The impact of US tariffs on EU industries: results from a global input-output model (OeNB Bulletin Q3/25)

Martin Schneider, Richard Sellner / OeNB Economic Analysis and Research Department, Business Cycle Analysis Section konjunktur@oenb.at .

JEL classification: C67, F13, F14

Keywords: US tariffs, European industries, trade war, input-output analysis

In this study, we analyze the economic effects of US tariff increases on EU industries during President Trump’s second term. We simulate three different scenarios of escalating tariffs and find that EU GDP could decline by up to 0.5% in the medium term. Our simulations show a wide range of impacts for different industries, with pharmaceutical manufacturing being hit hardest. Ireland, Denmark, Belgium and Germany are among the most affected countries according to our analysis.

Highlights

Opinions expressed by the authors of studies do not necessarily reflect the official viewpoint of the Oesterreichische Nationalbank or the Eurosystem.

Abstract

We simulate the impacts of the US tariff increases of President Trump’s second term (cut-off date May 12) on the EU industries using detailed data at the product group level on tariffs, tariff elasticities and a global input-output model. We simulate three different tariff scenarios. In the most severe scenario with 25% tariffs on US imports from the EU for steel, aluminum, automobiles and automobile parts and 20% on all other products including pharmaceuticals, GDP in the EU would fall by 0.5% in the medium term (five years) against the baseline of no tariff increases. The pharmaceutical manufacturing industry is particularly affected, with output declining by 10.4%. Other heavily affected industries are: manufacture of other transport equipment, of basic metals and of motor vehicles. Looking at individual countries, Ireland is expected to be hit hardest, followed by Denmark, Belgium and Germany. The results are subject to several limitations. Investment and consumption restraint due to high uncertainty, trade diversion of Chinese goods to the EU and relocation of production to the US could further dampen economic activity in the EU on top of the direct trade effects estimated in this study.

1 A short timeline of the US trade war of 2025 so far

Since the start of his second term in office on January 20, 2025, US President Trump has introduced numerous new US import tariffs. In the following, we outline the scope of some of these tariffs that are relevant for the EU countries. Those are the tariffs on steel, aluminum, automobiles, automobile parts and the broad-based tariff increases of April 2, 2025. ¹ The 25% tariffs on steel and aluminum products came into effect on March 12, covering all countries exporting to the US. On April 3, the 25% tariff on automobiles took effect and was extended to also cover automobile parts on May 3. Imports from Canada and Mexico satisfying the rules of origin requirements of the United States-Mexico-Canada Agreement (USMCA) were exempt from these tariffs. The escalation gained momentum on April 2 (“liberation day”), when President Trump imposed additional tariffs of 10% on a broad range of products from all major trading partners of the US, effective as of April 5. The official aim of these tariffs is to reduce the bilateral trade deficits of the US with its trading partners via “reciprocal tariffs.” To achieve this, even higher country-specific tariff rates, replacing the 10% tariff for 57 trading partners ² , came into force on April 9. This increased the tariffs for goods imported from the EU to 20%.

Table 1

However, several countries and product groups were exempt from these tariffs. Among the excluded countries are Canada, Mexico, Russia, North Korea, Belarus and Cuba. ³ Additionally, the specific products listed in annex II of the executive order (including energy, semiconductors, copper, pharmaceuticals and lumber) and products already subject to tariffs under other decrees (steel, aluminum, automobiles and automobile parts) were exempt. On April 11, a presidential memorandum extended the list of exempt products to also cover smartphones and electronic devices containing semiconductors in general. Yet, some products were subject to multiple tariffs. For instance, automobile parts made of aluminum or steel were to face a tariff rate of 50%. An executive order ⁴ on April 29 clarified that the tariffs on aluminum, steel and automobiles / automobile parts do not stack. ⁵

On April 9, President Trump partially backtracked in light of the turmoil in financial markets and lowered the April-2 tariffs on imports from all countries except China to 10% for 90 days. This lowered the additional tariff for the EU from 20% to 10%. The US President also announced that, should the EU fail to reach an agreement with the US by the end of those 90 days, the tariffs were going to revert to 20%. We also expected additional tariffs on previously exempt products in this case. On April 1, the Secretary of Commerce launched an investigation to determine the national security impact of the importation of drugs and pharmaceutical ingredients. The International Emergency Economic Powers Act (IEEPA) authorizes the US President to act in response to threats to US national security in the event of a national emergency. This includes the imposition of tariffs. Thus, we expected the removal of exemptions for pharmaceutical products after the end of the 90-day suspension period.

Since the tariffs described above affect the EU directly, we compiled a detailed list of affected products from the US 2025 Harmonized Tariff Schedule (HTS) Revision 12 (May 13, 2025). ⁶ For the tariffs on US imports from Canada, Mexico and China, we relied on a few assumptions to simplify the analysis.

Since February, almost all products imported from Canada and Mexico are subject to a 25% tariff unless they comply with the rules of origin requirements of the USMCA trade agreement. We set the mean tariff rate for Mexican and Canadian products to 25% times the share of non-USMCA conforming exports. As an estimate for these non-conforming exports we use 50% for Mexico and 62% for Canada as was communicated by a White House official (see CNBC news coverage ).

Tariffs for products from China were increased by 10% in early February and then by another 10% in early March 2025. Since China retaliated against the tariffs from April 2, the US President increased the additional tariffs on imports from China from an initial 34% to ultimately 125%. Furthermore, the US ended the de minimis exemptions (i.e. duty-free status) of low-value shipments from China. On May 12, China and the US reached an agreement which lowered the April-2 tariff rate imposed on US imports from China to 10%.

2 Description of scenarios

The above timeline demonstrates that so far during Trump’s second term, his trade policy has been very dynamic. To reflect the rapidly changing environment we developed three scenarios that we consider to be relevant and likely. In all three scenarios, we assume that the tariffs on steel, aluminum, automobiles and automobile parts will stay at the levels of our cut-off date (May 12). This amounts to additional tariffs of 25% for all countries (including EU countries) and an additional 45% for China (which also include the 20% from February and March). Furthermore, we do not assume any retaliatory measures from trading partners and have not included the retaliation measures already implemented by China against the US. Note that while our analysis focuses on the results for EU countries, all simulations also include indirect impacts of the tariff increases on non-EU countries.

The first scenario (S1 “90 days”) represents the status quo (cut-off date: May 12). Pharmaceutical products are exempt and all countries (including EU countries and China) benefit from the 90-day relief from the April-2 tariffs. Therefore, all products imported to the US from the EU are subject to an additional 10% tariff rate, all imports from China to an additional 30%. Scenario 2 (S2 “Liberation”) depicts the status quo after the 90-day relief. In this scenario the country-specific “reciprocal” tariff rates are applied to all products covered by the executive order issued on April 2. This corresponds for instance to 20% for EU countries and 145% for China. However, pharmaceutical (HS3 2-digit: 30) products are still exempt. In the third scenario (S3 “Pharma”) we assume that the exemption for pharmaceutical ⁷ products is lifted for each country. As we explained above, this seems likely given that an investigation on national security grounds has been launched in early April. All other assumptions are similar to S2.

Table 2

We mapped the tariffs at a detailed product level to the similarly detailed bilateral trade data for 2023 and calculated the trade-weighted average tariffs by product group and country. ⁸ Table 3 shows the average tariff of the EU-27 for the ten product groups that make up the highest share in total exports to the US in 2023. With a share of nearly 20% of all EU-27 exports, pharmaceutical products are on top of the list. Since most of the products included in this group are exempt, additional tariffs are only marginal in scenarios S1 and S2 but increase by 20 percentage points (pp) in S3 as per scenario assumption. Mechanical machineries (HS3 chapter 84) and automobiles and automobile parts (HS3 87) follow with shares of 16% and 12% of total EU exports to the US, respectively. As expected, steel, aluminum, automobiles and automobile parts show the highest average tariff increases in all three scenarios. ⁹ At the bottom of the ten most important export product groups are beverages (HS3 22) and cosmetics (HS3 33). Since none of the products in these groups are exempt from the tariffs, their average tariff rates increase by exactly 10 (S1) and 20 (S2 and S3) pp, respectively.

Table 4 shows the average tariff increases for the ten EU countries for which, in terms of shares of total exports, the US is the largest export market. These country-wide tariff exposures depend on the export structure of the specific country. In scenario S1, the EU exports to the US are subject to tariff increases of around 10 pp. However, this masks substantial country heterogeneity with much smaller increases for Ireland (2.4 pp) and Denmark (6.2 pp) and much larger ones for Slovakia (23.5 pp) and Germany (13.4 pp). The lower tariffs for Ireland and Denmark reflect their large shares of exports of pharmaceutical products to the US. Lifting the exemption for those products in S3 would increase the average tariff on Irish as well as Danish products to the US by around 10 pp. This would particularly hit Ireland, since nearly a third of its goods exports are shipped to the US market. By contrast, Slovakia already faces a high tariff increase of 23.5 pp in our least severe scenario S1. Ending the 90-day relief or imposing additional tariffs on pharmaceuticals leaves the tariff increases nearly unchanged. This reflects the higher share of goods with higher tariff rates of 25% in Slovakian exports, with automobiles and automobile parts playing a particularly important role. We observe a similar pattern for Germany and Austria. These two countries also export noticeable amounts of steel and pharmaceutical products to the US.

Table 3

Table 4

3 Method

We simulated the effects of the tariffs on Austria using a global input-output model which is based on the OECD's global input-output table for 2019. ¹⁰ The model comprises 77 regions (76 countries and the rest of the world) and 45 sectors / product groups according to the International Standard Industrial Classification of All Economic Activities (ISIC) Rev. 4 classification. It maps the interdependencies between the economic sectors of the countries by depicting national and international production networks, international trade in finished products and the value added resulting from production in a matrix structure. For each industry, in addition to the domestic intermediate input linkages, the imports flowing into domestic production are also included in detail by industry and country of origin. The model can be used to simulate the effects of demand shocks on the economic sectors in all countries.

3.1 Deriving the decrease in import demand using tariff elasticities

The global input-output model requires changes in final demand or intermediate inputs as simulation inputs. To translate the change in US tariffs to changes in US import demand we employ tariff elasticities. For example, a tariff elasticity of –0.5 means that a tariff increase of 1 pp leads to a reduction in import demand of 0.5 pp. In the empirical literature, the magnitude of elasticity estimates can differ substantially depending on the methods and data used and which countries and goods are covered. For illustration purposes, we cite two examples from the large literature. Devarajan et al. (2023) estimated elasticities of around –0.6 for developing and –1.4 for high-income countries. A survey of Schwarz et al. (2020) of 3,524 previous estimates found a median elasticity of –3.8 with a range from –2.5 to –5.1.

The elasticities we use are based on two sources. To cover the variation between product groups, we use the trade elasticities from Fontagné et al. (2022), which are available at the Harmonized System (HS) 3 (2007) subheadings (6-digit) product level (5,052 products). ¹¹ They found substantial variation between products, ranging from –3.6 for highly differentiated products like footwear to –19 for standardized products like mineral products. Since the elasticities found by Fontagné et al. (2022) are rather large (Median of –5.6) – potentially reflecting long-term relationships – we use the aggregate elasticities estimated by Boehm et al. (2023) to cover the medium-term perspective (five years). Using a thorough identification strategy, they find a short-term (one year after the tariff increase) elasticity of –0.76. After five years, the elasticity increases to –1.24. Since we are interested in medium-term effects of the tariff increases, we scale the elasticities of Fontagné et al. (2022) to a (US trade-weighted) average of –1.24 (the five-year elasticity of Boehm et al., 2023). This gives us detailed elasticities at product level \(\beta_{k}\):

\(\beta_{k} = (\varepsilon_{k} - 1)/\omega\), (1)

with \(\varepsilon_{k}\) being the HS3 6-digit trade elasticity from the dataset of Fontagné et al. (2022) and \(\omega\) being the scaling factor chosen so that the US import-weighted mean tariff elasticity equals ­–1.24. The distribution of the scaled elasticities is displayed in chart 1. Boehm et al. (2023) also estimated elasticities for a subset of aggregate HS chapters (2 digits). These include, among other items, plastics, wood, paper, textiles, metals and machinery. We may compare those elasticities for the medium term (five years) with the trade-weighted (US imports from EU-27) scaled versions of the Fontagné et al. (2022) elasticities we use in this analysis. Boehm et al. (2023) reported their estimated sector-specific elasticities only in a graph. This complicates a comparison. However, their estimates for the fifth year roughly vary between –0.2 and –4.0. Our scaled estimates for comparable HS chapters vary between –0.5 and –1.3, which is within the range of their results.

Chart 1

To convert the increases in US tariffs into a reduction in US import demand, we first allocate the tariff increases to the 5,052 6-digit HS3 product groups of each exporter (for further details, see annex I: data and tariff mapping). We then calculate the percentage change in each tariff relative to its corresponding baseline tariff rate. Our baseline tariff rates stem from the latest MAcMap-HS6 tariff data (i.e. 2019), which matches the product classification of the trade and elasticity data we use. These data include the WTO most-favoured-nation (MFN) and preferential tariff rates but excludes tariffs imposed via special and temporary legislation such as the tariffs from Trump’s first term. The tariff change multiplied by the tariff elasticities ¹² gives the change in US import demand at the level of HS3 subheadings (6-digit) for each exporter. We then aggregate these impacts to the dimensions of our global input-output model (77 countries x 77 countries x 45 product groups) using mapping matrices (see annex I). Formally this is:

\(\mathrm{\Delta}\ln\left( X_{US,j,k} \right) = \beta_{k}\left\lbrack \ln\left( 1 + {\widehat{\tau}}_{US,j,k} \right) - ln(1 + \tau_{US,j,k}) \right\rbrack\), (2)

with \(X_{US,j,k}\) being the US imports of trading partner \(j\) of HS3 6-digit subheadings \(k\). The difference operator is given by \(\mathrm{\Delta}\) which yields the percentage change of US imports in equation (2). The initial (before the tariff increase) US import tariff rate with trading partner \(j\) and product group \(k\) is given by \(\tau_{US,j,k}\). The new tariff rate – i.e. after the increase implied by the respective scenario– is given by \({\widehat{\tau}}_{US,j,k}\). ¹³

We assume that the exporters affected by the tariffs do not adjust their prices. Several empirical studies on the impact of tariffs during Trump's first term found evidence in favor of that assumption (see Fajgelbaum et al., 2020; Cavallo et al., 2021 or Amiti et al., 2019). As Amiti et al. (2019) note, however, this contrasts with studies on price reactions to exchange rate shocks or empirical estimates that imply rising export supply curves (Gopinath, et al., 2010; Broda et al., 2008). One of the reasons for the lack of transmission cited by Amiti et al. (2019) is that Trump's tariffs may not be perceived as permanent measures by exporters due to the high level of uncertainty (his erratic trade policy). In such a case, it would be difficult for exporters to increase previously lowered prices if the tariffs are lifted. As trade policy in Donald Trump's second term is similarly erratic and characterized by even greater uncertainty (as measured, for example, by the Trade Policy Uncertainty Index), we do not assume a strong reaction in export prices this time either and therefore assume a full transmission to US import prices. However, it should be noted that the higher tariff rate increases of Trump’s second term could very well have impacts on export prices, especially if strong non-linearities are present.

3.2 Simulating the impacts of the US tariff increases with a global input-output model

Imports before and after the tariff increase for each exporter and product group are mapped from the HS3 6-digit product group level to the ISIC Rev. 4 industry classification (see annex I: data and tariff mapping). The change in US imports ¹⁴ at this level is then used in the global input-output model to simulate the direct and indirect impacts on value added.

In our global input-output model, simulating a demand shock involves quantifying how changes in final demand propagate through the world economy’s interconnected industries. The Leontief inverse matrix \(L = (I - \theta)^{- 1}\), where \(\theta\) is the matrix of technical coefficients, plays a central role. \(\theta\) is derived by dividing intermediate inputs per industry and country by the output of industry i in country j . When a demand shock \(\epsilon = \left( \epsilon_{1},\ldots,\epsilon_{s} \right)\) hits the final demand vector d , the resulting change in sectoral outputs is calculated as \(\mathrm{\Delta}x = L\epsilon\). In our global input-output model, the final demand vector has the dimension 3,465 (=77 countries x 45 goods). This framework allows us to trace cascading impacts of reduced US imports across the global production network.

Note that the global input-output model delivers the impacts of the tariff increases from a comparative static perspective. As such, these impacts do not have a specific time span (short, medium, long) or development path attached to them. However, one could argue that it takes some time for the tariff impacts to feed through the entire production network, which we simulate by employing the Leontief inverse matrix. Because of that and since we are scaling the tariff elasticities to impacts after five years (as explained above), we would argue that our results represent medium-term effects. ¹⁵

4 Results

Chart 2 gives an overview of the results of all three scenarios for the most affected industries and countries. The bars for scenario S1 show the change of value added relative to the baseline in %, the stacked bars for scenarios S2 and S3 show the additional impact relative to S1 (S2) in the medium term, i.e. after five years. In scenario S1 (“90 days”: with 25% tariffs on steel, aluminum, automobiles and automobile parts from the EU), real GDP in the EU and the euro area decreases by 0.20% (see the right panel of chart 2). The countries with the strongest declines are Ireland, Denmark, Belgium, Germany and Slovakia.

In scenario S2 (“Liberation”: with additional tariffs of 10% for all other goods from the EU that are not exempt), the additional GDP loss (EU: –0.08 pp) is relatively small compared to scenario S1. The size of the additional impact on the EU’s GDP in scenario S3 (“Pharma”: with additional tariffs of 20% on pharmaceutical exports from the EU) is very similar to the impact in the first scenario (–0.19 pp).

Chart 2

In the following, we will focus on the discussion of the results of the most encompassing scenario S3. For more detailed results of S1 and S2, consult annex II: detailed results in tables. The US tariff increases cover a wide range of industries, with manufacture of pharmaceutical products (C21) being hit particularly hard with a decline of value added of 10.4% in the EU. The US is the single largest export market for pharmaceutical products from the EU. These products accounted for around 20% of all EU exports to the US in 2023. In 17 out of the 27 EU countries, this industry shows the highest relative losses in value added, with Ireland (–21.2%), Denmark (–15.2%) and Belgium (–10.1%) being hit hardest (see table 5). The second most affected industry is manufacture of other transport equipment (C30), which encompasses the production of a wide range of vehicles and equipment used for transportation, excluding motor vehicles, trailers, and semi-trailers (which are covered under NACE C29). Value added in this industry falls by 3.1% in the EU, with Poland (–5.7%), France (–4.7%) and Latvia (–3.5%) being most severely affected. The third highest loss of value added for the EU (–1.7%) we find for basic metals (C24), with Greece, Ireland and Sweden being hit hardest.

Table 5

Looking at the impacts on different countries, there is one country that stands out in terms of GDP losses (see table 6). If all US import tariffs of scenario S3 will materialize, the GDP of Ireland could decline by 2.7%. This is mainly due to the important role of pharmaceutical exports to the US, but also to a high exposure in other relevant industries (repair and installation of machinery and equipment (C31–33), fishing and agriculture (A03) and manufacture of basic metals (C24)). The second most affected country is Denmark, with a GDP loss of 1.0%, driven by substantial declines in manufacture of pharmaceuticals (C21) and losses in manufacture of chemicals (C20), and fishing and agriculture (A03). The GDP of Belgium could decline by 0.7%, mainly due to manufacture of pharmaceuticals (C21), mining support activities (B09) and manufacture of vehicles (C30). Germany ranks fourth among the most affected EU countries.

Table 6

5 Does the use of the global input-output table from 2019 distort our results?

Our calculations are based on the OECD global input-output table from 2019. Since then, the trade relations of some European industries with the US underwent significant changes. Total exports of goods of the EU to the US in nominal terms increased by 38%. The share of the US in extra-EU exports rose from 18% to 21%. For pharmaceutical products, the increase of EU exports into the US amounts to 85%. This implies that our global input-output model underestimates the impact of the tariffs for industries and countries where the role of the US as export market increased. In order to assess the relevance of these changes, we compare for each EU country j the ratio of exports of good i to the US, relative to the output of the industry i that produces this good for the years 2019 and 2024. ¹⁶ An increase of the exports to US-to-turnover ratio indicates that the impact of the US tariffs on this industry is larger than the result of the global input-output model suggests.

Chart 3 shows the changes of this ratio for the six most affected industries for those EU countries where data are available. For the EU and the euro area, the ratio changes are modest for most industries / goods. For manufacture of transport equipment (C30), the role of exports to the US has declined by 12% for the EU. For the five other most affected industries, the ratio increased by about 10–20%.

Chart 3

Source: Author’s calculations.

This increase does, however, not imply that the actual impact on the particular industry is 10–20% above our results. It only means that the direct exposure to the US and hence the potential direct effect is higher by this amount. However, the total impacts also contain the indirect exposures via production linkages with industries of countries outside the EU. For countries with strong downstream linkages in international production networks these indirect effects play a stronger role. ¹⁷ Depending on how the ratios of exports to US to turnover have changed in non-EU countries since 2019, this may either increase or decrease the bias of the total impact. ¹⁸

The industry where the direct role of exports to the US increased most is manufacture of machinery and equipment (C28). For the EU the export-to-turnover ratio increased by 24%. Five countries show an increase of more than 40% (Denmark, Austria, Latvia, Czechia and Greece, see chart 3). For manufacture of pharmaceutical products (C21), the informational content of the chart is limited for at least two reasons. The first is that Ireland – the country which is affected most strongly by the US tariffs for these goods – publishes no data for turnover in this industry. The second reason is that the regional export data for Denmark (which, for instance, exports the successful drug Ozempic to the US) were re-classified over time. Data from the Comext database on international trade in goods show a decline of 95% of exports to the US in 2017 compared to the previous year. In December 2017, the drug Ozempic was approved in the US. At the same time, exports classified under the region code QZ (countries and territories not specified for commercial or military reasons in the framework of trade with third countries) increased tenfold. ¹⁹ These likely contain exports to the US. The high growth rate for Greece’s exports of pharmaceutical products (+820%) is mainly due to low exports to the US in 2019 and, consequently, a low impact of US tariffs on this industry in Greece.

For manufacture of other transport equipment (C30), the role of the US as export destination country fell between 2019 and 2024. This is especially the case for Italy, Belgium, Sweden, the Netherlands and Austria. For the other three of the six most affected industries, the exports to US-to-turnover ratio increased by 10 to 20% for the EU, indicating no significant underestimation of the effects due to the use of the global input-output tables from 2019.

6 Summary and discussion

The main result of our study is that there is strong heterogeneity regarding the impacts of US tariffs on different European industries. To our knowledge, we are the first to investigate the effects in such detail, i.e. at product- / industry-level. The aggregate results of our study for the EU – although not accurately comparable due to different scenario assumptions and underlying models – are in line with recent studies from Bruegel, the Kiel Institute for the World Economy and the European Commission (Barata de Rocha et al. (2025), Hinz et al. (2025) and European Commission (2025)). For Austria, our results are close to current estimates provided by WIFO und the Institute for Advances Studies, which found short-term effects on Austrian GDP (i.e. within one year) of –0.35% (WIFO) and –0.20% (IAS). Both estimates are consistent with our medium-term effects of –0.5%. ²⁰

While our global input-output approach offers valuable insights, several limitations should be considered: It assumes constant input-output relationships, does not capture adjustments in prices or general equilibrium effects ²¹ and omits potential effects from retaliatory tariffs or wider economic feedback mechanisms (i.e. on commodity prices). Potential production relocations and the impacts of heightened policy uncertainty ²² on investment and household consumption are not comprehensively addressed. Finally, the additional turmoil in global financial markets could significantly exacerbate the downturn, increasing the likelihood of a global recession. Given all of these considerations, we take our model results to lie at the lower end of the expected impact range.

Despite these caveats, there are some important policy conclusions that can be drawn from our analysis. For the EU as a whole, the negative impact is not negligible but manageable. However, certain industries (especially, manufacture of pharmaceuticals and transport equipment) might suffer considerably. In the short term, ongoing diplomatic initiatives to address trade disputes and to negotiate potential tariff reductions are essential for curbing further escalation dynamics. Furthermore, providing targeted support to highly exposed industries might be necessary. In the medium term, the EU should strengthen its efforts to diversify export markets, to reduce reliance on the US and to increase resilience against unilateral tariffs. In addition, enhancing integration of the EU internal market and improving the resilience of European supply chains can help cushion the effects of external trade shocks. These policy recommendations can help mitigate the negative ramifications of US tariffs and promote sustainable economic growth in the EU.

7 References

Amiti, M., S. J. Redding and D. E. Weinstein. 2019.  The Impact of the 2018 Tariffs on Prices and Welfare. In: Journal of Economic Perspectives 33 (4): 187–210.

Baker, S. R., N. Bloom and S. J. Davis. 2016.  Measuring Economic Policy Uncertainty. In: Quarterly Journal of Economics 131 (4): 1593–1636.

Barata da Rocha, M., N. Boivin and N. Poitiers. 2025.  The economic impact of Trump’s tariffs on Europe: an initial assessment. Bruegel Analysis. April 17, 2025.

Boehm, C., A. A. Levchenko and N. Pandalai-Nayar. 2023.  The long and short (run) of trade elasticities. In: American Economic Review 113 (4): 861–905.

Broda, C., N. Limão and D. E. Weinstein. 2008.  Optimal Tariffs and Market Power: The Evidence. In: American Economic Review 98 (5): 2032–2065.

Caldara, D., M. Iacoviello, P. Molligo, A. Prestipino and A. Raffo. 2020.  The Economic Effects of Trade Policy Uncertainty. In: Journal of Monetary Economics 109: 38–59.

Cavallo, A., G. Gopinath, B. Neiman and J. Tang. 2021.  Tariff Pass-Through at the Border and at the Store: Evidence from US Trade Policy. In: American Economic Review: Insights 3 (1): 19–34.

Coibion, O., D. Georgarakos, Y. Gorodnichenko, G. Kenny and M. Weber. 2024.  The Effect of Macroeconomic Uncertainty on Household Spending. In: American Economic Review 114 (3): 645–677.

DB Research. 2025.  What now? Assessing the impact of US reciprocal tariffs on Europe. April 3, 2025.

Devarajan, S., D. S. Go and S. Robinson. 2023.  Trade elasticities in aggregate models. Estimates for 191 countries. World Bank Policy Research Working Paper 10490.

European Commission. 2025.  The macroeconomic effect of US tariff hikes. Special topic of the Spring 2025 economic forecast. May 19, 2025. The macroeconomic effect of US tariff hikes - European Commission

Fontagné, L., H. Guimbard and G. Orefice. 2022.  Tariff-based product-level trade elasticities. In: Journal of International Economics 137: 1–25.

Gaulier, G. and S. Zignago. 2010.  BACI: International Trade Database at the Product-Level. The 1994–2007 Version. CEPII Working Paper 2010-23. October 2010. CEPII.

Gopinath, G., O. Itskhoki and R. Rigobon. 2010.  Currency Choice and Exchange Rate Pass-Through. In: American Economic Review 100 (1): 304–336.

Guimbard, H., S. Jean, M. Mimouni and X. Pichot. 2012.  MacMap-HS6 2007, an exhaustive and consistent measure of applied protection in 2007. International Economics Q2 2012: 99–122.

Hinz, J., I. Méjean and M. Schularick. 2025.  The consequences of the Trump trade war for Europe. Kiel Policy Brief No. 190. April 2025.

Liao, S., I. S. Kim, S. Miyano and F. Zhu. 2020.  Concordance: Product Concordance. R package version 2.0.0.  https://CRAN.R-project.org/package=concordance .

OECD. 2023.  OECD Inter-Country Input-Output Tables.  http://oe.cd/icio .

Quintana, J. 2024.  The Dynamics of Trade Fragmentation: a Network Approach. Working Paper Series, Banco de España. Forthcoming.

Schwarz, J., T. Havranek, Z. Irsova and J. Bajzik. 2020.  Armington elasticity and international trade models. Fifty years on. Vox EU Columns. Armington elasticity and international trade models: Fifty years on | CEPR .

8 Annex I: data and tariff mapping

As a data source for bilateral trade data, we use the BACI dataset for the year 2023 at HS3/2007 classification (see Gaulier and Zignago, 2010). This dataset covers 238 countries. As a starting point for the simulation of tariff increases, we take the ad-valorem MFN tariff rates at HS 6-digit level for 2019 contained in the Market Access Map HS6 dataset (see Guimbard et al., 2012) for 152 importing and 189 exporting countries. ²³ The trade elasticities available at HS3 6-digit level are taken from the replication files of Fontagné, Guimbard and Orefice (2022). ²⁴ As there is no tariff data on Taiwan available and the BACI data on trade ²⁵ only contains the approximate category “Asia, not elsewhere specified (iso 490),” we subsumed Taiwan under the category “rest of the world.”

The mapping of the HS3 (2007) subheadings (6-digit) import flows to the ISIC Rev. 4 classification, used in the global ICIO input-output table (see OECD, 2023), was carried out in two steps. First, the 6-digit was mapped to ISIC Rev. 3 2-digit level using the R-package “concordance” (see Liao et al., 2020). The resulting ISIC Rev. 3 2-digits were then mapped by the authors to the ISIC Rev. 4 aggregates of the ICIO IO table (for a list of the ICIO ISIC Rev. 4 industries, see table A-6). The products of the HS3 2-digit code of the pharmaceutical products were considered separately (under a newly created heading 24a, see table A-7), as they would otherwise have been assigned to an ISIC Rev. 3 2-digit code together with chemical products (24).

Tariffs in the US are recorded in the Harmonized Tariff Schedule (HTS) of the United States International Trade Commission (USITC) at detailed 8-digit HS product level. For example, 52042000 would correspond to cotton threads. Tariffs are imposed or exempt at this level. Given this level of detail in the tariff schedules and because usually not all products in an industry are affected by the tariffs of an executive order, a more detailed product group-specific recording of the impact of tariff increases is advised.

For the classification, we used the tariff schedule 2025 HTS Revision 12 (05/13/2025), which was the latest available schedule at the cut-off date (May 12) of the analysis. ²⁶ For each product at HS 8-digit level, it is possible to look up which duty applies to which country. However, we record the tariffs’ impact on the products at the HS 6-digit level, as our supplementary foreign trade, tariff and tariff elasticity data are also available at this level. ²⁷ For example, the HS subheading 840734 stands for the vehicle component “Engines; reciprocating piston engines, of a kind used for the propulsion of vehicles of chapter 87, of a cylinder capacity exceeding 1000cc.”

9 Annex II: detailed results in tables

Table A-1

Table A-2

Table A-2

Table A-3

Table A-4

Table A-5