Capacity Utilization

Capacity Utilization measures actual US industrial output against its sustainable maximum. That maximum is what a plant can sustain on a realistic schedule with normal downtime and available inputs, not a brief unsustainable peak. The Federal Reserve Board publishes it monthly in G.17, Industrial Production and Capacity Utilization.

The system combines production indexes with end-of-year rates from outside surveys, led by the Census Bureau Quarterly Survey of Plant Capacity Utilization.

It covers manufacturing, mining and utilities only, never services, construction or farming. It divides seasonally adjusted output by estimated capacity, so it is a computed ratio, not a poll of managers like purchasing manager indexes.

Source data describe the reference month. First estimates appear near the middle of the next month at 9:15 am Eastern time, while the annual revision comes at noon. Worker hours cover the pay period holding the 12th, and physical data become daily averages over working days.

Each month gets 6 estimates, 1 preliminary plus 5 monthly revisions, with about 76% of source data in the first estimate. Releases bring a summary, tables, charts and machine readable data, and series face yearly revisions with census, capacity and seasonal data. Yearly capacity gets a first estimate in February and later aligns with the yearly benchmark revision.

Components combine by value added weights.

• Manufacturing, 73.22%: core goods output, tracked through plant surveys with capital input and age.
• Durable manufacturing, 36.68%: vehicles, machines and electronics, the most cyclical part.
• Nondurable manufacturing, 34.82%: food, chemicals, paper and petroleum products for daily use.
• Other manufacturing, 1.72%: logging plus publishing kept from an older classification.
• Mining, 16.01%: oil, gas, coal and minerals, measured mainly from physical output data.
• Utilities, 10.77%: electric power and gas distribution, weather sensitive and usually the lowest use.

The May 15, 2020 release for April 2020 data showed shutdowns dragging industrial output down 11.2%, its largest monthly fall in 101 years. Manufacturing led with a 13.7% drop, and vehicle output fell over 70%, while mining fell 6.1% and utilities slipped 0.9%. Total use fell 8.3 points to 64.9%, 14.9 points below its long run average and 1.8 points below the 2009 low recorded since 1967. Manufacturing sat at 61.1% against 81.7% for mining and 71.1% for utilities, so parts of industry lived in different worlds.

1. Production indexes: yearly benchmarks set each series trend, while physical counts or production worker hours set monthly moves. Series aggregate since 1972 as value added weighted Fisher indexes.
2. Capacity indexes: December implied capacity equals the output index over the survey rate. An index of 100 with an 80% survey means 125, or 100 / 0.8 = 125. Final capacity comes from regressions on capital services and age or on physical measures.
3. Aggregate capacity: yearly use averages individual rates by capacity weight, then yearly capacity equals output over use. Monthly capacity interpolates between year end estimates with a Fisher index.
4. Utilization rate: the rate equals seasonally adjusted output over capacity times 100, and the ratio itself is unadjusted. Output of 100.8 with capacity of 132.6 means 76.0%, as an illustration.
5. Seasonal adjustment: each series adjusts alone with X-13-ARIMA-SEATS, and aggregates are Fisher indexes of adjusted parts.