Akshay DineshA factory receives a large order for aircraft, ships a batch of machinery, and ends the month with shelves full of parts. All three events describe manufacturing, but they do not describe the same moment. One points toward possible future work, another records completed sales, and the third shows resources waiting somewhere in the production process. Treating them as interchangeable can make a healthy factory look weak, or a temporary surge look like a lasting boom.
The U.S. Census Bureau brings these movements together in the monthly Manufacturers’ Shipments, Inventories, and Orders survey, usually called the M3 or factory orders report. Its headline is often the percentage change in new orders, yet the report becomes far more useful when orders are read beside shipments, unfilled orders, and inventories. The September 2026 release offers a clear example: total new orders rose 1.8 percent, but that single number cannot tell us how much factories produced, how quickly customers received goods, or whether higher dollar values partly reflected higher prices.
What the factory orders report actually measures
The M3 survey has been conducted monthly since 1957. The Census Bureau uses reports from thousands of large-company units to estimate month-to-month changes across manufacturing industries. In the September 2026 methodology notes, the agency described a panel of about 4,700 reporting units representing roughly 3,000 companies. Those units covered 92 industry categories that were combined into 65 publication levels.
New orders are commitments to buy products for immediate or future delivery, minus cancellations. They are supported by documents such as contracts, letters of intent, or letters of award. Orders are not collected for industries in which products are normally delivered as soon as they are made, because an order series would add little information there. A new order therefore signals demand and a possible future production commitment; it does not prove that the item has been built.
Shipments record the value of products sold by manufacturers. Unfilled orders are the backlog remaining at the end of the month. Inventories cover materials and supplies, work in process, and finished goods held at month-end. Read together, these measures describe a flow: customers place orders, factories use inputs and perform work, finished products are shipped, and anything not yet completed or delivered remains in inventory or backlog.
Orders, shipments, and backlog occupy different points in time
Imagine a manufacturer begins the month with $300 million in unfilled orders. It receives $100 million in new orders, customers cancel $5 million, and the company records $70 million in sales. In a simplified accounting picture, the ending backlog would be $325 million: the starting backlog plus net new orders, minus sales. The factory may have been busy even if new orders fell, because it could still be working through commitments placed months earlier.
The reverse is also possible. A burst of new orders may lift the headline while shipments barely move because production takes time. Aircraft, turbines, ships, and complex machinery can remain in a backlog for months or years. A rising backlog can suggest future work, but it can also reflect delays, capacity constraints, parts shortages, or a concentration of long-cycle products. Context decides whether it is encouraging, concerning, or simply normal for the industry.
Ratios help connect these quantities. The unfilled-orders-to-shipments ratio compares the backlog with the current pace of shipments, while the inventories-to-shipments ratio compares stocks with that same pace. If the first ratio rises, the backlog is growing relative to shipments. If the second rises, inventory is accumulating faster than goods are leaving factories. Neither change has one automatic meaning: businesses may be preparing for stronger demand, or they may be holding goods that are moving more slowly than expected.
Why the headline can mislead
Transportation equipment can swing the total sharply because airplanes, vehicles, and defense products are expensive and are often ordered in large batches. That is why the report publishes totals excluding transportation and excluding defense. When the headline and the excluding-transportation measure move in different directions, a reader should ask whether one volatile industry is doing most of the work. The September 2026 full report illustrates the issue: transportation equipment accounted for $5.3 billion of the $6.5 billion increase in durable-goods orders.
Durable and nondurable goods also behave differently. Durable goods are intended to last, such as machinery, appliances, and vehicles; their orders can be postponed when businesses or households feel uncertain. Nondurable goods include products such as food, chemicals, paper, and petroleum products that are consumed more quickly. Some nondurable industries report shipments rather than a meaningful separate order pipeline, so the full factory orders report cannot be read as one perfectly uniform ledger.
The figures are seasonally adjusted, which helps remove recurring calendar patterns, but they are not adjusted for inflation. A higher dollar value can reflect more physical output, higher prices, or both. The report also excludes semiconductor manufacturing from its new- and unfilled-orders figures, a detail that matters when readers try to use the headline as a complete picture of high-technology demand. Factory orders are valuable evidence, but they are not a direct measure of real production, profits, employment, or the entire industrial economy.
A practical way to read each release
Start with the total change in new orders, then resist the temptation to stop there. A careful reading can be done in a short sequence:
- Check the revision. Compare the prior month’s revised figure with the number reported earlier. Revisions can change the direction or size of the apparent trend.
- Remove volatile categories. Compare the headline with totals excluding transportation and defense to see how broad the movement is.
- Compare orders with shipments. Orders moving faster than shipments may add to future work; shipments moving faster can draw down an existing backlog.
- Inspect unfilled orders. Look for persistent movement in the backlog, not just a one-month jump.
- Read inventories by stage. Materials, work in process, and finished goods point to different places where stocks are building.
- Use several months and other indicators. Industrial production, employment, prices, and business surveys can confirm or challenge the story suggested by factory orders.
This order of operations separates signal from spectacle. A one-month rise dominated by aircraft orders deserves less confidence than a broad increase across machinery, metals, electronics, and other industries. A revised series also deserves more weight than the first estimate alone. Economic reports are measurements under revision, not final scores.
What the September 2026 release shows
In September, seasonally adjusted new orders for manufactured goods rose $12.2 billion, or 1.8 percent, to $677.2 billion. Shipments rose 1.3 percent to $668.5 billion, while unfilled orders increased 0.9 percent to $1.6238 trillion. Inventories rose 0.4 percent to $976.0 billion. On the surface, orders, shipments, backlog, and inventory all moved upward.
The relationships add nuance. The unfilled-orders-to-shipments ratio slipped from 6.86 to 6.84 even though the dollar value of the backlog increased, because shipments were also rising. The inventories-to-shipments ratio fell from 1.47 to 1.46 while inventories grew, again because the shipment pace increased faster. Those falling ratios do not contradict the higher dollar totals; they show why denominators matter.
The release also revised August new orders from the earlier $663.5 billion and 0.1 percent increase to $665.0 billion and a 0.3 percent increase. That revision did not overturn the direction, but it changed the starting point for September. Meanwhile, petroleum and coal products led the rise in nondurable shipments and inventories, a reminder that current-dollar changes can be sensitive to prices as well as quantities. The responsible summary is not simply that factories were “booming.” It is that demand and shipments rose, backlogs remained elevated, inventories continued to build, and volatile industries played a meaningful role.
Read factory orders as a moving production pipeline
The most useful question is not whether the headline number is good or bad. Ask where activity is entering the pipeline, where it is leaving, and where it is accumulating. Orders point toward commitments, shipments record products sold, backlog preserves unfinished obligations, and inventories reveal goods at several stages of production. Ratios, revisions, industry detail, and price context keep those pieces in proportion.
That approach turns the factory orders report from a monthly percentage into a map of industrial timing. It cannot predict the economy by itself, and its panel is not a probability sample with measurable sampling error. Still, when its measures are read together, it shows something few single indicators can: how demand, production, delivery, and stocks are moving through the manufacturing system at the same time.
