Industrial Production & Capacity Utilization (G.17)

The scoreboard of American factory activity, kept by the board that can watch it in real time. Industrial Production & Capacity Utilization (G.17) is the Federal Reserve Board's monthly read on U.S. industrial output: a family of roughly 300 production indexes spanning every major market and industry group, each paired with a capacity utilization rate saying how much of the nation's installed capacity is actually running. Every series enters through the same door - a benchmark value-added weight, restated at successive reference years (2022, 2017, 2012, 2007, 2002, 1997) - so a line like Machinery (NAICS 333, $218,390 million of benchmark value added) can be compared honestly against fabricated metals, primary metals or any other branch of manufacturing. The full historical database reaches back to 1919, about 1,270 monthly observations deep.

For the Industrial Gases slice this is upstream context with teeth: gas demand - hydrogen for refineries, oxygen for steel and glass, nitrogen for everything that must stay inert - moves with manufacturing throughput, and the G.17 is the standard monthly measure of that throughput. In Datadory's catalog of 1,744 datasets across 159 viable industries this record sits in the Industrial Gases slice and scores 9/10 for quality, top band, with a verified field dictionary behind it. [Get a sample of this dataset](#request) and we cut the series to the industry groups and months you name.

Four rows pulled during the August 2026 research pass, showing the weighting architecture exactly as it presents itself:

table : TABLE 1 - Industrial Production: Market and Industry Groups
series: Total index               code: B50001   value_added_2022_usd_m : 4,258,456   proportion : 100.00%

table : TABLE 1
series: Machinery                 code: G333     naics: 333   value_added_2022_usd_m : 218,390   proportion : 5.13%

table : TABLE 1
series: Fabricated metal product  code: G332     naics: 332   value_added_2022_usd_m : 248,765   proportion : 5.84%

table : TABLES 7 and 8 - Capacity Utilization
series: Machinery                 code: G333     naics: 333   util_rate : monthly percent of installed capacity in use

The rows expose how the index is built before any level arrives: every series carries its Federal Reserve code, its NAICS mapping and its benchmark value added, and the proportion column states how much of the total index each line explains. Machinery weighs in at 5.13% of the $4.26 trillion base; fabricated metal products carry 5.84% - together just under 11% of U.S. industrial production by value added. The same G333 line reappears in the capacity tables, where its utilization rate is tracked alongside capacity growth. Request a sample and the extract returns the monthly index levels and utilization rates for exactly these codes, keyed to the months you name.

Seven fields define the record, all marked verified - mapped against the release documentation rather than guessed from summaries:

• Series Description - the market or industry group a line covers, e.g. Machinery.
• Code - the Federal Reserve series identifier: B50001 for the total index, GMF for manufacturing, G333 for machinery.
• NAICS - the matching North American Industry Classification System code (333 for machinery).
• Value added (millions of dollars) - the benchmark dollar weight for the series, carried for each of the six reference years.
• Proportion - that weight expressed as a share of the total index.
• Seasonally adjusted index - the monthly production level itself, on a 2017=100 basis for current vintages.
• Capacity utilization rate - percent of installed capacity in use, tracked by NAICS industry.

Stage-of-processing and product-detail splits, plus the diffusion-index tables flagged in the catalog size estimate, sit under additional fields on request - their column layout was not published as a formal data dictionary, so they ship once confirmed against your use case rather than promised blind.

• Geography: United States, measured nationally. Sub-national cuts are not part of the record's delivered surface.
• Temporal: monthly observations from 1919 in the full historical database - roughly 1,270 months - with periodic annual revisions restating recent years, so long panels stay consistent across vintages.
• Granularity: national monthly indexes by market group and NAICS industry, machinery carried at NAICS 333; stage-of-processing and product detail join via the on-request layer.

That combination - a century of history, restated on known reference years - is what makes the record useful as a cycle yardstick rather than a news ticker: any current print can be placed against the 2008-09 collapse, the 2010s expansion or the post-2020 rebound using like-for-like weights.

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

You pick the channel and the cadence; the field dictionary above travels unchanged across all three. Teams rebuilding a century-long panel tend toward file extracts loaded once and joined against their own demand or revenue tables; products surfacing a macro gauge inside an app take a scoped feed; analysts running models in SQL take warehouse load. Name the industry-group codes and months when you request the sample - the sample ships first either way, and changing cadence afterward is a settings conversation, not a re-integration project.

Production and utilization together answer two different questions - how much is being made, and how close the economy is running to its ceiling - which is why the record shows up in unrelated teams' workflows:

• Industrial gas demand planning - model hydrogen, oxygen and nitrogen offtake against manufacturing throughput instead of lagging shipment reports; the monthly index is the throughput variable most gas-demand models are missing.
• Capex-cycle positioning - read machinery production (G333) and its utilization rate as a real-time gauge of equipment investment, ahead of capital-goods order books turning into shipments.
• Equity research overlays - map utilization onto margins for operating-leverage-sensitive sectors: machinery, fabricated metals, primary metals, chemicals.
• Macro dashboards and recession timing - track the total index and aggregate utilization as cycle-state variables with nearly eight decades of comparable history.
• Credit and risk work - treat sustained utilization declines at NAICS 333 borrowers as an early deterioration signal before it reaches financial statements.
• Model features - feed a tidy, seasonally adjusted, century-deep panel into forecasting work as an exogenous variable, with the six-benchmark weight history keeping backtests honest.

Investors and quant researchers get the canonical monthly cycle variable - production levels and utilization rates ready for factor construction and regime detection. Market researchers and consultants get a century of restated history for sizing and cycle-positioning work that survives scrutiny. Data scientists and ML engineers get a seasonally adjusted panel with codes, NAICS mappings and weights already attached, so no derivation leaks into the pipeline. Sales and growth teams selling into manufacturing get a demand-direction check before committing outreach budget to capital-intensive accounts. Competitive intelligence and product teams get the market backdrop separating a rising tide from a good quarter. Journalists, academics and students get figures attributed to the Federal Reserve Board, citable without footnote gymnastics. All delivered daily, weekly, or hourly.

Provenance note - the G.17 is compiled by the Federal Reserve Board (Board of Governors of the Federal Reserve System), the same statistical programme behind the U.S. financial accounts. It is the reference series other output measures are benchmarked against, not the reverse.

Revision note - weights are benchmarked at successive reference years (2022, 2017, 2012, 2007, 2002, 1997) and annual revisions restate recent years. Confirm the current vintage's base year from the latest release notes before quoting absolute index levels; growth rates and utilization comparisons are robust across vintages.

Completeness note - the seven-field dictionary on this record is verified. The stage-of-processing, product-detail and diffusion-index cuts lack a published column dictionary, so they are quoted under additional fields on request and confirmed before anything ships.

Where to go next - for trade flows behind the production numbers, pair with Eurostat International Trade in Goods (Comext); for the demand side of the same equipment cycle, AMT U.S. Manufacturing Technology Orders (USMTO) tracks orders before they become production; for European gas-demand modelling, eGon industrial gas demand for Germany covers the CH4 and H2 side; and the head-to-head of the two flagship records lives at Eurostat Comext vs G.17.