index.php

Pinar Gunes

View profile

Gunes, P. (2026) The United States’ forced-labour tariffs: Who really pays for them?, CITP Briefing Paper 35

Key points

• The newly implemented US tariffs cause the largest welfare losses among smaller and more exposed economies, including Costa Rica, Vietnam and Ireland.
• The UK fares somewhat worse than the EU average; this could be due to market size and/or the UK’s greater exposure to US demand.
• Whilst the US gains, the gain is modest because the tariff applies to only roughly 38% of US trade, and US imports are only 14% of GDP, and it also masks substantial differences across industries, with some sector output falling significantly.
• When a country retaliates to the US tariffs alone, it cannot meaningfully shift the outcome: its own welfare barely improves, its exports to the US fall further, and the US remains a net gainer in every case.
• When partners retaliate together, the US terms-of-trade advantage becomes a loss of −0.43 per cent, and several partners that lost under the unilateral tariff move into modest positive territory.
• This creates a collective-action problem. Effective retaliation requires countries with very different exposures, interests and preferred targets to coordinate their response.

Pinar Gunes

Background

Since returning to office in January 2025, the Trump administration has made tariffs a central instrument of US economic policy. It began by imposing broad duties under the International Emergency Economic Powers Act (IEEPA), including the so-called fentanyl tariffs on Canada, Mexico and China and the much wider “Liberation Day” reciprocal tariffs announced in April 2025. In February 2026, however, the Supreme Court ruled that IEEPA did not provide the President with the authority to impose those tariffs, invalidating one of the administration’s main trade-policy tools. The administration then shifted to other statutory authorities: a temporary 10% global import surcharge under Section 122, which allows temporary import restrictions to address serious balance-of-payments problems, and existing Section 232 national-security tariffs on products including steel, aluminium, copper and automobiles. The Office of the United States Trade Representative (USTR) subsequently launched Section 301 investigations into issues including industrial overcapacity and forced-labour enforcement. It is this final measure that concerns us here.

In July 2026, the United States (US) imposed additional tariffs of 10% to 12.5% on goods from 60 trading partners, following a forced-labour investigation under Section 301 of the Trade Act of 1974, which authorises action against foreign practices judged to be unfair or discriminatory 1. On paper, this looks like another turn of the tariff ratchet. But the economics of this action differ from those of blunt, across-the-board duties that have dominated the debate. Its effective coverage is narrower, heavily qualified by exemptions and origin-specific — and those design features, far more than the headline rate, determine who actually pays.2

This analysis applies a quantitative trade framework3 to assess the short- and long-run trade welfare effects of the newly implemented US tariffs4 and identify who might bear their costs. In considering the impact of tariffs, what matters is not simply how much the bilateral tariff has changed, but how much it has changed relative to other countries, whose tariffs may also be changing. The framework thus traces how tariffs ripple across countries and sectors and, crucially, distinguishes5 immediate effects from those that emerge once trade patterns have fully adjusted.

The aim of this Briefing Paper is to analyse how the effects vary across countries and over time, what might be the impacts across different US sectors and to what extent trading partners can mitigate the consequences of US policy through retaliation. The analysis also considers in more detail the effects on the UK.

All the results are based on running simulations using what is technically known as a ‘New Quantitative Trade Model’ (NQTM). The model comprises 77 countries and 22 sectors and thus allows for the interactions at the sectoral and country level.

Winners and losers

Table 1 summarises the effects of the forced-labour tariffs on welfare and real US imports for ten selected economies including the US in the short run and long run. The two measures capture different dimensions of the impact. The changes in US imports show how much more or less each country sells to the US, while welfare captures the broader change in real income after changes in import prices, export earnings and tariff revenue are taken into account.

The nine partners are chosen to illustrate the main patterns in the results: large, diversified exporters (China, Germany and Japan), United States-Mexico-Canada Agreement (USMCA) partners (Canada and Mexico), the UK, and smaller, more US-exposed economies with relatively large welfare losses (Costa Rica, Vietnam and Ireland).

The US gains through the classic ‘terms-of-trade’ channel. As a large buyer, it can shift part of the tariff burden onto foreign exporters, who respond to the tariff by lowering their prices. The tariff also generates revenue that is assumed to be returned to US households. Together, these effects increase US welfare by 0.32 per cent in the short run.

The gain is modest because the headline tariff applies to only roughly 38% of US trade, and US imports are only 14% of GDP. The measure excludes steel, aluminium and copper already covered by Section 232— the national-security provision under which the US separately taxes those metals; goods from Canada and Mexico entering duty-free under USMCA; pharmaceuticals; civil aircraft; informational materials; and numerous products listed in country-specific annexes. Once these exemptions are taken into account, the trade-weighted tariff increase in most sectors is only a fraction of the headline rate.

Table 1. Welfare and US-import effects of the forced-labour tariffs

Notes: Model estimates percentage changes. Exports to the US refer to the change in the real value of each partner’s exports to the United States. The final column is the share of each partner’s total manufacturing exports going to the United States in the 2022 model baseline. SR = short run; LR = long run. No export-to-US figure applies to the United States itself. Canada A reflects the forced-labour tariff only; Canada B additionally includes the US Section 338 tariffs on Canada (a 50 per cent duty on approximately US$20 billion of dairy, alcohol and motor-vehicle products).

On the other hand, the burden across trading partners depends not only on the decline in their US sales but also on their exposure to the US market and their ability to redirect exports elsewhere. China, for example, loses more than one-tenth of its US sales but experiences only a small welfare loss because the US accounts for a smaller share of its overall market as given in the last column of the table (16.8%). Germany and Japan show a similar pattern. Costa Rica, by contrast, suffers large declines in both US sales and welfare because it depends more heavily on US demand and has fewer alternative markets. More generally, the largest welfare losses are concentrated among smaller and more exposed economies, including Costa Rica, Vietnam and Ireland.

Canada and Mexico illustrate how preferential treatment can alter the effects that a tariff would otherwise have. Goods entering the US duty-free under USMCA are exempt from the forced-labour tariff. This largely shields Mexico and makes its exports relatively more competitive as US demand shifts away from those suppliers facing higher tariffs. US imports from Mexico therefore rise, and Mexico records a modest welfare gain. Canada still experiences a welfare loss, however, showing that preferential access does not eliminate all adjustment costs. The precise effects depend partly on how often eligible exporters claim duty-free USMCA treatment. However, the underlying mechanism is clear: a preferential carve-out can substantially change who gains and who loses. Note that the table reports two rows for Canada to separate two distinct US actions6. Canada A shows the effect of the forced-labour tariff alone, from which most Canadian goods are shielded under USMCA; the small welfare loss reflects only limited residual exposure and indirect effects. Canada B adds the separate Section 338 tariffs7 the US imposed on Canada in August 20268 — an additional 50 per cent duty on roughly US$20 billion of dairy, alcohol, motor-vehicle and other products, which are not exempted under the USMCA. Facing both measures, Canada B records a long-run welfare loss of −0.58 per cent, roughly double that under the forced-labour tariff alone, and its exports to the US fall rather than rise.

The effects also change over time. Because the tariffs are imposed unilaterally and other countries do not retaliate, affected exporters can redirect sales towards markets that remain open. In the model, this is captured by a higher elasticity of substitution in the long run. These elasticities are estimated using bilateral manufacturing trade data for 170 countries over 1995–2022.9 The short-run elasticity captures the immediate response of trade flows to a tariff change, while the long-run elasticity incorporates the adjustment period. Welfare losses remain broadly similar in the long run for most of the countries reported here; the US gain, however, declines from 0.32 to 0.26 per cent because exporters with better alternatives are less willing to lower their prices for the US market, weakening the terms-of-trade benefit.

Which US industries are most affected?

The modest aggregate US welfare gain also masks substantial differences across industries. Given a fixed total labour supply, aggregate US real output remains essentially unchanged. The main domestic effect is therefore a reallocation of production: some industries expand as workers and resources move away from sectors that contract. The small aggregate effect should therefore not be interpreted as meaning that the effects are small everywhere. At the sector level, the adjustments can be substantial: in some sector-country combinations, real output falls by 3 per cent or more. Changes of this magnitude can have meaningful consequences for firms and workers in the affected industries, even when they are largely obscured in economy-wide averages. Table 2 shows the five largest expansions and contractions, measured as percentage changes in real output.

The expanding sectors are primarily import-competing consumer-goods industries, including textiles, apparel, leather and furniture. As the tariff raises the price of affected foreign goods, US demand shifts towards domestically produced alternatives. These industries consequently expand, with the effects becoming more pronounced as the economy adjusts over time.

The contracting sectors tend to rely more heavily on foreign sales. These include basic metals, chemicals, communications equipment, office and computing equipment, and other transport equipment. For these industries, higher input costs and changes in domestic and foreign demand can outweigh the protection provided by the tariff.

The scale of these responses also depends on sector-specific trade elasticities, which measure how readily buyers switch between suppliers when relative prices change. Hence, sectors with higher elasticities, such as basic metals and other transport equipment, experience larger changes in trade and production, while lower-elasticity sectors,10 such as motor vehicles and tobacco, adjust less.

The long-run trade elasticity is larger in every sector as firms need time to reorganise supply chains, find alternative suppliers and redirect sales. The model does not explicitly simulate firms reorganising supply chains, finding alternative suppliers or redirecting sales; instead, these longer-run adjustment possibilities are captured through the estimated long-run elasticities.

Table 2. Largest changes in US sectoral real output (%)

Notes: Percentage change in real output. Top five rows expand; bottom five contract. Aggregate US real output is essentially unchanged following Gunes (2026) — these are compositional shifts. Trade elasticities are estimated in Gunes (2026) using bilateral manufacturing trade data for 170 countries over 1995–2022. SR is the impact elasticity; LR is the fully adjusted elasticity implied by the estimated persistence of trade flows.

Where does the UK stand?

The UK is one of the 60 targeted economies and faces the slightly lower headline tariff rate of 10%. The action also includes UK-specific product carve-outs, including a broad medical-device exclusion. Even so, when it comes to the consequences, the UK fares somewhat worse than the EU average. It is one of the few large, advanced exporters among the more affected economies: its estimated welfare loss is greater than Germany’s or Japan’s, and its US sales fall sharply in the long run.

Part of the explanation may lie in market size rather than anything specific to the tariff. When US demand contracts, EU members can draw on a large internal market and redirect trade among neighbouring countries facing the same shock; an economy outside such a bloc has fewer straightforward opportunities to divert its exports. The UK’s position outside the EU may therefore be one factor, although establishing this mechanism would require further analysis. More broadly, a country’s capacity to absorb an external shock depends partly on the size of the preferential market to which it has access.

Another potential explanation is the UK’s greater exposure to US demand. In the 2022 model baseline, 15.9% of UK manufacturing exports went to the US compared with 9.9% for Germany. This suggests that the UK is more exposed to a contraction in US demand than at least some major EU exporters (France 9.6% or Italy 11.9%). Its position outside the EU may further limit the scope for redirecting trade within a large preferential market, but the model does not separately identify this channel. It should therefore be treated as a plausible contributing factor rather than a causal estimate.

At the sector level, the UK results reveal considerable variation beneath the aggregate loss. Table 3 reports long-run changes in real output for the ten sectors that contract most sharply in the UK, together with the five that fare best, and sets these against the corresponding changes in Germany, Mexico, Canada and China for comparison. The hardest-hit UK sectors are concentrated in machinery, textiles, electrical machinery, leather and footwear, and communications equipment, where output falls by between 0.6 and 1.5 per cent as US demand shifts away from taxed suppliers and competition in third markets intensifies. At the other end, a handful of sectors expand: coke and petroleum, printing and publishing, basic metals and other transport equipment each gain as resources are reallocated towards activities less exposed to the tariff.

The cross-country columns show that these are not universal patterns. Several sectors that contract for the UK expand for Mexico, whose USMCA-exempt access makes its exports relatively more competitive precisely where others are penalised, while the effects for Germany and China are generally more muted, reflecting their larger and more diversified markets. More broadly, the cross-country comparison also shows that modest aggregate welfare effects can conceal much larger sectoral adjustments: in some sector-country combinations, real output changes by around 3 per cent or more. Changes of this magnitude can have meaningful consequences for firms and workers in the affected industries.

Table 3. Sectoral real output by country, long run (%)

Notes: Percentage change in real output under the US forced-labour tariff (long run). Sectors are ordered by the UK’s results — its ten largest declines, then its five largest gains. The same sectors that contract for the UK often expand for Mexico as production shifts towards the USMCA-exempt market. All figures are central-scenario model estimates.

What if partners retaliate?

The results so far assume that the US acts alone and that no partner responds. Although partners have not retaliated against the forced-labour tariff so far, retaliation remains a possibility, as the US continues to threaten new tariffs. Two further sets of simulations, following the structure of Table 1, report the long-run result. See Table 4. In the first simulation, every targeted partner simultaneously imposes the same tariffs on US exports and on the same sectors that the US has levied on them (“all retaliate”). In the second simulation, each partner retaliates on its own, one at a time, while the US measure remains in place against everyone (“retaliates alone”). The original US-only long-run welfare effect from Table 1 is shown for reference.

The contrast is stark. When partners retaliate together, the US terms-of-trade advantage disappears: the American welfare gain of +0.26 per cent becomes a loss of −0.43 per cent, and several partners that lost under the unilateral tariff — Canada, the UK, Germany and Japan — move into modest positive territory. However, when a partner retaliates alone, it cannot meaningfully shift the outcome: its own welfare barely improves, its exports to the US fall further, and the US remains a net gainer in every case. Retaliation alters the US calculus only when it is collective; acting alone, no single partner has the market power to turn the terms of trade back in its favour. This, however, creates a collective-action problem: while countries may benefit from responding together, coordinating such a response is difficult because governments differ in their exposure to US tariffs, their willingness to escalate, and the products they would prefer to target.

Canada again appears twice. Canada A retaliates by mirroring the forced-labour tariff alone, whereas Canada B mirrors the full set of US measures it faces, including the Section 338 tariffs. Because Canada B is hit by both US actions, it stays worse off than Canada A across every column: even retaliating — collectively or alone — does not restore it to the near-neutral position that the lightly-taxed Canada A reaches when others retaliate.

Table 4. Retaliation — long-run effects (%), mirroring Table 1

Notes: Long-run percentage changes only, reported to mirror Table 1. “US-only” repeats the baseline forced-labour tariff (Table 1, long run). “All retaliate” has every targeted partner simultaneously levy the same rates, on the same sectors, on US exports. “Retaliates alone” is a separate counterfactual for each country, in which only that country retaliates while the US tariff stays in place against all partners; the final column gives that country’s own change in real exports to the US. Canada A mirrors the forced-labour tariff only; Canada B mirrors the combined forced-labour and Section 338 tariffs it faces.

What are the implications for policymakers?

The headline rate tells only part of the story. A tariff’s economic impact depends on which products and countries it targets, how strongly affected sectors respond, how well targeted countries can absorb the shock, and who else is being targeted and by how much.

At the sector level, the response depends partly on the trade elasticity — in other words, on how easily buyers can switch between suppliers when relative prices change. When switching is easy, with trade elasticity being high, a tariff can sharply redirect trade and production. On the other hand, when suitable domestic or foreign alternatives are limited, buyers remain tied to their existing suppliers, so the tariff has a smaller effect on trade flows and output and a larger effect on prices.

At the country level, being targeted does not necessarily lead to a large economy-wide loss. The outcome depends on how reliant the country is on the tariff-imposing market and whether its exporters can redirect sales elsewhere. Smaller economies that depend heavily on US demand tend to bear greater costs. Larger, more diversified economies, such as China, by contrast, can lose substantial US sales without experiencing a similarly large welfare loss.

The time horizon also matters. The immediate effects of a tariff can differ substantially from its long-run consequences as trade responds more strongly once adjustment has had time to occur. It is important to note, however, that the model used here does not directly trace firms reorganising supply chains, buyers searching for new suppliers, or exporters entering alternative markets. Rather, it compares counterfactual outcomes using estimated short- and long-run trade elasticities, which capture differences in the degree of adjustment across horizons without explicitly modelling the transition path itself.

Retaliation changes the picture, but only when it is sufficiently broad and coordinated. A single partner striking back does little to alter the US calculus: although retaliation may improve that partner’s position somewhat, it does not overturn the US terms-of-trade advantage. When many partners retaliate together, by contrast, the US gain becomes a loss. This creates a collective-action problem. Effective retaliation requires countries with very different exposures, interests and preferred targets to coordinate their response, making precisely the form of retaliation that has the greatest economic leverage the most difficult to organise.

Finally, small aggregate effects should not be mistaken for small effects everywhere. In the US, aggregate real output remains essentially unchanged in the model, but this masks substantial reallocation across industries. Looking across countries as well, particular sector-country combinations experience changes in real output of around 3 per cent or more. Changes of this size can have meaningful consequences for firms and workers in the industries concerned even when economy-wide effects are modest. The distributional consequences of a tariff can therefore be significant even when its aggregate consequences are not. For policymakers, the implication is clear: tariffs cannot be assessed by their headline rates or target lists alone. Product coverage, exemptions, trade elasticities, dependence on the importing market, access to alternative destinations, the time allowed for adjustment and the risk of retaliation all determine where a tariff ultimately bites.

Footnotes

1. Office of the United States Trade Representative (USTR) (2026), ‘Notice of Actions in Section 301 Investigations of Acts, Policies, and Practices of Various Economies Related to the Failure of Each Economy to Impose and Effectively Enforce a Prohibition on the Importation of Goods Produced with Forced Labor’, Federal Register, 91 FR 47318, 28 July.
2. This analysis covers the forced-labour action alone. Brazil’s separate 25% Section 301 tariff took effect on 22 July 2026, while the 50% Section 338 duties on specified Canadian goods are scheduled to take effect on 19 August 2026. Both measures have their own product lists and, where coverage overlaps, apply in addition to the forced-labour duties.
3. Developed in Gunes, P. (2026), “When Tariffs Bite: The Short-Run vs Long-Run Welfare Effects of Trade Wars.” Available at: /publications/when-tariffs-bite-the-short-run-vs-long-run-welfare-effects-of-trade-wars
4. The tariff measures, specified at the eight- and ten-digit HTS level, were first reduced to six-digit HS codes to match the available 2022 US import data. These codes were then converted from the HS 2022 classification to HS 2007 and subsequently mapped to two-digit ISIC Rev. 3 sectors using the relevant product concordances. For sectors containing both affected and exempt products, 2022 US import values were used to calculate the share of imports subject to the tariff. The tariffs imposed by the US are sourced from the United States International Trade Commission (USITC), Harmonized Tariff Schedule of the United States (2026), Revision 13, published 28 July 2026.
5. Before turning to the results, it is important to note that these are model-based estimates calibrated to 2022 trade data. The 2022 baseline is used because it is the latest year for which sufficiently comprehensive bilateral trade and tariff data are available to conduct the general-equilibrium analysis. This should not materially affect the qualitative results, however, because the model is designed to isolate the effects of the tariff shock itself rather than reproduce current trade levels exactly. Regardless of the baseline year, the estimates should therefore be interpreted as indicating the likely direction and broad magnitude of the effects rather than as precise forecasts.
6. The White House (2026). Proclamations imposing additional duties on certain products of Canada under Section 338 of the Tariff Act of 1930 (effective 22 August 2026). Washington, DC.
7. The White House (2026). Proclamations imposing additional duties on certain products of Canada under Section 338 of the Tariff Act of 1930 (effective 22 August 2026). Washington, DC.
8. The product coverage and rates of Canada’s counter-tariffs are sourced from Department of Finance Canada (2026), List of products from the United States subject to counter-tariffs effective September 8, 2026, Government of Canada, Ottawa, available at: https://www.canada.ca/en/department-finance/news/2026/08/list-of-products-from-the-united-states-subject-to-counter-tariffs-effective-september-8-2026.html (accessed 1 September 2026). The US–Canada trade values at the commodity level, which weight the counter-tariff lines to compute each sector’s effective duty, are sourced from United States International Trade Commission (2026), USITC Interactive Tariff and Trade DataWeb: U.S. total exports to Canada, 2022, by Harmonized Tariff Schedule (HTS) commodity, FAS value, Washington, DC, available at: https://dataweb.usitc.gov (accessed 1 September 2026).
9. Gunes, P. (2026), “When Tariffs Bite: The Short-Run vs Long-Run Welfare Effects of Trade Wars.” Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7167799
10. For a more detailed explanation and analysis of how trade elasticities are estimated, see Gunes (2026).