US Census International Trade Data - Schedule B 8429 Construction Machinery Exports

It is the US government's own account of America's earthmoving-machinery trade, one tidy row at a time. The Census Bureau is the official compiler of US export and import statistics, and commodity detail runs to the 10-digit Schedule B level for exports. Heading 8429 of the Schedule B 2026 chapter 84 file covers "self-propelled bulldozers, angledozers, graders, levelers, scrapers, mechanical shovels, excavators, shovel loaders, tamping machines and roadrollers", and it breaks into queryable subheadings: 8429.11 track-laying bulldozers (split new versus used/rebuilt), 8429.19 other bulldozers, 8429.20.0000 graders and levelers, 8429.30 scrapers (divided at a 13.7 cubic-meter bucket threshold) and 8429.40 tamping machines and road rollers (vibratory versus other). Each carries its own 10-digit number and unit of quantity.

Every observation joins three axes: a commodity code, a trading partner and a month. Roughly 17,000 active 10-digit export codes exist in a typical year, crossed with around 240 partner countries, and each cell carries dollar value for the month and year-to-date, unit counts, and vessel, air and containerized value splits. The underlying records originate from Electronic Export Information filed in Customs' ACE system - real shipment declarations, not estimates - which is why analysts treat these figures as the ground truth of the trade.

Flat, monthly, one row per commodity code per partner - shaped to load straight into a warehouse table. Illustrative rows in the delivered column order:

year: 2026   month: 6   e_commodity: 842911   comm_desc: Track-laying bulldozers and angledozers, new
cty_name: Canada   unit_qy1: NO.   all_val_mo: 12345678   all_val_yr: 98765432   qty_1_mo: 450
ves_val_mo: 9000000   air_val_mo: 500000   cnt_val_mo: 7000000   last_update: 2026-08-06

year: 2026   month: 5   e_commodity: 842940   comm_desc: Tamping machines and road rollers, vibratory
cty_name: Australia   unit_qy1: NO.   all_val_mo: 4210550   all_val_yr: 18930400   qty_1_mo: 96
ves_val_mo: 4180300   air_val_mo: 0   cnt_val_mo: 30250   last_update: 2026-08-06

year: 2026   month: 6   e_commodity: 842952   comm_desc: Self-propelled shovel loaders, new
cty_name: Mexico   unit_qy1: NO.   all_val_mo: 6874300   all_val_yr: 31204880   qty_1_mo: 214
ves_val_mo: 1204000   air_val_mo: 180500   cnt_val_mo: 5489900   last_update: 2026-08-06

The block already says something. Canadian buyers took $12.3 million of new track-laying dozers in June alone - $9 million of it moving by vessel - while Australian demand concentrated almost entirely in ocean freight, as you would expect for 20-tonne machines crossing the Pacific. The Mexico row shows the opposite pattern: nearly 80% of the value moved containerized, consistent with cross-border parts-and-units flows. Value divided by units gives an implied price per machine - here roughly $27,000-$44,000 average per unit across the three rows - a free sanity check on any pricing assumption. Request a sample and you get rows scoped to your own code list and country set.

Sixteen documented fields define the observation: two commodity descriptions, the aggregation-controlled code itself, monthly and year-to-date dollar values, quantity with its unit code, partner identifiers, three transport-mode value splits, customs district, calendar keys and a refresh timestamp. Field definitions are verified against the publisher's own variable metadata rather than inferred from documentation prose.

The dictionary below is the verified core. Deeper columns confirmed during sample preparation fold out on request rather than being promised blind.

Geography - United States totals first, then three ways to cut them: by partner country (roughly 240 of them, four-character codes and names included), by US state, and by customs district or port on the port-level endpoints. That last cut matters for logistics people: it shows not just that $12 million of dozers went to Canada but through which border district they actually crossed.

Temporal - January 2010 to present on the trade endpoints, extended every month on press-release day, with each record stamped by a LAST_UPDATE date so revisions are visible instead of silent. The companion end-use series, seasonally adjusted, reach back to 1994 when your analysis needs a longer arc than fifteen years.

Granularity - monthly flows per commodity code at HS2, HS4 or HS6 aggregation up to full 10-digit Schedule B detail, crossed with country, state or port. You choose the aggregation level; the schema underneath does not move, so a dashboard built at HS6 keeps working when you drill to 10 digits.

API, files, or your warehouse. Daily, weekly, or hourly.

Pick the channel your team already works in: REST-style queries for live lookups against a specific code and country, flat files sized for overnight warehouse loads, or a direct pipe into Snowflake, BigQuery or Redshift. Cadence is yours to set - and to change when your models change.

Every delivery ships with the full field dictionary, sample rows for validation, and a schema that holds steady between refreshes.

• Partner-country demand tracking - rank the markets actually buying American dozers, excavators and rollers month by month; see our market researchers use cases page.
• OEM demand sensing - monthly dollar values and unit counts are the highest-frequency official proxy for heavy-equipment order books, arriving weeks ahead of annual production surveys.
• Tariff and reshoring reporting - ground every headline in 10-digit official figures with a named government publisher behind them, the standard newsroom-data-desk move.
• Import-surge monitoring - watch the mirrored import side for exactly the 8429 subheadings where foreign machines threaten domestic share; more on our competitive intel product teams use cases page.
• Quant features and models - a fifteen-year monthly panel with verified definitions drops straight into a feature store; see our data scientists use cases.
• Logistics mix analysis - vessel, air and containerized value splits reveal how heavy iron physically reaches each market, useful to freight planners and equipment financiers alike.

Data scientists and ML engineers get a fifteen-year monthly panel with stable codes and verified field definitions - no scraping archaeology, no silent schema drift. Investors and quant researchers get the freshest official read on construction-machinery demand, a high-frequency complement to quarterly OEM filings. Market researchers and consultants get country-by-country demand sizing with citable provenance behind every chart. Journalists, academics and students get primary government statistics standing behind tariff and trade stories. Competitive intelligence teams get early warning on import pressure in precisely the subheadings they compete in. Developers building data products get one monthly schema that pipes into dashboards and alerting without per-release surgery.

Four things worth knowing upfront.

First, exemptions create blind spots. Export figures derive from Electronic Export Information filed in Customs' ACE system, and very low-value shipments that qualify for exemption never enter the record - irrelevant for bulldozers, material for anyone joining this to small-parts trade.

Second, the import mirror speaks a different dialect: import detail is keyed to HTS codes rather than Schedule B numbers, so a two-way analysis needs a code mapping, which we supply with the sample.

Third, monthly figures get revised in later releases, and the LAST_UPDATE field makes that visible. Build pipelines that re-pull rather than cache if finality matters to your series.

Fourth, classification follows configuration - the scraper line splits at a 13.7 cubic-meter bucket threshold, so a machine can change codes when its bucket changes. Anyone comparing multi-year series should aggregate to subheading level before drawing trend conclusions.