Loading workbench page
Fetching primary parquet sources and computing exhibits.
Loading chart...
Fetching primary parquet sources and computing exhibits.
US aggregate trade statistics conceal wide regional heterogeneity. Texas alone supplies roughly one in every five export dollars the country ships abroad; Louisiana and Alaska run undiversified petroleum-and-fish baskets; Washington and Kentucky live on Boeing and GE aircraft. This page unpacks the fifty-state export landscape at the origin-of-movement level using the Census Bureau's foreign-trade state series, 2015-2024.
Kemeny & Storper (2020, Journal of Economic Geography) document the rise of 'superstar regions' in advanced economies: export and high-wage employment increasingly concentrate in a small number of metro areas and states, with diminishing catch-up from the rest. The first test at the state level is simple distributional arithmetic, what share of US exports do the top four states command, and how skewed is the distribution? Moretti (2012, The New Geography of Jobs) made the same argument for labour markets: the divergence of American regions is not a rounding error on the national average, it is the story.
The top-ten states differ not just in level but in trajectory. Texas and Louisiana tie closely to petroleum and LNG prices; Washington's line tracks the Boeing 737 MAX grounding (2019) and post-pandemic recovery; Michigan moves with auto cycles and the UAW; New York spikes and falls with refined-gold and diamond re-exports routed through JFK Customs District. USITC (2024) DataWeb historical series match the Census origin-of-movement totals to the dollar at state level.
Autor, Dorn & Hanson (2013, AER) showed that US regions differed sharply in their exposure to the China import shock, with the Rust Belt absorbing a disproportionate share of the employment pain. A decade on, the same regional heterogeneity shows up on the export side: compound annual growth rates over 2015-2024 span a range of many percentage points, with petroleum, LNG, and oilseed-exporting states at the top and several traditional manufacturing states clustered near or below zero. The figure below shows the fifteen highest-CAGR and fifteen lowest-CAGR states over the period. Caveat: these are current-dollar CAGRs uncorrected for commodity price levels, so oil & gas booms mechanically flatter the top group; they still reflect real capacity additions in LNG export terminals and shale production.
State export baskets are astonishingly narrow once disaggregated. Five of the fifteen largest exporters have a single HS2 chapter account for more than 30% of the state's outbound value. Hausmann & Hidalgo's (2009, PNAS) product-space logic scales down to the subnational level: states specialise where they have productive knowledge, and the dominant chapter reveals which industrial complex the state is plugged into. Aircraft in Washington and Kentucky comes from Boeing Everett and GE Aviation respectively; vehicles in Michigan, Indiana, South Carolina, and Alabama trace the Detroit-to-Upper-South auto corridor; precious stones in New York is the Manhattan diamond wholesale trade; fish in Alaska is the Bering Sea cold-water fishery.
| state | top HS2 | industry | value | share |
|---|---|---|---|---|
| TX · Texas | 27 | Mineral fuels & oil | $217.1B | 47.8% |
| CA · California | 84 | Machinery | $33.5B | 18.2% |
| NY · New York | 71 | Precious stones & metals | $34.7B | 36.9% |
| LA · Louisiana | 27 | Mineral fuels & oil | $48.3B | 55.7% |
| IL · Illinois | 84 | Machinery | $13.8B | 16.8% |
| FL · Florida | 85 | Electrical machinery | $12.7B | 17.5% |
| MI · Michigan | 87 | Vehicles | $23.1B | 36.8% |
| IN · Indiana | 30 | Pharmaceuticals | $13.6B |
The Herfindahl-Hirschman Index summarises how concentrated a state's export basket is across HS2 chapters: HHI = Σ_c share_c², where shares are in percent. A perfectly diversified state across 100 equal chapters would score 100; a state that exported a single chapter would score 10,000. Kemeny & Storper (2020) argue that specialisation in high-complexity activities is precisely what sustains superstar-region premia, but narrow specialisation in a commodity chapter is a different beast, closer to the resource-curse geography of North Dakota's Bakken, Louisiana's Gulf Coast LNG, and Alaska's North Slope. The map colours states by HHI across HS2 chapters in 2024: the deep-specialisation oil-and-fish periphery versus the diversified-manufacturing core.
Texas and California together carry roughly one-third of US goods exports. They do it with almost opposite industrial mixes: Texas is an oil-and-chemicals complex that the shale and LNG build-out scaled up through the 2010s, while California is a diversified tech-and- agriculture basket. The figure below contrasts each state's top-8 HS2 chapters in 2024 as share of the state's total exports; the share numbers are cleanly divergent across the board. Moretti's (2012) 'divergent geography' framing applies squarely: the two largest export engines in the country run on different fuels.
US exporters do not face a single tariff when they land abroad; they face the destination country's MFN schedule, weighted by the chapter mix of the state's export basket. Autor-Dorn-Hanson (2013, AER) built their China-shock identification on exactly this kind of regional-mix weighting. For each state we compute taus = Σh (vs,h / Vs) × MFNhworld where MFNhworld is the simple average MFN tariff across all WTO reporters on HS2 chapter h in the latest TRAINS year, and vs,his the state's HS2 export value in 2024. Higher tausmeans the state ships more of what the world taxes heavily, typically agriculture and food, and less of what countries keep near zero, like aircraft and semiconductors. Fajgelbaum & Khandelwal (2022, Annual Review of Economics) use the same construction to benchmark state-level retaliation exposure in the 2018-2019 trade war.
A standing question in regional development is whether export diversification or specialisation drives growth. Imbs & Wacziarg (2003, AER 93(1): 63-86) document an inverted-U pattern across countries: economies first diversify as they develop, then re-specialise at high income. At the US-state level over a single decade the question is simpler: do states with concentrated export baskets (high HHI in 2024 across HS2 chapters) compound at a different rate than diversified states over 2015-2024? The scatter below plots each state on the HHI-CAGR plane. Commodity-state outliers in the upper-right (concentrated and fast-growing in current dollars) are the shale-and-LNG capacity story; manufacturing states in the lower-left (diversified but stagnant) are the Autor-Dorn-Hanson (2013) regional adjustment story written on the export side.
US export geography is not a diffuse national average: it is a Texas-California duopoly plus a set of narrow commodity and aerospace specialists, with a broad middle of manufacturing states whose export engines have been idling or contracting in current dollars for a decade. The pattern aligns with Autor-Dorn-Hanson's (2013) China-shock geography on the import side, Kemeny-Storper's (2020) superstar-region geography on the level side, and Moretti's (2012) divergent labour-market story on the outcome side. Every number on this page is reproducible from the two Parquet tables referenced in the SQL blocks; both derive from the Census Bureau foreign-trade state-exports API at daily frequency.
Method: state totals and HS2 chapter breakdowns are pulled directly from the Census Bureau foreign-trade state-exports endpoint (api.census.gov/data/timeseries/intltrade/exports/statehs), ALL_VAL_YR at December of each year = cumulative January-December total in current USD. Origin of Movement basis: the state where the goods began their journey to the port of export, not the state of production. This is the standard Census series cited in DataWeb, Trade.gov, and the BEA annual International Transactions release. 50 states + DC, 510 state-year observations 2015-2024. Heald-Hathaway (2025, JOLE forthcoming) document how state-level Section-301 exposure scales with the HS-weighted share of each state's export basket in tariffed chapters; Autor-Dorn-Hanson (2013) apply the analogous import-side weighting. A state-level export-weighted tariff is constructed as tau_s = Σ_h (v_s,h / V_s) × tau_h, where v_s,h is state s's exports in HS chapter h and tau_h is the foreign MFN schedule; Figure 6 applies this construction against the world MFN simple-average schedule from WITS TRAINS. Extending to retaliatory (e.g. China List 3) rates is the natural next step.
| 22.5% |
| WA · Washington | 88 | Aircraft & spacecraft | $17.4B | 30.2% |
| OH · Ohio | 84 | Machinery | $9.9B | 17.3% |
| PA · Pennsylvania | 30 | Pharmaceuticals | $7.6B | 14.3% |
| GA · Georgia | 88 | Aircraft & spacecraft | $10.9B | 20.3% |
| KY · Kentucky | 88 | Aircraft & spacecraft | $18.8B | 39.2% |
| NJ · New Jersey | 71 | Precious stones & metals | $5.8B | 13.4% |
| NC · North Carolina | 30 | Pharmaceuticals | $11.9B | 27.8% |
| TN · Tennessee | 84 | Machinery | $7.1B | 18.0% |
| SC · South Carolina | 87 | Vehicles | $12.7B | 33.5% |
| MA · Massachusetts | 84 | Machinery | $6.9B | 19.7% |
| OR · Oregon | 85 | Electrical machinery | $13.3B | 39.4% |
| AZ · Arizona | 85 | Electrical machinery | $7.2B | 22.1% |