In a rapidly evolving tariff environment, policymakers need credible, updatable estimates of the potential consequences of alternative U.S. tariff scenarios. We develop a transparent, readily accessible tool that reflects the global production structure and explicitly incorporates cross-country supply-chain linkages. Using this tool, we estimate the effects of announced U.S. measures — and reciprocal actions by trading partners such as China — on output (GDP) and prices (inflation) across major economies.

Our baseline U.S. price effect is about one percentage point higher than prominent alternatives — approximately 3 pp compared with ≈ 2 pp in Yale Budget Lab (2025b) — while the estimated effects on Russian prices and output are negligible. Beyond quantification, we disclose the tool itself. Methodologically, we also show — to our knowledge, for the first time — that estimated impacts depend systematically on the level of data aggregation. Moving from aggregate to more granular input–output data increases the estimated U.S. price effects and reduces the estimated output contraction for China.

Section 2 reviews the related literature and provides a concise historical over­view of U.S. tariff policy. Section 3 outlines the methodological framework, including the mathematical formulation and key assumptions. Section 4 describes the data — both aggregate and highly disaggregated global input–output tables (IOTs) — and points to our publicly available prototype for recomputing scenarios. Section 5 develops the theoretical intuition for tariff transmission in IOTs and formalizes how aggregation choice (country-only versus country — industry) shapes results. Section 6 presents the scenario findings. Section 7 concludes and discusses limitations.

The United States has alternated between periods of protectionism and liberali­zation. Several milestones illustrate the scale and impact of policy swings (Fig. 1):

Fig. 1.

Average U.S. import duty and key events, 1821–2025 (%). Source: Yale Budget Lab (2025b).

1. Tariff of Abominations (1828).1 A highly protectionist package that raised average tariff rates to nearly 49%, shielding Northern and Western agricultural producers from foreign competition.

2. Morrill Tariff (1861).2 Enacted at the onset of the Civil War, initially increasing duties to around 26%, with subsequent hikes to 48% by 1865.

3. Smooth–Hawley Tariff (1930).³ Raised duties on more than 20,000 imported­ goods. Contemporary estimates suggested a 40% contraction in U.S. foreign trade; later empirical work placed the direct import effect at 4–8%, with secondary­ spillovers of 8–10% (Irwin, 1996).

4. Reciprocal Trade Agreements Act (1934). Authorized mutual tariff reductions, marking the beginning of sustained trade liberalization.

5. General Agreement on Tariffs and Trade (1947).4 Established a multi­lateral framework for lowering trade barriers and supporting postwar recovery.

6. Recent reciprocal tariffs.

• February — March 2025. The U.S. imposed a 10% tariff on all Chinese goods, followed by an additional 10% surcharge. China retaliated with duties of 10–15% on U.S. agricultural imports. The U.S. also applied 25% tariffs on imports from Canada and Mexico, raising the import-weighted average tariff to 27–30%. ⁵
• April 2025. On April 2, the U.S. announced tariffs covering 183 countries, scheduled to take effect on April 9. Hours before implementation, a 90-day suspension reduced tariffs to 10% for more than 75 countries. By contrast, tariffs on Chinese goods were raised gradually to 145%, prompting China to impose duties of 125% on U.S. imports. Certain products (e.g., electronics with semiconductor content) were temporarily exempted.
• August 2025. The U.S. revised schedules for about 70 countries and announced new duties on pharmaceuticals, copper, and semiconductors. 6 By this point, effective U.S.–China customs duty rates stood at approximately 55% and 33%, respectively.

For additional discussion of the 2025 measures and potential spillovers to the European Union, see Klinova and Kondratyev (2025).

A substantial literature quantifies the effects of tariffs across three sets of economies: (i) countries imposing duties, (ii) countries directly targeted, and (iii) third countries experiencing indirect, or secondary, effects. Most studies conclude that any short-term protection for selected industries is offset by higher prices, resource misallocation, and employment losses (York, 2018a, 2018b; CBO, 1986; Hufbauer & Elliott, 1994).

Researchers apply a range of methods. Partial- and general-equilibrium models remain common (e.g., Irwin, 1996). Difference-in-differences designs treat tariff hikes as quasi-natural experiments (e.g., Cigna et al., 2021; Amiti et al., 2020). Input — output tables (IOTs) are also widely used to simulate outcomes under alternative scenarios (Leontief, 1973; Sayapova and Shirov, 2019). Analysts at the Yale Budget Lab (2025a, 2025b, 2025c) recently applied IOTs to evaluate the effects of heightened U.S. protectionism on the domestic economy.

It is equally important to assess broader, indirect consequences. Trade restrictions alter global demand, supply-chain linkages, and relative prices, affecting countries not directly targeted. Given the diversity of channels, IOT-based estimates are best interpreted as lower bounds. By analogy with pandemic-era applications — where IOT models were used to quantify disruptions to production and consumption (Ponomarenko et al., 2020) — our tariff estimates are conservative approximations of potential impacts.

Prior to mathematical modeling, it is important to understand the boundaries of the quantitative tools’ application. Hence, we consider a couple of extreme cases to form expectations about the likely direction of impact when tariffs are imposed.

First, think of two economies, A and B, that rely only on one another. Let one of them (country A) impose a tariff on its neighbor B. In this simplistic setup, all earnings for the countries originate from international trade. A tariff spike makes it impossible for country B to pay the customs duty, as in the previous trade iteration, country A paid as before, with no margin to offset the tariff. Hence, contraction of home production in both A and B follows: for B, because it cannot sell as many goods as in the previous step; for A, because there are insufficient imports from B to finalize production and consumption in country A.

One possible response by country B is a money injection to nominally pay for the newly imposed duties. Mises (1953) underlines that money exchanges do not take place simultaneously, at a zero time tick. Even though the exchange rate might be floating — where, in the long run, money injection results in currency depreciation — in the immediate time, country A may not notice money injection by country B. This will give more money at the disposal of country A’s citizens. Due to our assumption of perfect connectivity of the two countries, country A will spend all its nominal surplus on the same amount of goods from country B. Hence, price rises occur in country B, disregarding the fact that country B did not impose duties against country A.

However, if both countries impose mutual (same) tariffs against each other, then seemingly no effects take place, as there is merely a larger money reallocation between the two countries. Yet, the first-mover effect matters. As Hayek (1931) stressed the importance of who receives money first, it is important which country is to pay tariffs first. After paying tariffs first, more money arriving may trigger relative price revisions and changes in production chains. Thus, when the same money is to be paid for the counter-tariff, production may already deviate from the initial condition. We might wish that tariffs of the same size in monetary terms (with respect to local money supply) have no effect, but in reality, there will be an effect — though hard to quantify — particularly when economic systems are as complicated as modern ones.

Now consider a scenario where two countries, C and D, are not as interdependent as A and B. In this case, regular incomes exceed expenses on imports. If C and D impose mutual tariffs, the tariff spike can be paid from the residual portion of incomes (to the detriment of some home expenditures), with seemingly no impact on international trade. As in the previous case, the actual effect will still be non-zero, as the reduction of expenditures in certain home areas will spark perturbations in production chains and structures.

The takeaway from these theoretical cases is twofold. First, tariffs may spark price rises in countries that do not impose tariffs of their own if the latter countries resort to money injection, as the new money will not be absorbed by the tariff-setter forever; it will flow back. Second, non-perfectly dependent countries imposing mutually equivalent tariffs may be wrongly expected to have no impact on local production, as just a larger sum of money circulates due to tariffs. The effect in both situations occurs because there is always a time lapse for money to flow through the economy. Most quantitative tools — general equilibrium models and input–output­ tables — assume an instantaneous exchange of goods for money, neglecting such effects related to money proliferation. Though we recognize the limitations of the quantitative tools employed, these remain the second-best options available.

Prior to running quantifications and delivering interpretations for the effect sizes, we study to what degree our findings might be sensitive to the choice of granularity level. To do so, we consider 4 scenarios for which we run an IOT cost computation under a new tariff setup according to the methodology presented above, at two layers of granularity. First, we consider country-level tables (just two countries against each other). Second, we look at their artificial industry structure (we assume each country C has two industries, O1 and O2). Scenario representation and full cost assessment are shown in Fig. 2.

Fig. 2.

Schematic of input–output structures and tariff impact assessment cases. Source: Compiled by the authors.

We assume each country’s own cost is 40 units, and an additional 20 units are spent on goods from the other country. Final consumption equals 50 units for each country. We vary scenarios by the industry structures within countries and types of import — export relationships. In cases 1 and 2, we assume even cost structures within countries. We distinguish cases 1 and 2 by the source of imports (from which industry of the other country). Case 1 might be deemed a symmetric trade setup, as industry 2 from country 2 imports from country 1 and at the same time exports its products to country 1. By contrast, case 2 might be called an asymmetric trade, as industry 2 of country 2 (C2 O2) imports from country 1, while its first industry (C2 O1) exports to country 1.

We look deeper into case 1 by segregating it into cases 3 and 4. In cases 3 and 4, we examine concentrated industries (too related to each other in case 4 or not related at all in case 3). For all 4 granular cases, we have the same country-aggregated input–output tables (see the top-right part of Fig. 2). We consider equivalent tariffs set by each country (see the top-left part of Fig. 2).

In most cases, we find that industry-level (more granular) IOTs deliver a higher impact of tariffs on full production costs. Hence, the more granular the data used, the higher the impact estimate should be expected. However, case 2 is an exception. Case 2 (even internal production structure and asymmetric trade structure) yields lower tariff impact estimates.

The key takeaway here is that we need to benchmark tariff impacts against ­results employing comparable levels of granularity. On the one hand, the most granular­ level might be considered the most accurate (true) one. On the other hand, even for the rest of the world (ROW), the estimates refer to the aggregate level. Hence, the actual impact on the rest of the world might be more elevated than we present.

Nevertheless, we recognize that indirect implications of various U.S. trade policy scenarios might be more considerable than the estimates we obtained. These implications include the following.

1. Considerable growth of uncertainty. This may decrease investment activity and shift investments toward more reliable assets (e.g., gold).

2. Changes in the structure of domestic consumption. As popular imports become more expensive, consumers and producers might begin to prefer more affordable goods. This, in turn, can lead to a more fundamental transformation of the production structure in IOT terms.

3. Reorganization of logistics routes with no changes in the consumption and production structure. Similar to the first episode of the U.S. — China trade confrontation in 2018, logistics routes may change to transport goods from countries with higher duties through those with moderate tariffs. This will increase delivery times and costs, ultimately translating into higher domestic prices.

4. Use of a relatively old database of IOTs (2014). The production structure could have changed over the past 10 years. However, we are limited by available data, which are published with a significant lag. For instance, the latest available­ OECD database9 (for 2020) was published in 2023. Available estimates by the Yale Budget Lab, based on more recent IOTs (GTAP v7; Corong et al., 2017), are similar in scale to ours, which are based on WIOD (2016). As an alternative, long-term IOTs for 30 years from Woltjer et al. (2021) could be used. However, although these were published just prior to 2022, they incorporate data only up to 2000. Consequently, there are no grounds to assume that the production chain structures of 1965–2000 have remained unchanged 25 years later, especially given the significant transformations in the global production structure since the pandemic.

5. In addition, it is important to note that the data we used, which are the only relevant and currently available data, do not reflect the strengthened economic ties between Russia and China. As stated by Krylov and Pakhmutov (2025), these ties have led to increased business cycle synchronization between the two economies. Therefore, our estimates of the effects on Russia may require upward expert adjustments. Naturally, these estimates will be more reliable once updated international IOTs incorporating data for the period after 2022 are released.