Fetching primary parquet sources and computing exhibits.
tariff scenarios
What happens if USA sets a 25% tariff on China's HS 854231?
A scenario-driven tariff simulator: the direct partial-equilibrium effect is computed under a CES-Armington import demand with substitution elasticity sigma = 4(Saito 2004 IMF WP 04/36 on HS2 aggregates; Broda & Weinstein 2006 QJEreport a lower median near 3; policy simulations routinely use 3 to 5). Upstream pass-through is decomposed using OECD TiVA foreign value-added intensity (Koopman, Wang & Wei 2014), and downstream consumer impact is allocated across destination activities by TiVA FVA intensity weighted by export share. Amiti, Redding & Weinstein (2019 JEP) document near-complete pass-through of the 2018 US tariffs onto domestic prices, which justifies treating the USA importer and consumer as the primary incidence bearers in the diagrams below.
originCHN
destinationUSA
HS6854231
baseline tariff (MFN 2024)0.0%
scenario tariff25.0%
bilateral flow (2024)$3.65B
sigma4
Direct bilateral impact
With the world-price line horizontal under the small-country assumption, the tariff raises the in-USA price from P0 = 1 to P1 = (1 + t1) = 1.25. Imports slide down the CES demand curve from Q0 to Q1, and the shaded regions partition the welfare change into revenue (a transfer to the treasury), consumer surplus loss (paid by the importer), and the deadweight Harberger triangle (Harberger 1964 AER P&P) which no one collects.
Figure 1
Direct tariff incidence, CHN exporting HS 854231 into USA, 2024
Baseline bilateral flow is $3.65B. The 25% scenario tariff (up from MFN 0.0% filed in 2024) raises the landed price by 25.0% percentage points. Under Armington with sigma = 4, imports contract from $3.65B to $1.49B (-59.0%). Tariff revenue collected: $373.6M. Consumer-surplus loss: $642.8M. Harberger deadweight loss: $456.0M. This is a tariff hike from the current MFN schedule.
Method: small-country Armington PE, Q1 = Q0 * ((1+t1)/(1+t0))^epsilon, DWL = 0.5 * (t1-t0)^2 * |epsilon| * P0 * Q0 / (1+t0) (Harberger 1964, AER P&P 54(2): 58-76). Bilateral flow proxied as origin's world exports of this HS6 times destination's share of world imports (country_year_product_ext). MFN rate from WITS-TRAINS via CEPII pref_tariff_hs6, reporter 840, 2024. sigma = 4 (Saito 2004; Broda-Weinstein 2006 QJE).
Upstream pass-through via TiVA
A tariff on CHN's HS 854231 is not just a bill paid by the final importer: part of that price is cost pass-back to whichever economies supply inputs embedded in CHN's production. Leontief (1936 REStat) gives the accounting identity that gross output decomposes into direct and indirect value added contributions via the (I − A)−1inverse; Koopman, Wang & Wei (2014 AER) operationalise that for gross exports. We approximate the partner decomposition by combining CHN's economy-wide foreign value-added share in gross exports (OECD TiVA EXGR_FVA, activity aggregate _T) = 15.8% (OECD TiVA 2020) with CHN's bilateral sourcing mix from BACI.
Figure 2
Estimated upstream hit from a 25% tariff on CHN HS 854231
Downstream burden across USA activities
On the destination side, the question is which end-use sectors actually bear the tariff bill. We weight USA activities by their foreign-value-added intensity (TiVA EXGR_FVA by activity, 2020) times their share of USA's total exports, and allocate the imported-intermediate tariff bill across activities by that intensity. Activities with high FVA content in exports are the ones whose cost base rises most when an intermediate like HS 854231is taxed: the textbook Grossman & Rossi-Hansberg (2008 AER) 'trading tasks' incidence channel.
Figure 3
Estimated downstream burden by USA activity, 25% tariff scenario
No data available for this chart.
TiVA FVA-by-activity series not resolved for USA.
Method: intensity_j = FVA_pct_j * (EXGR_j / EXGR_total) from TiVA (iso3=USA, measure=EXGR_FVA, unit=PT_EXGR, partner=W, 2020) times TiVA EXGR level in USD. Shares renormalised across labeled ISIC activities so they sum to the imported-intermediate bill DeltaP * bilateral flow. Grossman & Rossi-Hansberg (2008) AER 98(5): 1978-1997. Composite rollup codes (_T, C16T18, etc.) excluded to avoid double-counting.
Multi-sector welfare preview
Caliendo & Parro (2015, Review of Economic Studies 82(1): 1-44) generalise Eaton-Kortum to multiple sectors with input-output linkages and show that the welfare consequences of a tariff depend on the sector substitution elasticity sigmas, the base expenditure share on the tariffed origin (1 − lambdaii), and the I-O loop through domestic industries. Ossa (2014, AER104(12): 4104-4146) uses the same apparatus to quantify trade-war payoffs. Arkolakis, Costinot & Rodríguez-Clare (2012, QJE127(1): 51-80) prove that in a broad class of trade models the gain from trade is dW/W = (1 − lambdaii)1/(1−sigma), so high-sigma, small-domestic-share sectors deliver the largest swings. As a preview of the sectoral heterogeneity, the bars below rank HS6 lines inside chapter 85 by their Broda-Weinstein-style sigma (trade_elasticity.parquet) and colour by their USA MFN. A full Caliendo-Parro solve requires an ICIO cube that is scheduled but not yet ingested in the workbench.
Figure 4
Within-chapter HS85 elasticity and MFN profile, USA 2024
Amiti, Redding & Weinstein (2019, JEP 33(4): 187-210) document that the 2018 US tariff schedule did not just compress imports of Chinese goods, it re-sorted them across third countries: Vietnam, Mexico, Korea and the EU absorbed a substantial share of the diverted flow. The destination's import elasticity with respect to the tariffed origin therefore depends on how concentrated the global supply of this HS6 is. The bars rank the world's top exporters of HS 854231 in 2024 (excluding China); the more weight in the right tail, the cheaper the substitution menu the destination importer faces under the 25% scenario.
Figure 4b
World's top alternative exporters of HS 854231 (excluding China), 2024
A 10-pp uniform hike: revenue, deadweight, and terms-of-trade
A canonical textbook question is how a tariff's welfare cost decomposes into a transfer to the treasury (revenue), a triangle nobody collects (deadweight), and any terms-of-trade improvement the tariff buys by pushing world prices down. Under the small-country assumption pass-through is complete and the ToT channel is zero (Amiti, Redding & Weinstein 2019 JEPfind near-complete pass-through of the 2018 US tariffs to US retail prices). Cavallo, Gopinath, Neiman & Tang (2021, Review of Economics and Statistics) decompose the same tariffs into border pass-through (~1.0) and retail pass-through (0.93 at an 18-month horizon), leaving 7% of the tariff borne by foreign producers as a ToT gain. We apply a +10 percentage-point uniform hike on top of the current schedule and attribute PT = 0.93 to the consumer, 1 - PT = 0.07 to the foreign producer. The rectangles+triangle sum to the consumer's ex-ante marginal outlay; colours match the Harberger (1964) diagram in Figure 1.
Figure 5
Decomposition of a +10 pp uniform tariff on CHN HS 854231 into USA
On the $3.65B baseline flow, a 10 pp hike (from 0.0% to 10.0%) collects $249.2M in revenue, imposes a $307.0M consumer-surplus loss, wastes $73.0M as a Harberger triangle, and buys $17.4M in terms-of-trade improvement (at PT = 0.93). Net to the USA economy: -$40.4M. The ToT gain is small in PE because (1 - PT) is only 0.07 on the Cavallo et al. central estimate, and shrinks to zero under full pass-through as in Amiti-Redding-Weinstein. In a full multi-sector Caliendo-Parro (2015) solve the ToT channel is endogenous and can flip sign once retaliation is modelled (Ossa 2014, AER).
Method: Q1 = Q0*((1+t1)/(1+t0))^(-sigma) with sigma = 4. Revenue = (t1-t0)*P0*Q1. CS loss = P0*(t1-t0)*(Q0+Q1)/2. DWL = 0.5*(t1-t0)^2*sigma*P0*Q0/(1+t0). ToT transfer = (1 - PT)*DeltaP*Q1 with PT = 0.93 (Cavallo, Gopinath, Neiman & Tang 2021 RESTAT, 18-month retail pass-through). Harberger (1964) AER P&P 54(2): 58-76.
Sigma sensitivity: how much does the elasticity choice matter?
Every headline number above is built on sigma = 4. The policy literature uses a range: Broda & Weinstein (2006, QJE) report a median HS10 sigma near 3; Saito (2004, IMF WP 04/36) around 4 on HS2 aggregates; CGE practitioners commonly use sigma in [3, 5]; higher values (6-8) emerge from highly disaggregated estimates on differentiated goods. At this scenario's bilateral flow of $3.65B and a tariff change from 0.0% to 25.0%, we recompute the import-volume contraction, the Harberger deadweight triangle, and tariff revenue under sigma ∈ {2, 4, 6, 8} to show the elasticity-choice footprint. DWL scales roughly linearly in sigma; the volume contraction scales more steeply because sigma enters as an exponent. Revenue is non-monotone in sigma: higher sigma shrinks Q1 faster, which shrinks the revenue rectangle even though t1 is held fixed.
Figure 6
Elasticity sensitivity: DWL, revenue, and import-volume response under sigma in {2, 4, 6, 8}, 25% scenario on CHN HS 854231 into USA
Retaliation: symmetric tariff response and the DWL doubling result
Ossa (2014, AER104(12): 4104-4146) and Caliendo & Parro (2015) show that trade wars are not one-sided: when the destination imposes a tariff, the origin optimally retaliates, and the welfare loss doubles because the Harberger triangle is now incurred on both sides of the border. Fajgelbaum, Goldberg, Kennedy & Khandelwal (2020, QJE 135(1): 1-55) document that US agricultural exports were targeted almost dollar-for-dollar in the 2018-19 retaliation by China, EU, and Mexico. Under a small-country PE assumption on both legs, a symmetric counter-tariff of25% imposed by China on USA's reciprocal bilateral flow generates a mirror Harberger triangle and revenue rectangle. The global welfare loss is the sum of both DWLs; neither side captures the other's revenue. We bring in the reciprocal bilateral flow (USA exports to China of the same HS6) and compute the retaliation decomposition.
Figure 7
Symmetric retaliation decomposition: USA tariffs China at 25%, China retaliates at 25%
Forward-leg DWL on the $3.65B bilateral flow is $456.0M. Symmetric retaliation at 25% on $2.92B of China's reciprocal imports of HS 854231 from USA (at 80% coverage per Bown 2019 PIIE) generates a mirror DWL of $364.8M. Total global deadweight: $820.8M, roughly 1.8× the one-sided figure. Each treasury keeps its own revenue ($672.4M combined), but the combined consumer-surplus loss is $1.16B. This is the canonical 'trade war is net-negative even at Nash equilibrium' result (Johnson 1953 ; Kennan & Riezman 1988 ; Ossa 2014). In a full CGE solve with multi-sector I-O linkages the retaliation DWL can exceed the forward DWL because third-country substitution is imperfect (Fajgelbaum et al. 2020).
How to use this page
Edit the URL to change scenario: ?origin=CHN&destination=USA&hs=854231&rate=25. ISO3 codes are uppercased server-side; HS6 must be six digits.
For live slider interaction on the same PE math see /tariff-lab. That page is a client component; this one is server-rendered so the URL alone reproduces any scenario shown in a report or screenshot.
The upstream decomposition here uses aggregate TiVA FVA; a full ICIO-inverse decomposition (Koopman-Wang-Wei 2014) requires a partner-resolved I-O cube that is scheduled but not yet ingested.
References
Amiti, M., Redding, S. J., & Weinstein, D. E. (2019). 'The Impact of the 2018 Trade War on US Prices and Welfare.' Journal of Economic Perspectives 33(4): 187-210.
Arkolakis, C., Costinot, A. & Rodríguez-Clare, A. (2012). 'New Trade Models, Same Old Gains?' Quarterly Journal of Economics 127(1): 51-80.
Armington, P. S. (1969). 'A Theory of Demand for Products Distinguished by Place of Production.' IMF Staff Papers 16(1): 159-178.
Broda, C., & Weinstein, D. E. (2006). 'Globalization and the Gains from Variety.' Quarterly Journal of Economics 121(2): 541-585.
Caliendo, L., & Parro, F. (2015). 'Estimates of the Trade and Welfare Effects of NAFTA.' Review of Economic Studies 82(1): 1-44.
Grossman, G. M., & Rossi-Hansberg, E. (2008). 'Trading Tasks: A Simple Theory of Offshoring.' American Economic Review 98(5): 1978-1997.
Harberger, A. C. (1964). 'The Measurement of Waste.' American Economic Review, Papers & Proceedings 54(2): 58-76.
Koopman, R., Wang, Z., & Wei, S.-J. (2014). 'Tracing Value-Added and Double Counting in Gross Exports.' American Economic Review 104(2): 459-494.
Leontief, W. W. (1936). 'Quantitative Input and Output Relations in the Economic System of the United States.' Review of Economics and Statistics 18(3): 105-125.
Ossa, R. (2014). 'Trade Wars and Trade Talks with Data.' American Economic Review 104(12): 4104-4146.
Saito, M. (2004). 'Armington Elasticities in Intermediate Inputs Trade: A Problem in Using Multilateral Trade Data.' IMF Working Paper 04/36.
Top exporter to CHN in 2024: Korea, Rep. (KOR) at 7.2% of CHN's imports. First-order upstream hit on that partner under full pass-back of DeltaP = 25.0% on bilateral flow $3.65B: $10.3M. The ten partners below together would bear roughly $79.1M, compared to the direct importer-side pass-through of $912.0M (~DeltaP * Q0).
Method: hit_k = FVA_origin (TiVA EXGR_FVA, activity=_T, partner=W, 2020) * s_k (BACI bilateral_year share of CHN's total imports, 2024) * DeltaP * bilateral flow. Leontief (1936) REStat 18(3): 105-125; Koopman, Wang & Wei (2014) AER 104(2): 459-494. This is a first-order approximation; a full input-output decomposition requires a non-partner-aggregated ICIO cube, which the current workbench does not yet include.
The top three alternative suppliers (S19, KOR, MYS) together account for 47.4% of world exports of HS 854231 in 2024, against China's share of 13.5%. When China's share is large and the next three suppliers are small, the substitution menu is thin and the import-volume contraction in Figure 1 under-states the welfare cost. When the next three suppliers together rival China, third-country diversion is cheap and the Harberger triangle is approximately the correct welfare measure (Amiti, Redding & Weinstein 2019).
Source: CEPII BACI 2024 via country_year_product_ext, top 12 world exporters of HS 854231, excluding CHN. Shares computed against world imports of the same HS6 (worldHs6Total = $427.66B). Reference: Amiti, Redding & Weinstein (2019, JEP 33(4): 187-210) on third-country substitution in the 2018 US tariffs.
At sigma = 2 the DWL is $228.0M and the import-volume response is -36.0%; at sigma = 8 these become $912.0M and -83.2%. Revenue moves from $583.7M (sigma = 2) to $153.0M (sigma = 8). The central-estimate results at sigma = 4 reported in Figures 1-5 sit near the middle of this range. A policy reader should never cite a single DWL/revenue number without the sigma sensitivity band: Broda-Weinstein (2006) show sigma uncertainty is easily a factor of two on sector-level estimates.
Method: holds t0 = 0.0% and t1 = 25.0% fixed at this scenario's values; varies sigma in {2, 4, 6, 8}. Q1 = Q0*((1+t1)/(1+t0))^(-sigma); DWL = 0.5*(t1-t0)^2*sigma*Q0/(1+t0); revenue = t1*Q1. All under small-country Armington (Armington 1969 IMF Staff Papers 16(1): 159-178; Harberger 1964 AER P&P 54(2): 58-76). Sigma range: Broda & Weinstein (2006 QJE 121(2): 541-585) HS10 median ~3; Saito (2004 IMF WP 04/36) HS2 ~4; CGE practice 3-5; upper bound 6-8 from disaggregated differentiated-goods estimates.
RES
IER
Method: forward leg as Figure 1 (small-country Armington, sigma = 4). Retaliation leg assumes China imposes symmetric 25% tariff on USA's reciprocal bilateral flow of the same HS6, proxied as 80% of the forward flow (Bown 2019, PIIE Policy Brief 19-8, documents ~80% retaliatory coverage in 2018-19 US trade war). Both legs computed with pre-retal tariff = 0 for clean mirror illustration. Total DWL = sum of both Harberger triangles (Johnson 1953 RES; Kennan & Riezman 1988 IER 29; Ossa 2014 AER 104; Fajgelbaum, Goldberg, Kennedy & Khandelwal 2020 QJE 135).