U.S. industrial output rose.
The factory story was more uneven.
September 2026 U.S. industrial output rose 0.7%, but manufacturing gained 0.3% and autos fell 3.0%. Read the Fed data beyond the headline.
A stronger headline, a divided production floor
U.S. industrial production increased 0.7% in September 2026, the Federal Reserve reported on October 16. That was a stronger monthly reading than August's revised 0.1% gain. But the headline does not describe a uniform factory surge: manufacturing rose 0.3%, mining rose 0.2%, and utilities jumped 3.5%. Motor vehicles and parts moved in the opposite direction, falling 3.0%. [1][2]
The most useful reading is therefore both positive and qualified. More industrial output was produced in September, while the composition of that increase differed sharply across sectors. Electricity and natural-gas utilities supplied a conspicuously strong movement; the factory component grew more modestly, with gains in some industries offsetting weakness in others. A national index can rise without describing the experience of every plant, worker or supplier.
For the third quarter, total industrial production increased at a 2.8% annual rate. Manufacturing's corresponding quarterly rate was just 0.7%. These are annualized quarter-to-quarter changes, not September's monthly changes and not forecasts for the next year. Keeping those time periods separate is essential to understanding why a relatively strong September headline can coexist with a much less dramatic manufacturing quarter.
Industrial production is not another sales total
The Fed's industrial production index measures real output in manufacturing, mining, and electric and gas utilities. It is designed to describe production volume rather than current-dollar receipts. That makes it a different lens from the Census Bureau's retail-and-food-services sales report, which measures dollar sales and is not adjusted for price changes. [3][4]
The distinction is especially relevant this week. Our earlier analysis of September retail sales showed that gasoline-station receipts accounted for most of the estimated monthly dollar increase. A rise in receipts can reflect prices, quantities and the mix of products sold. Industrial production asks a different question: how much output did the covered industries produce? Readers should not subtract one headline from the other or treat them as competing estimates of the same object. [5]
The coverage also differs. G.17 is not a measure of the entire economy. It does not include the full range of services or construction activity itself, although it does report production of construction supplies. A 0.7% rise in industrial output is not a 0.7% rise in GDP. The industrial sector is economically informative, but it is a defined slice of national activity rather than a substitute for a comprehensive output measure.
Utilities moved much faster than factories
The utilities index rose 3.5% in September, following increases of 1.1% in July and 1.5% in August. The detailed industry table shows electric utilities up 3.7% in September and natural-gas utilities up 2.1%. Utilities output was 8.7% above its September 2025 level. Manufacturing, by comparison, was up 1.1% over the year. [1][2]
Those figures establish a difference in measured output, not a complete causal explanation. The release does not assign the utilities increase to a specific share of data-center demand, household cooling, industrial use or another end user. It would be tempting to fold the result into a broad story about artificial intelligence or the weather, but doing so requires evidence beyond the aggregate production table.
Utilities and manufacturing also occupy different weights in total industrial production. The release's 2025 proportions are 12.40% for utilities and 75.89% for manufacturing. A larger percentage move in a smaller component does not translate one-for-one into the headline. Conversely, a large enough movement in that smaller component can materially influence the total. The comparison should preserve both the growth rate and the component's size, rather than ranking percentages alone.
The auto decline sits inside a small manufacturing gain
Motor vehicles and parts fell 3.0% in September. Within durable manufacturing, the overall index nevertheless edged up 0.1%. Primary metals rose 1.6%, machinery rose 1.4%, and electrical equipment, appliances and components rose 1.3%, according to the detailed table. The result was a set of offsetting movements, not an industry moving in one direction. [2]
There are several reasons to resist translating that auto number into an immediate claim about car buyers. Production is not retail sales. Dealers can sell vehicles from inventory, factories can adjust production schedules, and imports and exports connect domestic output to markets beyond U.S. households. The G.17 figure alone does not identify which mechanism drove the September decline or which manufacturer's operations explain it.
The report also contains similarly named but distinct series. Motor vehicles and parts is an industry grouping. Automotive products is a consumer-goods market grouping, which fell 2.8% in September. The two figures need not match because the classifications answer different questions. One organizes production by the industry making it; the other organizes it by the market in which the output is used. Switching between them without labeling the series can create a false discrepancy.
Nondurables improved in September but not over the quarter
Nondurable manufacturing rose 0.4% in September, led by gains in printing and support and in chemicals. Printing and support rose 1.9%, while chemicals rose 1.0%. But the quarterly picture was weaker: nondurable manufacturing declined at a 1.6% annual rate in the third quarter. A positive latest month does not erase the performance of the preceding months. [1][2]
The industry table gives concrete examples of the variation. Food, beverage and tobacco production increased 0.4% in September but was 1.5% below a year earlier. Paper production fell 1.1% in the month and was 3.2% below September 2025. Plastics and rubber products declined 0.9% in September, while petroleum and coal products increased 0.7%.
These figures should not be turned into a league table of corporate profitability. An industry can produce more while facing higher costs, weaker margins or a less favorable product mix. It can also produce less while selling inventory or changing its operations. G.17 provides the output side of the picture. Earnings, hiring, prices and orders need their own evidence.
The semiconductor number belongs to the quarter
The Fed highlighted a 28.4% annualized third-quarter increase in semiconductors and related electronic components as one contributor to durable manufacturing's 3.8% annualized quarterly gain. That is not a 28.4% September increase. Nor does it mean every electronics industry expanded at that speed. [1]
The broader computer-and-electronic-products industry rose 0.1% in September and was 9.5% above a year earlier. Those numbers describe a wider grouping and different comparison periods. A reader can accurately report all three figures, but only if the labels travel with them: semiconductors in the quarter at an annual rate, broader electronics in the month, and broader electronics over twelve months.
An annualized rate expresses what a quarterly pace would look like if compounded for a year. It is a reporting convention, not a promise that the pace will continue. For perspective, a 28.4% annualized rate corresponds to roughly 6.4% over a single quarter using the standard compounding conversion. That is an explanatory calculation from the published rate, not an additional Fed forecast. The conversion helps keep an impressive number attached to the period it actually describes.
Business equipment was flat, not uniformly weak
Business-equipment output was unchanged in September, but that total combined different movements. Transit equipment declined 1.7%, information-processing equipment declined 0.9%, and industrial and other equipment increased 1.3%. Over twelve months, the business-equipment total was up 6.7%. [2]
The flat monthly figure therefore does not establish that all business investment stopped. It describes domestic output in this market grouping, not a complete investment-spending account. Businesses can purchase imported equipment, draw on inventories or invest in structures and intellectual property that are not represented by this one series. Production of equipment and total capital expenditure are connected but not identical.
This is a useful discipline for interpreting the report's other market groups as well. Construction-supplies output rose 1.0%, business supplies rose 0.9%, and materials rose 0.7%. These are signals about the production of particular inputs and products. They are not direct counts of construction projects, company expansions or newly created jobs.
Capacity utilization rose to 76.9%
Total industrial capacity utilization increased from 76.4% in August to 76.9% in September. The change is 0.5 percentage point, not 0.5% of the previous rate. The September level remained 2.5 percentage points below the Fed's 1972–2025 average of 79.4%. [1]
Capacity utilization relates an output index to a corresponding capacity index. The Fed describes capacity as sustainable maximum output under a realistic work schedule, allowing for normal downtime and sufficient inputs to operate existing capital. It is not a count of workers who are busy, a percentage of machines switched on at a particular moment or a measure of whether every plant can produce the same product. [3][6]
Nor is 100% a normal target against which every lower reading represents failure. Industries need maintenance, face different demand patterns and have different operating characteristics. The Fed's explanatory notes emphasize that broad utilization aggregates normally remain well below 100%. Historical comparisons should therefore use the relevant series and period, not an assumption that every unused percentage point can be converted immediately into profitable production.
Mining, factories and utilities have different operating rates
Manufacturing utilization edged up to 75.9%, still 2.3 percentage points below its long-run average. Mining stood at 87.5%, 2.3 points above its own average. Utilities rose to 73.8%, remaining substantially below their historical norm. These are three different positions relative to three different benchmarks. [1]
It would be misleading to infer that mining must be healthier than manufacturing simply because its utilization number is higher. The sectors have different capital structures and operating patterns. A comparison within each sector against its own history is more informative than treating their percentages as grades on a common test.
Capacity itself can change, too. The release reports year-over-year capacity growth of 1.2% for total industry, 1.0% for manufacturing and 3.1% for utilities, while mining capacity was down 0.1%. Utilization depends on both output and capacity, so it can remain subdued even while production grows if potential output also expands. Conversely, a higher operating rate can partly reflect a smaller capacity base. The numerator and denominator both matter.
An index level is not a growth rate
September's total industrial-production index was 104.0, with 2017 set to 100. That means the measured level was about 4% above the base-year average, not that output rose 104% in September. The reported monthly increase was 0.7%, and the twelve-month increase was 2.2%. They are three different ways of describing the same series. [1][3]
Manufacturing's index was 98.4, compared with mining at 124.5 and utilities at 117.3. Those levels can illustrate changes relative to each series' base-year reference, but they are not comparable quantities of physical goods. A point in mining is not the same physical unit as a point in utilities. The indexes aggregate different outputs through a statistical framework rather than converting electricity, minerals and manufactured goods into a single literal product.
The published rounded index levels may not reproduce every reported percentage exactly. The Fed calculates percentage changes from unrounded indexes. For example, a reader using a one-decimal-place level can get a slightly different result from the official change. That is a rounding issue to check before treating a small arithmetic difference as evidence of an inconsistency. [4]
Why this estimate will be revised
The September figures are preliminary. The Fed's explanatory notes say a month's first production estimate is normally revised over the following five months as more source information arrives. The program combines physical-product information, such as quantities produced, with input measures such as production-worker hours when direct output data are not suitable or available. [4]
The current release already changes the recent baseline. August's total production gain is now 0.1%, compared with a previously published rounded estimate of no change. August manufacturing is now shown down 0.2%, rather than down 0.3%. Those revisions are part of the October 16 vintage and should be used consistently when discussing the new September estimate. [1]
This does not make the first reading useless. It makes it an estimate with a known updating process. Analysts who want to compare releases should preserve dates and distinguish what was known initially from what later information showed. A chart that silently combines an old August number with a new September number can introduce a mismatch that the official release itself has already resolved.
November brings a scheduled benchmark change
The Fed plans an annual revision on November 24, 2026. The revised indexes will use 2022 as the base year and incorporate additional annual benchmark data, including Census manufacturing data for 2023 and 2024. Capacity estimates will incorporate newer survey and other source information. Seasonal factors and some estimation methods may also change. [1]
Changing the base year is not itself an economic event. If an index is rescaled so that a different year equals 100, the number printed next to the same underlying period can change without a factory producing anything different that day. Benchmark revisions can also alter estimated growth history, which is a separate matter from the cosmetic rescaling of the reference level.
For practical comparisons, save the release date and the series definition. After the revision, use the revised history consistently rather than joining old-base and new-base index levels as though they belonged to the same scale. The next regular monthly release is scheduled for November 17; the annual revision follows a week later. Both dates are scheduled statistical releases, not forecasts of what the next production number will show. [7]
What September does, and does not, establish
September added evidence of industrial expansion: total output rose, manufacturing improved, and most major market groups increased. It also showed a less even picture underneath the headline, with utilities advancing much faster than factories, motor vehicles and parts declining and business equipment flat. The third-quarter factory growth rate remained modest beside the total industrial rate.
For a manufacturer, the relevant question is how its own industry and customers compare with those components, not whether the national headline is encouraging in isolation. For households, the report offers context about production rather than a direct reading of wages, job security or the cost of living. For investors, it is one input among many, not a standalone instruction to buy or sell a sector.
The clearest summary is that U.S. industry produced more in September, but the increase was not a broad, uniform factory acceleration. Follow the components, distinguish output from sales, and keep monthly changes separate from annualized quarterly rates. Those distinctions preserve the useful signal in a stronger headline without turning it into a claim the data cannot support.
A more complicated factory floor.
Comments
Post a Comment