Freight Analysis Framework Version 5 (FAF5)

Freight Analysis Framework Version 5 (FAF5) is the US DOT's origin-destination freight database, produced by the Bureau of Transportation Statistics with support from the Federal Highway Administration and developed at Oak Ridge National Laboratory. It integrates the 2017 Commodity Flow Survey, Census Bureau international trade records and sector-specific sources into a single consistent frame covering every mode - truck, rail, water, air, pipeline and their multimode combinations - moving among US states and major metropolitan areas.

For ground-cargo work the frame answers the question carrier-keyed feeds cannot: which commodities move between which regions, in what tonnage and dollar volume, by which surface mode, and how that mix compounds over time. Release FAF5.7.1 holds a 2017 base year, 2018-2024 estimates and forecasts running to 2050. Datadory scores it 10 out of 10 on its quality rubric - a mark held by just 145 of the 1,744 cataloged datasets against a 7.81 average. Get a sample of this dataset and judge the rows before anything ships.

Grid-shaped, keyed, flat - one origin-destination commodity flow per observation. Straight off the verified dictionary:

dms_orig dms_dest sctg2 dms_mode trade_type dist_band      tons   value
111      311      35    1        1          001-250       1250.4 23800.0
# domestic move: FAF zone 111 -> zone 311, electronics, truck, shortest band

Read the anatomy rather than the digits - every number above is the documented example value for its column, not a live figure. An observation resolves its coordinates first: origin zone (dms_orig), destination zone (dms_dest), two-digit SCTG2 commodity (sctg2), mode (dms_mode, where 1 means truck), trade type (1 domestic, 2 import, 3 export) and optionally the distance band. Only then do measures attach - thousand tons, million dollars at 2017 constant prices, current-dollar value and million ton-miles.

Import and export rows swap the domestic keys for foreign-region codes and carry fr_inmode or fr_outmode for the leg on either side of the border:

fr_orig  dms_orig sctg2 fr_inmode trade_type
8101     111      35    1         2
# foreign-origin electronics entering at entry region 111, truck on the inbound leg

Because every year repeats the identical grid, stacking observations builds a commodity-by-corridor panel without reshaping, and forecast years simply add scenario columns beside the mid-range figures.

Fifteen verified variables define the flow grid, every definition below checked against the official data dictionary during the cataloging pass. Four columns do most analytical work. dms_orig and dms_dest pin the corridor at FAF-region scale, so Chicago-to-Dallas questions resolve once instead of county by county. sctg2 carries the 43 commodity groups - electronics, pharmaceuticals, machinery, mixed freight - that decide what belongs on a truck at all. dms_mode isolates truck (code 1) from rail, water, air, pipeline and multimode combinations, and trade_type separates domestic flows from imports and exports, each side carrying its own cross-border mode column. The measures - tons, value, current_value and tmiles - sit on top of that key, with dist_band adding the distance cut.

Additional cuts on request. The high and low growth scenario columns live in the expanded forecast cut, a state-level geography cut covers 50 states plus DC, and county-level experimental estimates extend the grid one notch finer. Field definitions meet the verified standard held by 1,495 of the 1,744 datasets (85.7%) in the Datadory catalog.

Geography - 132 domestic FAF regions: states, the state portions of large metropolitan areas, and remainders of state, plus eight foreign regions for trade flows. Zones are metropolitan-scale by design, so a major gateway spans its whole metro rather than stopping at a city limit. A parallel state-level cut covers all 50 states plus DC when state frames suit the question better.

Temporal - a base year of 2017, recent-year estimates for 2018 through 2024, and forecasts at five-year increments from 2030 to 2050 under a mid-range path flanked by high and low economic growth scenarios. Reprocessed annual state-level history runs back through 1997-2012, so a single project can span a quarter century of structural change in how American freight moves.

Granularity - origin-destination pair by two-digit SCTG2 commodity by mode by trade type by year, optionally banded by shipment distance. That ladder serves a single-corridor commodity audit and a national modal-share model off the same base, which is why teams rarely outgrow it.

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

You choose the slice - one commodity's corridor history, the truck-only cut across all 132 regions, or the entire national grid with forecasts - and the shape: flat files for analysts, a feed for running pipelines, or landed tables in your own warehouse. Most teams take the historical depth once as a backfill and keep new vintages rotating on whatever cadence their models expect. Datadory normalizes the structural quirks described above - foreign-versus-domestic key swaps, mode-code semantics, scenario columns appearing only in forecast years - and attaches the full mode, commodity and region code sheets so no cell arrives as an unexplained integer.

• Market sizing - dollar-weighted corridor totals by commodity and distance band give ground freight a denominator with a federal pedigree; see market researchers x cargo ground transportation.
• Network planning - rank zone-pairs by tons, value or ton-miles before siting distribution capacity; metropolitan-scale endpoints make the ranking lane-specific rather than state-vague.
• Demand modeling - the repeated yearly grid feeds lane-choice and freight-demand models directly; part of the shelf serving supply-chain mapping and demand forecasting.
• Modal-share work - seven explicit mode codes separate truck from every alternative, so modal-shift claims survive scrutiny instead of resting on blended categories.
• Investment diligence - ton-mile trajectories plus the high-low scenario spread stress-test logistics theses; see investors & quants x cargo ground transportation.
• Reporting and coursework - government-produced commodity statistics make any freight claim traceable to a published figure; see journalists & academics x cargo ground transportation.

Market Researchers & Consultants get the federal benchmark under every ground-freight TAM model (market researchers view). Data Scientists & ML Engineers get a typed panel whose forecast targets ship beside the features - rare training material (data scientists view). Investors & Quant Researchers get scenario-spread projections that turn freight exposure into a testable range (investors view). Supply-chain strategists & network planners get measured corridor volumes for siting decisions. Developers & Data-Product Builders get a stable fifteen-column schema for lane-lookup features (developers view). Journalists, Academics & Students get attributable national figures with the methodology already written.

Stated plainly, because they shape the analysis:

• Zones, not addresses. FAF-region geography is metropolitan-scale by design; terminal-, carrier- or shipment-level questions need a companion feed layered underneath.
• Forecast years step in fives. Projections land at five-year increments to 2050, so interpolation between milestones is your modeling decision, not something the grid supplies.
• Constant versus current dollars is a real choice. value sits at 2017 prices while current_value is nominal; mixing them manufactures phantom growth rates.
• Distance bands cover the domestic portion only, and ton-miles measure effort, not route miles - routing work still needs a network layer on top.
• International legs classify by surface mode. A door-to-door chain with a flying middle reads partly as truck; decide whether you want lift segments or the full chain before filtering.

Because the raw release splits its dictionary into a separate workbook, swaps key columns between domestic and trade flows, scatters scenario columns across forecast cuts, and expects you to reconcile geography variants yourself. Datadory normalizes all of it into the single field dictionary above, resolves code sheets onto every row, flags which geography frame and which scenarios a delivery carries, and schedules recurring deliveries on your cadence. Browse the shelf on the cargo & ground transportation data hub or the best cargo ground transportation datasets list, then get a sample cut to your commodities, corridors and years.

Neighbors that bracket what the grid leaves open. Commodity Flow Survey (CFS) 2022 is FAF5's own upstream - shipment-level microdata where FAF5 publishes a finished matrix; the head-to-head runs on our CFS vs FAF5 comparison page. FHWA's Freight Performance Measurement Program adds highway travel-time reliability between the endpoints the grid only counts. ATA Economics' truck tonnage measures track the month-to-month pulse between framework vintages. Data.gov's freight catalog reaches everything else published on US freight, and UK road freight statistics extend the same questions across the Atlantic. Together: the matrix, the microdata, the highway reality, the monthly pulse and the international comparison.