
  • ISI had produced growth through the 1960s but left an economy unable to export enough to service debt accumulated against oil revenues, and the oil-price collapse removed the last support.
  • The “lost decade” that followed — inflation, capital flight, collapsing real wages, successive devaluations — destroyed the developmentalist model’s authority from within.
• Liberalisation therefore began well before any treaty. Mexico acceded to the GATT in 1986, unilaterally cutting maximum tariffs and abolishing most import licensing; Carlos Salinas de Gortari, president from 1988, privatised banks and state enterprises, deregulated foreign investment and opened the capital account.
• The decisive motive for seeking a treaty was not market access but credibility. Mexico had liberalised unilaterally before and reversed course; investors discounted the reforms accordingly, and capital would not come at the scale required.
  • A treaty with the United States converted domestic policy into international obligation, making reversal by a future Mexican government legally costly and diplomatically visible. This is the lock-in or commitment-device function of a trade agreement, and it is the single most important thing about NAFTA from a political-science point of view.
  • The insight generalises: a north–south agreement is often valued less for the tariff schedule than for the constraint it imposes on the signatory’s own successors — a deliberate exchange of policy space for credibility, which is also why such agreements are resented a generation later.
  • Salinas approached Washington in 1990 after failing to attract European capital; George H. W. Bush negotiated the agreement and Bill Clinton brought it into force. NAFTA had bipartisan American authorship, which is why the later bipartisan repudiation of it is so striking.
  • Salinas’s public case for the treaty in both countries was that Mexico wished to export goods rather than people — the migration argument examined below.

The maquiladora border

• The maquiladora programme had already built a manufacturing frontier before any treaty existed. Mexico launched the Border Industrialization Programme in 1965, directly in response to the end of the American Bracero guest-worker scheme in 1964, which had returned large numbers of seasonal workers to a border region with no industry.
  • The model was simple: inputs entered duty-free on condition that the output was re-exported, so a maquiladora was an assembly plant inside Mexico but outside Mexico’s tariff wall.
  • By 1985 maquiladoras had overtaken tourism as Mexico’s largest source of foreign exchange; by the mid-2000s they accounted for roughly half of Mexican exports, and today several thousand plants along the two-thousand-mile border employ on the order of a million workers.
• Three consequences shaped everything that followed.
  • NAFTA did not create the border economy; it generalised it. The treaty extended maquiladora-style duty-free treatment to the whole country and removed the re-export condition, which is why Mexican manufacturing spread inland to the Bajío rather than staying at the frontier.
  • Its labour practices became the template for the criticism of NAFTA: independent unionisation suppressed in favour of employer-friendly “protection contracts” tied to the state-linked Confederation of Mexican Workers, wages a fraction of American levels, and documented pregnancy testing and dismissal of pregnant workers in breach of Mexico’s ICCPR and CEDAW obligations.
  • The sector’s dependence on assembly rather than design left Mexico with high export volumes and low domestic value added — the structural weakness that the USMCA’s rules of origin now try to address from the American side rather than the Mexican one.

What NAFTA actually was

• NAFTA was signed on 17 December 1992 and entered into force on 1 January 1994, binding Canada, Mexico and the United States. The Zapatista uprising in Chiapas began the same day, and the coincidence was not accidental: the rebellion was framed explicitly as a response to the agreement’s agricultural provisions.

A free trade area and nothing more

• On the standard ladder of economic integration, NAFTA was a free trade area — the lowest rung — and it never climbed a step. It was not a customs union, a common market or an economic union, and it made no pretence of political integration.
• Because there was no common external tariff, rules of origin were unavoidable. In a customs union, goods face the same duty at any member’s border and then circulate freely; in a free trade area, members keep their own external tariffs, so without origin rules an exporter would route goods through whichever member charged least.
  • The whole architecture of North American integration therefore runs through origin rather than through borders, which is why an apparently technical matter became the central battleground of the renegotiation.
• There was no free movement of labour — only a temporary-entry category for certain professionals, the TN visa. Where free movement of persons is a treaty right in the European Union, North American integration deliberately liberalised goods, services and capital while leaving people under national control.
• The institutions were minimal by design. A Free Trade Commission of cabinet-level trade ministers met periodically; a small trinational Secretariat administered dispute panels; a handful of working groups existed. There was no legislature, no court, no commission with a right of initiative, no supranational law and no independent budget.
• There was no development fund — the sharpest institutional contrast with Europe. The European Communities accompanied the accession of Greece, Spain, Portugal and later central Europe with structural and cohesion funds, on the theory that a market between unequal economies needs a redistributive mechanism to sustain convergence. NAFTA joined economies with a per-capita income gap near six to one and transferred nothing.

A market between unequals with no transfer mechanism does not converge on its own; it converges only where the market happens to want it.

The core bargains: Chapters 11, 19 and 20

• Chapter 11 was NAFTA’s most contested provision. It gave foreign investors a right of action domestic investors did not have, before arbitrators rather than courts, against public measures.
  • Ethyl Corporation v. Canada, over a ban on a fuel additive, and Metalclad v. Mexico, over a refusal to permit a hazardous waste facility, set the pattern: environmental and public-health regulation challenged as indirect expropriation.
  • Critics in all three countries argued the mechanism produced regulatory chill. Its defenders argued it was the protection that made investment in Mexico’s uncertain legal environment possible — the lock-in argument in another form.
• Chapter 19 was the provision the United States most wanted gone and Canada most wanted kept. It reviews not whether a duty is justified but whether the investigating agency followed its own law, and Canadians regard it as the only realistic check on American trade remedies — above all in softwood lumber, unsettled by CUSFTA and unsettled since.
• Chapter 20 quietly failed. A party could obstruct a panel by refusing to appoint panellists or maintain the roster, and did; Canada and Mexico abandoned it and litigated at the WTO instead. A state-to-state mechanism that can be blocked at will is an obligation without a remedy.

The side agreements: NAALC, NAAEC and the CEC

• Congressional passage in 1993 required Clinton to answer labour and environmental objections without reopening a text already signed. The solution was two side agreements negotiated in parallel and concluded outside the core treaty.
  • The North American Agreement on Labor Cooperation (NAALC) committed each party to enforce its own labour law, established National Administrative Offices to receive public submissions, and provided for consultations and, in theory, sanctions.
  • The North American Agreement on Environmental Cooperation (NAAEC) created the Commission for Environmental Cooperation (CEC) in Montreal, with a submissions on enforcement matters process under which any person could allege that a party was failing to enforce its environmental law.
• Both were judged weak, and the reason is structural rather than a failure of will.
  • They obliged parties only to enforce their own existing law, setting no substantive floor — a state could comply by having weak law and enforcing it.
  • The remedies were slow and consultative: the labour sanctions route ran through a long ministerial process and was never completed to a penalty.
  • Above all they sat outside the core text, so a violation could not be met with the withdrawal of trade concessions as a breach of the tariff schedule could. A commitment that cannot be enforced with trade consequences is not a trade commitment.
  • The CEC’s factual records produced real transparency about enforcement failures, but it could not compel; the North American Development Bank and Border Environment Cooperation Commission addressed border infrastructure far below the scale of need.

The record, argued from both sides

• NAFTA did roughly what trade theory predicted: aggregate gains small relative to the economies involved, and losses large relative to the places that bore them. Neither the claim that it was an unqualified success nor the claim that it destroyed American manufacturing survives the evidence.

Trade and the continental supply chain

• Trilateral trade roughly tripled in NAFTA’s first two decades, from around $290 billion in 1993 to above $1 trillion by 2011, and reached about $1.2 trillion in 2019, the year before the USMCA took effect.
• The volume matters less than the structure of the trade it produced.
  • Most of it is intra-industry and intra-firm: not Canada selling wheat and Mexico oil, but the same firms moving partly finished goods between their own plants.
  • Automotive components cross a border several times before final assembly, so part of an American-assembled vehicle’s value is counted as an import repeatedly — which is why tariffs on North American trade are unusually destructive.
  • Imports from Mexico contain a far higher share of American-made content than imports from Asia, the strongest single argument that regional trade and domestic manufacturing are complements rather than substitutes.
• Agriculture and electronics followed the same pattern: Mexican fresh produce reorganised North American winter supply, Canadian energy and American refining became one system, and electronics assembly in Guadalajara and Ciudad Juárez fed American final markets.
• The region remains extraordinary in scale: US–Mexico goods trade reached a record $872 billion in 2025, US–Canada trade was of comparable order, and with Canada–Mexico trade the trilateral total is near $1.8 trillion. The three economies account for around 30% of world GDP and over 500 million people.

The American employment argument

• Ross Perot’s warning in the 1992 presidential debates that NAFTA would produce a “giant sucking sound” of jobs going south is the most quoted sentence in the history of trade politics, and the reason the employment question dominates American discussion of the agreement to this day.
• The estimates are wide, and the width is itself the finding.
  • The Congressional Research Service concluded that NAFTA’s net effect on American employment was “close to zero” and its effect on the American economy “relatively small”, while accepting that it shifted the composition of demand away from less-skilled work.
  • About 415,000 workers were certified for trade adjustment assistance as NAFTA-related between 1994 and 2001 — a real number of real dislocations, and under 1% of American employment.
  • The Economic Policy Institute, working from the growth of the bilateral goods deficit, put job opportunities lost or displaced at figures in the range of 680,000 to 770,000. Most economists reject the methodology, which treats every dollar of deficit as a subtraction from domestic production and ignores the gains from cheaper inputs.
• The attribution problem is the heart of it. The collapse in American manufacturing employment is concentrated after 2001, not after 1994.
  • David Autor, David Dorn and Gordon Hanson established that Chinese import competition following China’s WTO accession — the “China shock” — accounts for losses on an altogether larger scale: on the estimates they cite, roughly 2.4 million American jobs between 1999 and 2011 once supply-chain and local demand effects are included.
  • Their more important finding was about adjustment: wages, labour-force participation and employment in exposed local labour markets stayed depressed for a decade or more, and the offsetting gains that trade theory predicts did not appear where the losses fell.
  • This is why NAFTA is blamed for the China shock. The agreement was the visible, named, voted-on decision; normal trade relations with China was a technical vote in 2000. Politics attaches responsibility to the object it can identify.
• The distributional point survives the attribution correction, and it is the point that matters. Gains spread thinly across consumers and losses concentrated in particular industries and towns produce a politics in which the losers are organised and the winners are not — enough to explain the 2016 campaign without overstating the job numbers.

Mexico’s uneven outcome

• Mexico’s experience is not the mirror image of the American one, and treating it as such is the commonest error in accounts of NAFTA.
• What worked: manufactured exports grew several-fold, foreign direct investment rose sharply, Mexico became a serious producer of vehicles, aerospace components, medical devices and appliances, and the treaty survived the 1994–95 peso crisis, which arrived within a year and might otherwise have triggered reversal.
• What did not work: aggregate growth disappointed. Real GDP per capita growth since 1994 has averaged well under two per cent a year — below Mexico’s pre-1982 record and below Latin America’s better performers — and productivity outside the export sector barely moved.
• The geography of the gains is the sharpest finding. Industry concentrated in the north and the Bajío; the south — Chiapas, Oaxaca, Guerrero — was largely untouched and in some measures fell further behind.
  • Regional inequality inside Mexico widened under an agreement whose defenders had predicted convergence, because investment went where infrastructure, skills and proximity to the border already were.
  • A market opening reproduces existing advantage unless something counteracts it. Europe built cohesion funds; North America did not, and got the outcome the omission predicted.
• Mexico also acquired a persistent dual economy: a productive export sector alongside an informal sector accounting for well over half of employment, where productivity growth is close to zero.

Maize, the countryside and the south

• Agriculture is where NAFTA did the most identifiable harm, and the maize story is the one that must be told accurately.
• Mexican maize was the crop of two to three million smallholders, much of it rain-fed and grown for subsistence and local sale, and it met subsidised American corn produced at industrial scale.
  • The tariff-rate quota on American corn was to phase out over fifteen years; Mexico chose to admit above-quota volumes duty-free well before schedule, accelerating its own liberalisation because cheap feed grain served the livestock and processing industries.
  • Producer prices for Mexican maize fell steeply — estimates of the decline over the first decade run to around two-thirds — while consumer tortilla prices rose, because the milling and distribution chain was concentrated. Cheaper corn did not mean cheaper food.
  • Mexico dismantled CONASUPO, the state marketing and price-support agency, replacing it with PROCAMPO, a decoupled per-hectare payment large in political visibility, small in per-farm value and favouring larger holdings by design.
• Estimates of the resulting displacement vary with method and advocacy, but the direction is not in dispute: on the order of a million and a half to two million people left farming in the agreement’s first decade and a half, and the loss fell heaviest on the indigenous south.
• The Zapatista rising in Chiapas on 1 January 1994 named NAFTA and the prior repeal of the constitutional protection of ejido communal land as its grievances. The agreement’s most famous opponents were not American factory workers but Mexican peasants, and this is routinely forgotten.

The wage convergence that did not happen

• The most damaging single finding against NAFTA is that Mexican manufacturing wages did not converge on American ones. Standard trade theory, and the agreement’s own advocates, predicted that opening a labour-abundant economy to a capital-abundant one would raise wages in the former relative to the latter. It did not happen.
  • Mexican manufacturing compensation stayed at roughly a fifth to an eighth of the American level across the NAFTA period, and by some measures the relative gap widened rather than narrowed.
  • Mexican manufacturing wages fell below Chinese ones during the 2000s — unpredicted in 1993, and fatal to the simple “jobs moved to cheap labour” story, since the cheap labour was cheaper still elsewhere.
• Why convergence failed is the analytically interesting question, and there are three serious answers, which are complements rather than rivals.
  • The peso crisis of 1994–95 cut dollar wages roughly in half at the outset, and the recovery started from that floor rather than from the pre-crisis level.
  • Labour institutions suppressed the transmission from productivity to wages. Mexico’s system of employer-negotiated “protection contracts” with unions the workers had not chosen meant that productivity gains in export plants were not bargained into pay. This is exactly what the USMCA later attacked.
  • The reserve of underemployed rural and informal labour was enormous, so export-sector expansion drew workers in without bidding wages up — the classical surplus-labour result.
• The corrective came from domestic policy, not the treaty: the minimum wage was raised sharply from 2018, roughly doubling in real terms. It has run ahead of productivity growth with visible costs in informality — but it did more for Mexican wages in five years than the agreement did in twenty-four.

The migration paradox

• NAFTA was sold in part as an alternative to migration: prosperity in Mexico would keep Mexicans in Mexico. Net migration from Mexico to the United States did indeed fall to approximately zero around 2010 and turned mildly negative thereafter, as Pew Research Center analysis established. Reading that as the treaty working is a serious error of causal reasoning, and reading it as coincidence is a serious error the other way.
• The initial effect ran the wrong way. Annual migration from Mexico rose through NAFTA’s first decade, roughly doubling between the early 1990s and 2000, and the undocumented Mexican population in the United States rose from about two million in 1990 to a peak near six million in the mid-2000s.
  • The mechanism is the agricultural one already described: displacement from maize farming released labour faster than manufacturing absorbed it, and the displaced were in the south while the new jobs were in the north or across the border. The peso crisis compounded it, and the maquiladora boom drew people to the border, where onward movement was cheapest.
• The reversal after 2007 has four causes, and NAFTA is at most one of them.
  • Mexican demography. Total fertility fell from around 6.7 children per woman in 1970 to near replacement by the 2000s, so the cohorts that supplied migration became smaller — the largest single factor, and nothing to do with trade policy.
  • The 2008–09 American recession, which removed the construction jobs that had pulled Mexican workers north.
  • Enforcement and cost: heavier enforcement raised the price and risk of crossing and, perversely, ended the circular migration that had once seen workers return, converting temporary migrants into permanent residents.
  • Mexican economic stabilisation, to which NAFTA genuinely contributed — low inflation, a functioning financial system and northern manufacturing employment all raised the value of staying.
• The honest verdict is three-part. The claim that restricting migration was a major American gain from NAFTA is defensible only if “gain” means a long-run correlation rather than an intended outcome; the timing of the fall matches the recession and the demographic transition better than the treaty; and in the years when the treaty’s effects were strongest, migration rose.
• The composition of migration then changed entirely. By the 2010s the southern-border flow was predominantly Central American, and later Venezuelan, Cuban and Haitian, driven by violence and state collapse — a movement NAFTA never addressed and contained nothing to address.

Environment and the pollution-haven question

• The environmental case against NAFTA rested on the pollution-haven hypothesis: that firms would relocate to the jurisdiction with the weakest enforcement, and that competitive pressure would deter all three from tightening standards.
• The strong version was not confirmed. Relocation was driven overwhelmingly by labour costs, logistics and market access; compliance costs are a small share of total costs in most manufacturing, and multinationals often operate to a single global standard.
• The weak version has real support in specific places. Border industrialisation outran the water, sewage and hazardous-waste infrastructure to handle it, producing documented contamination along the Rio Grande and in the colonias, and Mexican enforcement capacity did not keep pace.
• The CEC produced valuable trinational data, but its recommendations bound no one and its submissions process was slow enough that findings often arrived after the harm.

What NAFTA lacked as an institution

• Beyond the record, NAFTA attracted a distinct set of criticisms as a piece of institutional design, and these are what the renegotiation actually addressed.

• Asymmetry of power is the structural fact. Where one economy is roughly ten times the size of the second and fifteen times the third, formal equality of the parties does not produce equality of bargaining: Canada and Mexico depend on the American market as the United States depends on neither.
• There was no mechanism to manage divergence. Where the EU answers a member’s difficulty with funds, derogations and conditionality, NAFTA had only the text.
• Nothing covered digital trade, data flows or e-commerce, because in 1994 they did not meaningfully exist. Over two decades the most dynamic part of the economy fell outside the agreement entirely.
• Nothing covered state-owned enterprises, currency practices or environmental and labour standards in enforceable form, all of which became standard content in later agreements.
• There was no scheduled point of renewal, so the agreement could only be changed by unanimous amendment or exited by withdrawal — an all-or-nothing design that made political pressure accumulate until it broke out as a threat to leave.

NAFTA’s deepest flaw was not what it contained but what it had no procedure for: revisiting itself.

The renegotiation: from withdrawal threat to signature

The withdrawal question and Article 2205

• By 2016 both American presidential candidates had turned against the agreement; Donald Trump described NAFTA as “the worst trade deal ever made”, and the Democratic position had moved from Obama-era proposals to revise it to outright opposition. A treaty with bipartisan authorship acquired bipartisan repudiation.
• NAFTA Article 2205 allowed a party to withdraw six months after giving written notice, with the agreement remaining in force between the others. The threat of invoking it was the central instrument of American leverage from 2017 to 2018, and it raised a genuine constitutional question.
  • The agreement had been approved by Congress and implemented by the NAFTA Implementation Act. A presidential notice of withdrawal would terminate the international obligation, but it would not by itself repeal the domestic statute or restore the pre-1994 tariff schedule, which is fixed in American law.
  • The result would have been legal chaos rather than a clean exit — an international withdrawal sitting on top of a domestic statute still granting NAFTA tariff treatment, with litigation certain and the constitutional allocation of the commerce power unresolved. That uncertainty was itself part of the leverage.
  • The episode illustrates something general about the regionalisation of world politics: when an agreement has no scheduled review, the only pressure valve available to a dissatisfied party is the threat of exit, and exit threats are far more destabilising than scheduled renegotiation.
• The consequences of an actual American withdrawal would have gone well beyond trade volumes — which is precisely why it never happened.
  • Continental automotive production would have faced duties compounding on the same value at every crossing.
  • Mexico’s post-1986 development strategy rested on the treaty’s credibility function; removing it would have devalued three decades of reform and pushed Mexico toward diversification, including toward China.
  • Canada’s energy exports and its security relationship with Washington would have been drawn into a trade dispute they had been insulated from.
  • For regionalism generally, a demonstration that the largest member of a regional agreement can threaten to dissolve it at will damages every arrangement built on asymmetric membership.

The negotiation, the protocol and entry into force

• Formal renegotiation began in August 2017 and ran through seven rounds and a long stalemate. The opening American demands were a five-year sunset clause, abolition of Chapter 19, sharply higher American content requirements and restrictions on procurement.
• The United States and Mexico concluded a bilateral understanding on 27 August 2018, and Canada joined on 30 September 2018 — the bilateral-first sequencing that would reappear in 2026. The agreement was signed on 30 November 2018 in Buenos Aires, in Spanish, English and French, and is known as CUSMA in Canada and T-MEC in Mexico.
• Ratification stalled in the Democratic-controlled House over enforcement, and the price of passage was a Protocol of Amendment agreed on 10 December 2019 — the most consequential changes in the whole process.
  • It created the facility-specific Rapid Response Labour Mechanism, which was not in the signed text.
  • It reversed the burden of proof on the “in a manner affecting trade” requirement for labour and environmental violations, so that a violation is presumed to affect trade unless the respondent proves otherwise, and removed the requirement to show a “sustained or recurring” pattern of violence against workers.
  • It stripped out the pharmaceutical provisions, deleting ten years of data exclusivity for biologics and additional exclusivity for new uses of existing drugs, on the argument that they would raise medicine prices.
  • It closed the panel-blocking loophole in state-to-state dispute settlement.
  • It added the requirement that steel and aluminium for automotive rules of origin be North American, with steel to be melted and poured in the region from the seventh year.
  • It required adherence to a longer list of multilateral environmental agreements, including CITES, the Montreal Protocol, MARPOL and Ramsar.
• Mexico ratified in December 2019, the United States enacted implementing legislation on 29 January 2020, Canada ratified in March 2020, and the agreement entered into force on 1 July 2020. The most enforceable labour provision in trade law was extracted at ratification by a legislature, not designed by negotiators.

Inside the USMCA

• The USMCA is still a free trade area: no common external tariff, no supranational institution, no free movement of people, and a Free Trade Commission of ministers rather than a commission with a right of initiative. The change is in what the agreement regulates, not in how deeply it integrates.

Tariffs, market access and the small print

• Duty-free treatment for the overwhelming majority of qualifying goods is carried over from NAFTA. What changed is the difficulty of qualifying.
• Certification of origin was simplified: no prescribed certificate is required, and origin may be certified on a commercial invoice by the importer, exporter or producer.
• De minimis thresholds were raised on American insistence for express carriers and e-commerce: Canada from C$20 to C$40 tax-free and C$150 duty-free, Mexico to US$50 tax-free and US$117 duty-free. Canadian and Mexican retailers objected to the exposure to untaxed parcels.
• Canadian dairy is the most-cited market-access concession. Canada retains supply management — production quotas, price setting and prohibitive over-quota tariffs — but granted expanded tariff-rate quota access and agreed to eliminate the Class 6 and Class 7 milk-price classes.
  • This produced the agreement’s first two state-to-state disputes: a 2021 panel found Canada’s reservation of quota shares for processors inconsistent with its obligations, and a second panel in 2023 largely upheld Canada’s revised allocation — dispute settlement does go against the United States.
• Government procurement coverage narrowed: Canada withdrew from the chapter entirely, relying on the WTO Government Procurement Agreement, so its obligations now run only between the United States and Mexico.

Automotive rules of origin and the labour value content rule

• The automotive rules are the heart of the agreement and the clearest expression of its philosophy.

• The labour value content rule is genuinely novel in trade law. No previous agreement conditioned tariff preference on the wage paid to the workers who made the good.
  • For passenger vehicles the threshold is 40%, for light trucks 45%, with credits for research and development and for engine and transmission assembly in high-wage plants.
  • The US$16 figure is a nominal amount fixed in the text, not indexed to inflation or wage growth, so its real value erodes automatically.
    • By the mid-2020s a growing share of Mexican skilled manufacturing had approached or crossed it, so a provision designed to be demanding is becoming less so — an argument for revision the United States has made in the review.
  • The intent is transparent: it is not to liberalise trade but to make low-wage production ineligible for the preference, pushing assembly toward the United States and Canada or pushing Mexican wages up.
• Rules of origin this strict are a tax on regional producers, not only a barrier to outsiders.
  • Compliance costs rose, and where meeting the 75% threshold costs more than the ordinary most-favoured-nation duty — low for many vehicles and parts — a producer may rationally pay the tariff and ignore the preference. Assessments by the US International Trade Commission and by industry find exactly this response in parts of the sector.
• The rules produced the agreement’s most significant legal dispute. Mexico and Canada challenged the American “roll-up” methodology — whether a core part that qualifies as originating counts as 100% originating in the vehicle’s own calculation. A panel ruled against the United States in January 2023, and compliance has been contested since. A won case that is not complied with tests whether the mechanism means anything.

Labour: Mexico’s reform and the Rapid Response Labour Mechanism

• Mexico’s labour law reform of May 2019 is the necessary precondition and is often omitted. It required that existing collective agreements be legitimised by a secret-ballot vote of the workers covered, created independent labour courts in place of the tripartite conciliation and arbitration boards, and established a Federal Centre for Conciliation and Labour Registration to run union elections and register contracts. Its purpose was to dismantle the protection-contract system.
• The Rapid Response Labour Mechanism (RRM) is the most enforceable labour provision in any trade agreement, and its design explains why.
  • It is facility-specific: a complaint concerns a named plant, not a country’s general record, which removes the political weight that made state-to-state labour complaints unusable.
  • It is fast, with timelines in weeks — a request for review, a period for the respondent state to investigate and remediate, then an independent rapid response labour panel.
  • The remedies bite at firm level: suspension of preferential tariff treatment for that facility’s goods, and denial of entry in repeat cases. The cost falls on the employer rather than the economy at large, which is why it is both usable and politically survivable.
  • The presumption on “affecting trade” from the 2019 protocol removes the evidentiary obstacle that made NAALC complaints hopeless.
  • It covers priority-sector facilities producing traded goods and is reciprocal in form, though nearly all cases have concerned Mexico.
• The record is substantial and its limits are equally clear.
  • Between mid-2020 and mid-2025 the United States initiated on the order of forty cases across roughly as many facilities — automotive and auto parts above all, and also mining, call centres, aviation, food processing, electronics and apparel.
  • Around two-thirds of concluded cases produced reinstatements of dismissed workers, several produced recognition of an independent union and a first collective agreement, and the United States Trade Representative estimated that some 42,000 workers had benefited by early 2025. Negotiated wage increases have run several points above inflation.
  • The limits: agriculture is excluded; coverage is confined to export-oriented facilities; Mexican labour authorities lack sanctioning power of their own; there is no duty to bargain in good faith once a union is recognised; and the workers reached are a small fraction of a manufacturing workforce above five million. Independent analysts find no measurable spillover to national wage levels, and enforcement depends on American funding and political will.

Environment brought into the core text

• The USMCA’s Chapter 24 puts environmental obligations inside the agreement, subject to the same state-to-state dispute settlement as any other chapter. This is the structural fix to the NAAEC’s central weakness.
• The obligations are broader than NAFTA’s: air quality, marine litter, sustainable fisheries management and a prohibition on subsidies contributing to overfishing — the last agreed years before the WTO’s own fisheries subsidies agreement — plus commitments on illegal wildlife, timber and fish trade, invasive species and ship pollution, and adherence to named multilateral environmental agreements.
• The CEC survives, with a submissions process carried forward on tightened admissibility: private remedies first, and no submission resting on media reports alone.
• What is absent is as telling as what is present: the chapter does not mention climate change. It also repeats the historically ineffective formula that it is “inappropriate” to attract investment by weakening environmental protection, rather than prohibiting it. This is a genuine structural improvement of modest substantive ambition.

Dispute settlement rebuilt

• Investor-state arbitration was eliminated between the United States and Canada after a three-year transition for legacy claims — Canada had been the most-sued NAFTA party and wanted it gone; American negotiators, sceptical of arbitration constraining American regulators, agreed.
• Between the United States and Mexico it was narrowed rather than removed.
  • Ordinary claims must now exhaust local remedies for at least thirty months and are confined to a reduced set of protections, excluding the expansive “fair and equitable treatment” and indirect-expropriation standards that generated the Chapter 11 case law.
  • A broader carve-out preserves fuller protection for covered government contracts in oil and gas, power, telecommunications, transport and infrastructure — where American capital is most exposed to Mexican state policy.
• Chapter 19 survives as Chapter 10, binational panel review of anti-dumping and countervailing duty determinations intact. Canada treated this as non-negotiable in 2018 and has treated it the same way since.
• Chapter 31 fixes the Chapter 20 paralysis by providing for automatic appointment of panellists from a standing roster where the parties fail to agree, so a respondent can no longer make a panel impossible. It also opens the process to third-party written submissions.
• The mechanism has been used more in five years than Chapter 20 was in twenty-six: panels have decided disputes on Canadian dairy quotas, American solar safeguards, automotive rules of origin and Mexico’s decree restricting genetically modified maize, the last decided against Mexico in December 2024, with Mexico repealing the disputed restrictions in early 2025. Compliance, however, has been uneven where the United States lost.

Digital trade and intellectual property

• The digital trade chapter is the clearest case of the agreement covering ground NAFTA could not have covered.
  • No customs duties on electronic transmissions, and no discriminatory treatment of digital products.
  • No forced data localisation: a party may not require computing facilities to be sited in its territory as a condition of doing business, and cross-border data transfer may not be restricted, subject to a legitimate-public-policy exception.
  • Intermediary liability protections modelled on American law, shielding platforms from liability for user content — a domestic legal settlement exported into a trade agreement, and one the European Union has moved away from.
  • Provisions on electronic authentication, open government data and source code.
• On intellectual property, standards rose: copyright of life plus seventy years, obliging Canada to extend its term, plus stronger trade-secret and anti-counterfeiting provisions.

Article 32.10 and the geo-economics of exclusion

• Article 32.10 requires notice at least three months before a party negotiates a free trade agreement with a “non-market economy”, disclosure of its objectives, and the text for review thirty days before signature.
  • If such an agreement is concluded, the other two parties may terminate the USMCA on six months’ notice and replace it with a bilateral agreement between themselves.
  • “Non-market economy” is defined by reference to a party’s own trade-remedy determinations — which for the United States includes China, and the provision is universally read as aimed at China.
• Legally, the clause adds little: any party could already withdraw on six months’ notice under Article 34.6, and triggering the substitution requires the other two to agree.
• Politically and analytically it matters a great deal, and it is the clearest single piece of evidence that the USMCA is a geo-economic instrument rather than a commercial one.
  • It converts a trade agreement into a membership condition in an economic bloc: access to the American market is made conditional on the alignment of external economic relations.
  • It has become a template. Similar clauses have appeared in later American trade arrangements, and using market access to shape a partner’s third-country policy is now a standard instrument.
  • The underlying anxiety is transshipment: that Chinese components, and later Chinese-owned plants in Mexico, use North American assembly as a back channel into the American market at preferential rates. Mexico’s imposition of tariffs on Chinese vehicles and its screening of Chinese investment are responses to American pressure on precisely this point.
  • Robert D. Kaplan has argued that the United States neglected the challenge to its position in its own hemisphere while attending to distant theatres. Recasting North American trade as a security question is that neglect being corrected in the most disruptive way available.

Article 34.7: the sunset and the review

• Article 34.7 gives the agreement a sixteen-year term. Absent extension, it terminates on 1 July 2036.
• A joint review is mandatory on the sixth anniversary of entry into force — 1 July 2026 — at which the parties confirm in writing, at the level of heads of government, whether they wish to extend for a further sixteen years.
• If any party declines to confirm, the agreement is not renewed but does not end. Instead Article 34.7.4 activates annual joint reviews, held every year until 2036, at each of which the parties may still confirm an extension. Confirmation at any point restarts a fresh sixteen-year term.
• It was an American demand, opening at a hard five-year expiry and moderated to sixteen years with a review because Canada, Mexico and business argued that an agreement expiring inside the payback period of an automotive plant would deter the investment it was meant to attract.
• The criticism was, and remains, that a built-in expiry undermines the very function the agreement performs. The value of a trade treaty for a smaller partner lies in its irreversibility; an agreement that must be re-consented to periodically supplies exactly the uncertainty it was meant to remove.

Managed trade, not deeper integration

• The USMCA fixed NAFTA’s enforcement deficit and did not fix its distributional or developmental one. Every criticism about the absence of a cohesion fund, the neglect of southern Mexico, the exclusion of agriculture from labour enforcement, and the asymmetry of a three-party agreement with one dominant member applies to the new text exactly as it applied to the old.
• Its economic philosophy is different in kind.
  • Raising a content threshold from 62.5% to 75%, adding a wage floor as a condition of preference and requiring steel to be melted and poured in the region are not steps toward freer trade. They are instruments that manage where production happens — closer to the Auto Pact’s conditional access than to the liberalising instinct of 1994.
• It is therefore best read as a template for the protectionist turn rather than as deeper integration, and it has been read that way by everyone drafting agreements since: labour value content, non-market-economy clauses, melt-and-pour rules and sunset provisions have all been proposed elsewhere on the American model.
• The counter-argument deserves full statement. Enforceable labour standards attached to trade preferences are something no other regional arrangement, the European Union included, has achieved against a partner with weaker institutions.
  • The EU exports labour standards through accession conditionality, which works only on states seeking membership and stops the moment they join. The USMCA’s mechanism works on a sovereign, non-acceding partner, at factory level, in weeks.
  • The reply is that the same instrument serves an American interest in raising a competitor’s costs — a provision can be simultaneously a genuine advance in labour rights and an instrument of industrial protection.

2025–26: tariffs, the joint review and the bilateral turn

The tariff environment

• From early 2025 the United States imposed tariffs on Canada and Mexico under the International Emergency Economic Powers Act (IEEPA), tied rhetorically to fentanyl and migration rather than trade, with USMCA-qualifying goods exempted. The exemption raised compliance sharply: most Canadian and Mexican exports were certified as originating and around 85% of North American trade continued duty-free.
• The Supreme Court struck the IEEPA tariffs down in Learning Resources, Inc. v. Trump, decided 20 February 2026, holding 6–3 that IEEPA does not authorise the President to impose tariffs: the statute never mentions duties, and reading “regulate… importation” as a taxing power would not survive the Export Clause. A plurality invoked the major questions doctrine. Thousands of refund claims followed.
• Sectoral tariffs under Section 232 of the Trade Expansion Act were untouched by the ruling, and they are Canada’s and Mexico’s principal grievance.
  • Steel and aluminium at 50%, raised from 25% in June 2025, with derivative products covered.
  • Automobiles and parts at 25%, with USMCA-qualifying vehicles taxed only on their non-American content — a mechanism that is itself an origin-based industrial policy.
  • Softwood lumber, plus furniture and cabinets; medium and heavy trucks from late 2025; copper at 50%.
• The administration substituted other authorities after the ruling, including a temporary across-the-board tariff under Section 122 of the Trade Act of 1974, capped at 15% for 150 days, and new Section 301 investigations. The legal position has moved repeatedly and any figure needs checking against the date it is used.

The 1 July 2026 joint review and what followed

• At the mandatory joint review on 1 July 2026, United States Trade Representative Jamieson Greer announced that the United States “did not agree to renew the USMCA in its current form”, and that the agreement was accordingly not renewed. Canada and Mexico both supported a sixteen-year extension.
• The precise consequences matter, and they are widely misreported.
  • The agreement remains fully in force. Tariff preferences, rules of origin, labour and environment chapters and dispute settlement all continue to operate. “Not renewed” is not “terminated”, and the USMCA has not expired.
  • Article 34.7.4 annual joint reviews are now running, every year to 1 July 2036.
  • Extension remains available at any time by written confirmation of the three heads of government, without formal renegotiation.
  • Only if no extension is confirmed by 2036 does the agreement lapse.
• The United States has pursued separate bilateral tracks rather than trilateral negotiation, which is itself the most significant structural development.
  • Three US–Mexico bilateral rounds ran from May to July 2026 on automotive rules of origin — Washington seeking content thresholds above 80% and an American-specific sourcing requirement — plus steel and aluminium coordination against Chinese material, energy, agricultural biotechnology and labour enforcement. Mexico has been accommodating on security and migration, resistant on content rules and energy.
  • Canada–United States talks collapsed on 21–22 August 2026. Washington had offered tariff relief on steel, aluminium, autos and lumber in exchange for supply-chain and enforcement commitments and formal USMCA renegotiation.
    • Prime Minister Mark Carney said the United States “asked too much and offered too little” and that late changes to the terms were “uneconomic, unfair”.
    • The United States imposed 50% tariffs on around $20 billion of Canadian goods, and Canada announced dollar-for-dollar retaliation from 8 September 2026.
• Both smaller partners have responded with diversification rather than concession alone.
  • Canada has concluded agreements with Indonesia, Ecuador and the United Arab Emirates, joined the EU’s SAFE defence-procurement instrument as its only non-European participant, and is negotiating with Mercosur, ASEAN, Turkey and India; non-American exports rose to $333 billion in 2025. Carney’s formulation that the old relationship of steadily increasing integration “is over” is a statement about regionalism, not only tariffs.
  • Mexico’s Plan México, launched 13 January 2025, aims at the world’s ten largest economies by 2030 through investment above 25% of GDP, 1.5 million manufacturing jobs and a 15% rise in domestic content. Its first-year results were poor — investment fell as a share of GDP and manufacturing employment declined — because nearshoring cannot proceed while the terms of access are unsettled.

What the episode means for regionalism

• An agreement with a built-in expiry, reviewed annually, negotiated bilaterally by its largest member, is a different kind of object from the European Union’s treaties, and the difference is not one of degree.
• Regional agreements can be de-institutionalised without being abolished. The USMCA is fully in force and simultaneously less certain than at any point in its life, because certainty was never located in the text; it was located in the expectation of renewal.
• Bilateralism inside a trilateral framework dissolves the region as a unit. Where the largest member negotiates separately with each partner, the smaller two cannot combine, and the agreement’s formal trilateralism becomes a container for two asymmetric bargains.
• Trade agreements are now instruments of security policy. Fentanyl, migration enforcement, defence spending, investment screening and alignment against China have all entered a negotiation nominally about tariffs — the strongest available evidence for the geo-economic reading of contemporary regionalism.

India and the North American trading order

• India has no agreement with any USMCA party, and its exposure to North America is therefore governed entirely by unilateral American, Canadian and Mexican measures — which is precisely why the USMCA’s design matters to Indian policy.
• The United States is India’s largest single-country trading partner, with two-way goods and services trade of roughly $239 billion in 2025 and a large Indian surplus, concentrated in pharmaceuticals, engineering goods, gems and jewellery, textiles and IT services.
  • The proposed India–US bilateral trade agreement has been repeatedly delayed; the Learning Resources ruling unsettled it further by removing the tariff authority that framed the negotiation, and talks paused in early 2026.
  • India–Canada goods trade is modest, roughly $9 billion in 2024–25, hostage to the political rupture of 2023–24 and the suspension of CEPA talks now cautiously resumed.
  • India–Mexico trade is larger — around $14–15 billion, Mexico being among India’s leading partners in the Americas, with Indian pharmaceutical, auto-component, steel and IT firms established there.
• Indian exporters are exposed to North American rules of origin in a specific and underappreciated way.
  • An Indian auto-component maker supplying a Mexican assembly plant contributes non-originating content, counting against the 75% threshold and against a labour-value requirement it can never satisfy. Stricter regional content rules are a barrier to Indian suppliers even though India is not a party.
  • Indian firms that invested in Mexico on nearshoring logic now face terms of access that can change annually, and Indian steel, aluminium and derivative exports face the Section 232 tariffs directly, at rates set without reference to any negotiation with India.
  • Mexico’s own response has hit India directly: from 1 January 2026 Mexico raised tariffs on some 1,400 product lines from non-FTA partners to between 5% and 50% — textiles, apparel, steel, aluminium, tyres, chemicals and leather — affecting close to $2 billion of Indian exports. A country squeezed inside a bloc passes the squeeze outward, and India, with no agreement, absorbs it.
• Three lessons follow for India’s own trade policy, and they are being applied.
  • Labour and environment clauses are now enforceable, not decorative. The USMCA’s facility-level mechanism, the EU’s carbon border adjustment and forced-labour rules and the sustainability chapters of recent European agreements all point one way. India’s position that labour standards belong at the ILO rather than in trade agreements must now be defended inside negotiations rather than at their threshold.
  • Non-market-economy clauses are a demand India should expect. Article 32.10 is the model for asking a partner to disclose or forgo negotiations with China, and India — in the SCO and BRICS, and carrying a large Chinese trade deficit — has an obvious interest in refusing to trade strategic autonomy for market access.
  • Sunset and review mechanisms deserve deliberate treatment. The India–UK Comprehensive Economic and Trade Agreement, signed 24 July 2025 and in force 15 July 2026, and the India–EU agreement concluded on 26 January 2026 and expected in force in 2027, are open-ended agreements with review provisions rather than expiry dates. A periodic review keeps an agreement current; a periodic expiry makes it a hostage.
• The USMCA model — managed, conditional, reviewable trade, with access tied to production location, wage floors, third-country alignment and security cooperation — is not an American aberration but the template of the present period, and Indian negotiators meet versions of it in every negotiation they have open.

Conclusion

• North America is the strongest case in world politics that deep economic integration does not require deep institutions — and the strongest case that integration without institutions is fragile.
• NAFTA produced continental supply chains and a tripling of trade while creating no court, no fund, no citizenship and no procedure for revisiting itself. It delivered aggregate gains that were real and modest, and concentrated losses that were real and severe, and it left the wage convergence its advocates promised undelivered.
• The USMCA answered the enforcement criticism and left the developmental one untouched. Its rules of origin, wage floor, non-market-economy clause and sunset provision describe a different project: not the removal of barriers among three economies, but the direction of production within a bloc defined against a fourth power.
• The 2026 joint review completes the picture. An agreement that remains fully in force while its largest member declines to renew it, reviews it annually and negotiates its future bilaterally is a regional arrangement in form and a set of bilateral bargains in substance. Whether it can hold to 2036 is now the open question of North American regionalism.