Loading workbench page
Fetching primary parquet sources and computing exhibits.
Loading chart...
Fetching primary parquet sources and computing exhibits.
A live register of the major post-2015 industrial-policy packages (IRA and CHIPS in the US, the Net-Zero Industry Act in the EU, Made in China 2025, India's PLI, Japan's GX, and Korea's K-CHIPS) with their stated legal vehicle, enactment date, committed public outlay where officially scored, sectoral scope, and the HS2 chapters they most directly target. Each entry is a cited fact, not an estimate: the source document is linked. Pair this with the BACI sector-exposure bars below to see the magnitude of global trade potentially in policy scope, and with the allied vs non-allied event study to see the post-2022 reshuffling already visible in customs data.
Industrial policy is back on the agenda of every major economy after a 40-year eclipse (Juhasz, Lane & Rodrik 2024, 'The New Economics of Industrial Policy,' Annual Review of Economics 16). The table below catalogues the biggest post-2015 packages, keyed to their statutory instrument. Where a headline public-outlay figure exists and has been scored by an official body (CBO, statutory text, or national gazette), it is reported verbatim; where no such figure exists, the entry says so. We do not invent numbers.
For each program, we translate its stated sectoral scope into HS2 chapters (the mapping is editorial and documented in the program card) and report world trade in those chapters in 2024. This is the exposure frame: the maximum slice of global commerce that could, in principle, be touched by the program's incentives, tariffs, or local-content rules. Actual intensity of treatment is much narrower and varies by HS6 line.
The IRA (P.L. 117-169, 16 August 2022) and CHIPS Act (P.L. 117-167, 9 August 2022) were signed within a week of each other. Both route incentives preferentially toward allied-origin production through local-content rules (IRA Section 30D vehicle credits, Section 45X advanced manufacturing credit) and foreign-entity-of-concern exclusions (CHIPS guardrails). The chart below shows the path of allied-origin vs non-allied-origin exports of goods in the combined HS2 chapters targeted by both acts (HS85, HS87, HS27), indexed to 2021 = 100. 'Allied' is operationalised as OECD members plus Taiwan following the friendshoring-discourse taxonomy used in Aiyar et al. (IMF WP/23/100) and Alfaro & Chor (NBER 31560, 2023). The BACI workbench does not include HS6 bilateral flows, so we cannot show US-specific import growth by origin here; the bloc-level export capacity shift is the closest proxy. Global Trade Alert would resolve this by adding intervention-partner resolution (see Figure 4).
Everything above is inevitably partial. Industrial policy operates through hundreds of instruments (grants, tax credits, loan guarantees, local-content rules, export controls, standards, procurement preferences) across national, sub-national, and SOE channels. No single agency compiles them. The Global Trade Alert project (Evenett & Fritz, 2023) is the canonical third-party database of such interventions, with over 60,000 state acts coded at the product and partner level since 2008. GTA is free for research via its public API. Ingesting it into this workbench would let us replace the seven hand-coded cards above with a machine-readable ledger, connect each intervention to BACI HS6 lines at the partner-resolved level, and run proper difference-in-differences event studies of the sort that Figure 3 can only approximate.
Until then, the three most-cited aggregate reads on post-2022 industrial-policy intensity are: Bown (2023, 'Industrial Policy for Electric Vehicle Supply Chains,' PIIE WP 23-1, and Bown & Clausing 2023 on subsidy races); Juhasz, Lane & Rodrik (2024) on the analytical revival of industrial policy in the economics literature; and DiPippo, Mazzocco & Kennedy (2022) on cross-country comparison of industrial-policy spending (China, US, Germany, Japan, Korea, Taiwan, France, Brazil).
Antràs & Chor (2013, Econometrica81:6) and the Antràs-Chor-Fally- Hillberry (2012, AER P&P) upstreamness measure show that shocks to upstream stages propagate with super-linear multipliers (Baqaee & Farhi 2019, Econometrica87:4). BEC-stage composition is the coarser cousin. IRA and CHIPS target stages up the chain (cells, wafers, modules, battery-grade materials), where Criscuolo et al. (2022, OECD) argue subsidy effectiveness is highest but leakage to foreign suppliers also greatest. Juhasz, Lane & Rodrik (2024, Annu Rev Econ 16) formalise the trade-off: concentrated upstream dependencies raise the option value of public support, but the welfare multiplier depends on the elasticity of substitution in final demand. The plot below is the descriptive base for that calculation.
For each HS2 chapter the two acts target (HS85, HS87, HS27), we compute HHI, CR4, and CR8 on exporter shares of world exports in 2024. HHI = 10,000·Σs2 (Hirschman 1945; Herfindahl 1950; US DOJ/FTC 2010 Horizontal Merger Guidelines); CR4 and CR8 are the cumulative shares of the top 4 and top 8 exporters (Bain 1951, QJE 65:3; Saving 1970, IER 11:1; Miller-Pauly-Sobel 2000, International Economic Review41:3). A chapter with HHI > 2,500 and CR4> 75% indicate concentration among exporting countries. Those measures alone cannot establish whether a subsidy can displace incumbents or how quickly entry would occur.
The 1,500/2,500 bands here follow the historical 2010 DOJ/FTC guidelines, withdrawn in 2023. The 2023 guidelines use a highly concentrated threshold above 1,800. These figures measure concentration across exporting countries, not firms in a defined market; the bands are descriptive benchmarks, not antitrust findings.
The Inflation Reduction Act's Section 45X credit pays producers per kilogram or per watt for lithium-ion battery cells/modules, solar cells and modules, wind turbine components, and critical-mineral processing; Section 30D conditions the EV credit on critical-mineral and battery-component sourcing. CRS report R47262 (Inflation Reduction Act: Clean Vehicle Credit, 2023) and Bown (2023, PIIE WP 23-1) map these credits onto the narrower HS6 lines below. We compare world-export shares for each HS6 in 2022 (the year of enactment, pre-rulemaking effect) and 2024(two credit years later), for the six largest exporters per line. Changes ≥ 1 percentage point are large in these fast-moving markets; changes ≥ 3 pp over only two years are unusually sharp. This is not a causal estimate: Chinese overcapacity, post-Covid demand rebounds, and EU battery-passport rules all overlap the IRA window, but it is the first-order descriptive check the Bown (2023, PIIE) literature calls for.
The Figure 7 panels read within-HS6 reshuffles between exporters; this figure asks the complementary capacity question. For each of the four headline subsidy-tracked sectors, solar PV, lithium-ion batteries, electric vehicles, semiconductors, we compute the cumulative growth rate of world export value from 2022 (IRA and CHIPS enactment year) to 2024. A sector where world trade grows much faster than the global-trade baseline (roughly 4-5 per cent a year in nominal USD per WTO 2024 World Trade Outlook) is a sector where subsidy packages helped expand total capacity; a sector where world trade is flat or shrinking is one where subsidy-driven domestic capacity is largely substituting for imports, consistent with the Bown (2023, PIIE WP 23-1) and Juhasz-Lane-Rodrik (2024) distinction between capacity creation and rent reallocation.
Until Global Trade Alert is ingested, the closest comparable cross-country measure of industrial-policy spendingis the 2019 reference-year cross-country reconciliation by DiPippo, Mazzocco & Kennedy (2022), Red Ink: Estimating Chinese Industrial Policy Spending in Comparative Perspective (CSIS, May 2022). Their study aggregates direct subsidies, state-investment-fund outlays, below-market credit, tax expenditures, and procurement preferences to a single 'direct industrial-policy spending' number as a share of GDP, for nine major economies in 2019, the last pre-pandemic base year with consistent accounts across all nine. The 1 per cent-of-GDP threshold is a natural cut because it separates China (the only economy in the sample above it) from the rest of the OECD-plus-Taiwan sample. The post-2022 wave (IRA, CHIPS, NZIA, GX, K-CHIPS, PLI) will lift the US, EU, Japan and Korea above their 2019 baselines; DiPippo-Mazzocco-Kennedy is the pre-wave benchmark against which any GTA-based re-estimation must land.
Figures 7 and 8 read the post-2022 reshuffle. The longer arc behind the subsidy race is a decade of Chinese capacity build in the same HS6 lines. This figure plots China's share of world exports for each subsidy-tracked HS6 from 2022 to 2024. Bown (2023, PIIE WP 23-1) and Juhasz, Lane & Rodrik (2024, Annu Rev Econ 16) frame the policy question explicitly as incumbency displacement: subsidies must offset accumulated learning-by-doing and scale economies in the dominant supplier, not merely shift a static cross-section. The IRA Section 45X and Section 30D credits, the CHIPS guardrails, the EU NZIA capacity targets, and Korea's K-CHIPS amendments all sit against the same incumbency baseline this figure draws.
The six figures above say this: the HS chapters where IRA / CHIPS / NZIA / CRMA / PLI / GX / K-CHIPS meet each other are concentrated (Figure 6), upstream (Figure 5), and have already begun to reshuffle along the allied vs non-allied seam (Figure 3). The policy question is whether the subsidies move the location of value, not just the location of the last step of assembly. Juhasz, Lane & Rodrik (2024) list the conditions: a learning externality inside the targeted industry, a binding concentration friction that the market cannot unwind on its own, and a credible sunset. China's dual-circulation response (14th Five-Year Plan, 2021) is the mirror: lean further into domestic consumption of refined outputs while using 2023 export controls (gallium, germanium, graphite) to keep incumbency leverage. The empirical base in Figures 5 - 6 says every program is fighting for the same narrow intermediate-stage HS chapters, and the concentration ratios are high enough that unco-ordinated subsidy races will mainly shift rents across jurisdictions rather than raise world output of the targeted goods.
SELECT p.chapter, SUM(cyp.export_value) * 1000 AS world_usd
FROM 'country_year_product/year=2024/*.parquet' cyp
JOIN products_all p ON p.code = cyp.product_code
WHERE p.chapter IN ('85','87','27','84','88','30','29') AND p.revision = 'HS96' AND cyp.export_value > 0
GROUP BY p.chapter;WITH prods AS (
SELECT code FROM products_all
WHERE chapter IN ('85','87','27') AND revision='HS96'
), ctry AS (
SELECT iso3, MIN(code) AS code FROM 'countries.parquet' GROUP BY iso3
)
SELECT year,
CASE WHEN c.iso3 IN (<OECD+TWN list>) THEN 'allied' ELSE 'nonallied' END AS bloc,
SUM(export_value) * 1000 AS v
FROM 'country_year_product/**/*.parquet' cyp
JOIN ctry c ON c.code = cyp.country_code
JOIN prods p ON p.code = cyp.product_code
WHERE year BETWEEN 2015 AND 2024 AND export_value > 0
GROUP BY year, bloc;Scroll horizontally to view the full chart.
WITH g AS (
SELECT year, c.iso3,
CASE product_code WHEN '850760' THEN 'batteries'
WHEN '854143' THEN 'solar modules'
WHEN '854142' THEN 'solar cells'
WHEN '870380' THEN 'EVs'
ELSE 'wind' END AS grp,
export_value AS v
FROM 'country_year_product_ext/**/*.parquet' cyp
JOIN countries c ON c.code = cyp.country_code
WHERE revision='HS22'
AND product_code IN ('850760','854143','854142','850231','850232','850233','850239','870380')
AND year IN (2022, 2024) AND export_value > 0
AND regexp_matches(c.iso3, '^[A-Z0-9]{3}$')
),
tot AS (SELECT grp, year, SUM(v) AS t FROM g GROUP BY grp, year)
SELECT g.grp, g.year, g.iso3, SUM(v)/t AS share
FROM g JOIN tot USING (grp, year) GROUP BY g.grp, g.year, g.iso3, t
ORDER BY grp, year, share DESC;SELECT product_code, year, SUM(export_value) * 1000 AS v
FROM 'country_year_product_ext/**/*.parquet'
WHERE revision = 'HS22'
AND product_code IN ('854142','854143','850760','870380','854231','854232')
AND year IN (2022, 2024) AND export_value > 0
GROUP BY product_code, year;WITH facts AS (
SELECT product_code AS code, year, c.iso3, export_value AS v
FROM 'country_year_product_ext/**/*.parquet' cyp
JOIN (SELECT iso3, MIN(code) code FROM 'countries.parquet'
WHERE regexp_matches(iso3,'^[A-Z0-9]{3}$') GROUP BY iso3) c
ON c.code = cyp.country_code
WHERE revision='HS22'
AND product_code IN ('854142','854143','850760','870380','854231','854232')
AND year BETWEEN 2015 AND 2024 AND export_value > 0
)
SELECT code, year,
SUM(CASE WHEN iso3='CHN' THEN v ELSE 0 END) / NULLIF(SUM(v), 0) AS chn_share
FROM facts GROUP BY code, year ORDER BY code, year;