Guide

Industrial production explained

U.S. industrial production (IP) is the Federal Reserve’s monthly index of physical output at factories, mines, and utilities — a hard quantity measure that complements survey-based indicators like the Purchasing Managers Index (PMI). Published in the G.17 statistical release, IP tracks how many units of goods and energy the industrial sector produces, not dollar sales or profits. Alongside IP, the Fed publishes capacity utilization — the share of installed plant and equipment actually in use. Utilization near long-run averages signals balanced supply; sustained readings above ~80% in manufacturing historically correlate with pipeline inflation pressure. IP feeds directly into BEA estimates for GDP industrial output and helps economists judge whether producer price moves reflect demand strength or cost shocks alone. This guide covers the three major sectors, index construction and base years, manufacturing subsector detail, capacity slack and inflation links, seasonal adjustment, the relationship to GDP and PMI, a Harbor Manufacturing monthly read worked example, an indicator decision table, common pitfalls, and an investor checklist.

What industrial production measures

The Federal Reserve Board compiles IP from physical product data — production hours, electric power generation, mine tonnage, and similar volume metrics — weighted by value added in the 2017 NAICS structure. The headline index is set to 2017 = 100. Month-over-month and year-over-year percent changes are seasonally adjusted and published around the 15th of each month at 9:15 a.m. ET (often the same morning as retail sales on the economic calendar).

IP is a real volume index — inflation-adjusted at the sector level using producer price deflators. A 0.3% monthly gain means factories and utilities produced 0.3% more physical output, not 0.3% more revenue. That distinction matters when comparing IP to nominal retail sales.

Three major sectors

Manufacturing — roughly 78% of the IP index. Covers NAICS manufacturing: motor vehicles, machinery, chemicals, computers, food processing, aerospace, and hundreds of other factory categories. Often reported as “manufacturing IP” separately from the total.
Mining — about 14% of IP. Oil and gas extraction, coal, metal ore, stone, and nonmetallic minerals. Highly volatile with commodity prices and rig counts; weather and Gulf Coast hurricanes move monthly prints.
Utilities — about 8% of IP. Electric power and natural gas distribution. Driven by heating and cooling degree days — a cold January can spike utilities IP even when factories are quiet.

Total industrial production combines all three. Markets frequently focus on manufacturing IP excluding motor vehicles and parts — autos are large and lumpy (plant shutdowns, chip shortages, model-year changeovers), similar to how retail sales watchers strip autos from headline consumption.

Capacity utilization — slack, bottlenecks, and inflation

Capacity utilization equals actual output divided by estimated maximum sustainable output at existing plants. The Fed estimates capacity from Census surveys, trade association data, and engineering ratios — updated less frequently than monthly production.

How to read utilization levels

Total industry utilization — long-run average (1972–2023) is about 79.6%. Readings persistently below average signal slack — idle equipment, room to grow without new investment.
Manufacturing utilization — long-run average near 78%. The manufacturing-only rate is what macro traders watch for goods inflation pressure.
Above long-run average — historically associated with rising goods prices and tighter labor markets in durable goods plants; not a precise trigger but a useful context variable for Fed watchers.
Below average for extended periods — suggests weak capex demand, excess inventory cycles, or structural decline in heavy industry share of GDP.

Utilization is not a real-time inflation forecast by itself. Energy shocks can push mining IP and utilization up while manufacturing weakens. Pair utilization with PPI final demand, wage growth in goods-producing industries, and PMI Prices Paid before drawing rate-path conclusions.

Manufacturing subsectors — what moves the index

The G.17 release publishes IP and utilization for major market groups and selected industries. Investors map subsectors to equities and supply-chain themes:

Motor vehicles and parts — largest single manufacturing category; dominates headline volatility. Watch ex-auto manufacturing IP for underlying factory momentum.
High-tech — computers, communications equipment, semiconductors. Correlates with capex cycles and electronics demand; sensitive to inventory corrections.
Chemicals — petrochemicals link to energy prices; fertilizers tie to agriculture.
Machinery and electrical equipment — business investment barometer; lags orders data from PMI New Orders by one to three months.
Aerospace and miscellaneous transportation — Boeing production schedules and defense budgets move this slice.
Consumer goods — split into durable (appliances, furniture) and nondurable (food, clothing, paper). Durable consumer IP connects to retail sales categories with a production-to-shelf lag.
Construction supplies — cement, lumber processing, HVAC equipment; ties to housing starts and nonresidential building.

Month-over-month changes in individual subsectors can exceed total manufacturing IP because weights differ. A semiconductor surge plus auto shutdown can net to a flat headline — read the Fed’s supplementary tables, not only the top-line paragraph in news wires.

Industrial production vs PMI — surveys vs hard data

The ISM and S&P Global manufacturing PMI are diffusion indexes from purchasing manager surveys — “better, same, or worse than last month.” IP measures actual physical output. The two series usually directionally agree over three to six months but diverge frequently in individual months:

PMI leads IP at turning points — New Orders and Production PMI components often soften one to two months before IP confirms the slowdown. PMI below 50 does not guarantee negative IP; it signals more managers reporting contraction than expansion.
IP is revised; PMI is not — IP incorporates updated source data across multiple months. PMI is a one-time survey snapshot.
Autos and utilities distort IP — PMI is manufacturing-only and excludes utilities entirely. Compare manufacturing IP ex-autos to PMI for the cleanest pairing.
Inventory swings — PMI can stay strong while IP flatlines if firms build inventory without shipping (or vice versa during destocking).

Macro desks use PMI as an early warning and IP as confirmation. Three consecutive months of contracting manufacturing IP carries more weight for recession watchlists than a single sub-50 PMI print, because IP is a coincident-to-lagging hard output measure tied to GDP accounting.

IP, GDP, and the business cycle

BEA uses Federal Reserve IP data when estimating real value added in manufacturing, mining, and utilities for quarterly GDP. Monthly IP helps nowcast GDP industrial production before the advance GDP release:

Goods-producing share of GDP — industry including construction is roughly 18–20% of U.S. output; manufacturing alone is near 11%. Services dominate, so weak IP alone rarely drives negative GDP — but sharp manufacturing contractions have preceded every post-war recession.
Inventory contribution — IP rising faster than final sales suggests inventory build (positive for current-quarter GDP, potentially negative later). IP falling while orders hold may signal destocking.
Link to employment — manufacturing payrolls in the establishment survey lag IP turns. Compare with unemployment in goods-producing industries.
International context — U.S. IP correlates with global trade volumes and German industrial production; use for multinationals with heavy U.S. plant exposure.

The NBER Business Cycle Dating Committee does not use IP as a primary coincident indicator (payrolls, real income, sales, and industrial production as a group matter). IP is one of four classic coincident indicators in the Conference Board Coincident Index — useful for confirming whether a soft PMI patch became real output loss.

Seasonal adjustment, revisions, and data quirks

Published IP changes are seasonally adjusted. Manufacturing has predictable shutdowns (auto retooling, holiday schedules); utilities follow weather patterns. Residual seasonality can inflate or depress winter months — compare year-over-year percent changes when month-over-month prints look extreme.

Revision pattern

First estimate — published ~15 days after month-end; subject to revision as source data arrive.
Annual revision — each spring/summer the Fed revises IP and capacity back several years using updated Census value-added weights and new capacity surveys. Trend breaks can appear retroactively.
Hurricanes and strikes — Fed footnotes often quantify one-off hits (Gulf Coast refining, UAW stoppages). Strip these when assessing underlying trend.

Mining IP volatility from OPEC decisions, rig counts, and pipeline outages can move total IP ±0.5% in a month with no manufacturing story. Always read the manufacturing-only line and ex-auto detail.

How investors use industrial production release days

G.17 releases at 9:15 a.m. ET — 45 minutes after 8:30 a.m. retail sales when both land the same day. Thin liquidity can persist from the earlier data wave; IP-specific trades focus on manufacturing surprise vs consensus and capacity utilization versus prior month.

Typical asset reactions

Industrials and materials equities — strong manufacturing IP ex-autos supports machinery, chemicals, and transport names; weak IP pressures cyclicals regardless of a prior-day retail sales beat.
U.S. Treasuries — hot IP plus rising utilization can sell off duration (growth/inflation risk); weak IP with falling utilization supports bonds, especially if concurrent with soft PPI.
U.S. dollar — resilient factory output supports USD in relative-growth trades versus regions with contracting manufacturing (e.g., eurozone IP weakness).
Commodities — mining IP and utilization inform near-term metals and energy demand expectations; manufacturing IP confirms whether commodity price moves reflect real consumption.
Fed expectations — capacity utilization trending above long-run average with firm PPI keeps tightening bias; sub-trend IP for three months with rising unemployment opens easing speculation — cross-check monetary policy guidance.

Long-horizon investors weight six-month manufacturing IP trends and utilization versus its long-run mean over single-month surprises driven by weather or autos.

Worked example: Harbor Manufacturing monthly read

Harbor Manufacturing produces precision hydraulic components at three U.S. plants. Each month after the G.17 release, the operations team runs a fifteen-minute checklist — complementing their PPI input-cost review and PMI New Orders tracking:

Read manufacturing IP ex-autos first — if total IP beats consensus on utilities and mining but manufacturing ex-auto misses, Harbor treats it as soft for their machinery end-market — not a capex acceleration signal.
Check capacity utilization for manufacturing — utilization rising toward 79%+ with firm orders justifies overtime and temp hiring; utilization below 76% triggers shift reductions before layoffs.
Scan machinery and electrical equipment subsector IP — Harbor’s customers are OEM machinery builders; this subsector leads their order book by four to eight weeks.
Compare to prior-month PMI Production index — if PMI Production expanded while IP contracted, Harbor flags possible inventory destocking at distributors and holds raw-material buys flat.
Note mining IP for steel and energy inputs — mining strength with weak manufacturing may mean input cost pressure without demand (margin squeeze scenario per their PPI playbook).
Write one paragraph for the ops journal — e.g. “September manufacturing IP −0.2% m/m (cons +0.1%); ex-auto −0.3%. Capacity utilization 77.4% (prior 77.8%). Machinery IP flat. Confirm Q4 overtime freeze; maintain 6-week raw steel inventory; no change to headcount plan pending October ISM.”

Harbor pairs hard output data with survey and price indicators before committing to production schedules — IP confirms whether PMI optimism translated into actual factory activity.

Indicator decision table

Question you have	Best indicator	Why
How much did factories actually produce last month?	Fed manufacturing IP (ex-autos)	Hard volume data; deflated real output
Are plants running near full capacity?	Manufacturing capacity utilization	Slack vs bottleneck signal for goods inflation
Will manufacturing soften before output falls?	ISM PMI New Orders and Production	Survey leads IP by one to three months at turns
Are wholesale input costs rising?	PPI final demand	Price index; IP tells you if volume supports pass-through
Total GDP industrial contribution?	BEA real GDP goods-producing / IP trend	IP nowcasts GDP manufacturing before advance print
Energy and commodity extraction volume?	Mining IP subindex	Separate from factory story; volatile with oil rigs
Weather-driven power demand?	Utilities IP	Heating/cooling driven; can distort total IP
Consumer goods production vs retail sales?	Consumer goods IP + retail sales control group	Production leads shelf sales; inventory bridge
Official recession dating?	NBER coincident index bundle	IP alone does not date recessions

Common pitfalls

Trading total IP while ignoring manufacturing ex-autos — utilities weather spikes and mining rig cycles distort the headline.
Equating one month of IP with recession — autos, hurricanes, and strikes create noise; use three- to six-month manufacturing trends.
Ignoring annual revisions — spring benchmark revisions can rewrite whether a quarter looked expansionary.
Assuming high utilization guarantees CPI acceleration — goods are a shrinking CPI weight; services inflation can dominate even with hot factories.
Comparing U.S. IP levels to eurozone IP without base-year context — index levels differ; use percent changes and national conventions.
Expecting IP and PMI to match monthly — surveys and hard data diverge during inventory swings; judge over quarters.
Forgetting IP is real but sectorally deflated — not directly comparable to nominal revenue or orders dollars.
Overlooking concurrent retail sales on the same morning — markets may have already priced a consumption narrative at 8:30; IP at 9:15 can reverse sector rotations.

Investor and analyst checklist

Track manufacturing IP, ex-auto manufacturing IP, and total IP as separate series.
Plot capacity utilization against its long-run manufacturing average (~78%).
Read Fed G.17 tables for market groups: consumer goods, business equipment, construction supplies.
Cross-check month-over-month IP with year-over-year to filter residual seasonality.
Compare to ISM Manufacturing Production and New Orders for the same month.
Note Fed footnotes for hurricanes, strikes, and unusual weather impacts.
Bridge IP trends to PPI final demand and industrial payroll growth.
Use six-month annualized IP growth for GDP nowcasting, not single prints.
Review annual revision history each summer for changed cyclical interpretation.
Pair with economic calendar timing — 9:15 a.m. ET release, often with retail sales same day.

Key takeaways

Industrial production is the Fed’s real-volume index of factory, mine, and utility output — G.17, 2017 = 100.
Manufacturing dominates IP; ex-auto manufacturing is the cleanest cyclical read.
Capacity utilization measures slack versus sustainable maximum output — context for goods inflation and capex.
PMI surveys tend to lead IP at turning points; IP confirms hard output for GDP nowcasts.
Strip utilities weather noise and mining commodity volatility before drawing trend conclusions.