Production responses, not delivery delays, drove most of July's PMI jump

By Cumulant Research

[ISM's July Manufacturing PMIManufacturing PMIThe Manufacturing Purchasing Managers' Index is an [ISM survey measure that equally weights new orders, production, employment, supplier deliveries, and inventories](https://www.ismworld.org/supply-management-news-and-reports/reports/ism-pmi-reports/pmi/july/). rose 2.3 points to 55.6, its highest reading since May 2022](https://www.ismworld.org/supply-management-news-and-reports/reports/ism-pmi-reports/pmi/july/). Because the headline combines five equally weighted components, the published changes allow a reconstruction of which survey responses supplied the increase.

The arithmetic does not support the narrow claim that delivery delays generated most of July's jump. It also does not establish that factories produced substantially more goods. ISM measures the breadth of reported change, while the [Federal Reserve's manufacturing index estimates real output](https://fred.stlouisfed.org/series/IPMAN).

[ISM equally weights new orders, production, employment, supplier deliveries, and inventories](https://www.ismworld.org/supply-management-news-and-reports/reports/ism-pmi-reports/pmi/july/). Multiplying each published June-to-July change by 0.20 produces contributions of 0.14, 1.26, 0.62, 0.30, and -0.04 point, respectively.

Those five contributions sum to 2.28 points, while the reported headline rose 2.3 points. The 0.02-point gap is consistent with the fact that ISM publishes the component readings and changes to one decimal place, but the exact source of the residual cannot be proved without unrounded values.

Think of the survey as a vote count, not a warehouse scale. A diffusion indexdiffusion indexA diffusion index measures how widespread change is by combining the share reporting improvement with half the share reporting no change, rather than counting how many goods were produced. can show that improvement was reported by a wider share of factories without showing whether each factory added one unit or one million units. The calculation can identify which survey category moved the headline, but it cannot measure the size of the physical-output change.

[ISM asks respondents whether conditions changed from the previous month and constructs diffusion indexes from the shares reporting positive, unchanged, or negative conditions](https://www.ismworld.org/supply-management-news-and-reports/reports/ism-pmi-reports/pmi/july/). Its panel is organized by industry and weighted according to industries' contributions to the economy.

[Twelve of the 18 manufacturing industries reported higher production in July, six reported no change, and none reported lower production](https://www.ismworld.org/supply-management-news-and-reports/reports/ism-pmi-reports/pmi/july/). Fifteen industries reported overall growth, and four of the six largest manufacturing industries expanded.

That breadth makes a production-led survey move more plausible than a result confined to one disrupted supply route. It remains descriptive evidence, however. The survey does not identify how much extra output each responding company produced, and the calculation cannot establish that reported production caused the headline increase through any mechanism beyond the PMI formula.

[The Supplier Deliveries IndexSupplier Deliveries IndexThe Supplier Deliveries Index rises when deliveries become slower, so congestion can lift the PMI even if production volume does not increase. rose from 57.4 to 58.9, and readings above 50 mean deliveries were slower](https://www.ismworld.org/supply-management-news-and-reports/reports/ism-pmi-reports/pmi/july/). Thirteen industries reported slower deliveries, while none reported faster deliveries.

The correct conclusion is therefore not that supply pressure disappeared. Slower deliveries added 0.30 point to the reconstructed PMI increase. They were a contributor, but their contribution was less than one-quarter of production's 1.26-point contribution.

The remaining reconstructed movement came from employment, which contributed 0.62 point; new orders, which contributed 0.14 point; and inventories, which subtracted 0.04 point. These figures follow directly from [ISM's published changes and 20 percent weights](https://www.ismworld.org/supply-management-news-and-reports/reports/ism-pmi-reports/pmi/july/).

The latest available government data provide useful backdrop but not a July verdict. In the current data vintage, the [Federal Reserve's manufacturing-output index rose from 97.0786 in January to 98.6995 in June](https://fred.stlouisfed.org/data/IPMAN), a 1.67 percent increase. The [BLS aggregate-hours index rose from 91.0 to 91.4](https://fred.stlouisfed.org/data/CES3000000016), or 0.44 percent.

Those figures show that the first half already contained an output rebound. They cannot confirm that the pace accelerated in July because neither government series yet contains a July observation. The [Federal Reserve labels June industrial-production data preliminary](https://www.federalreserve.gov/releases/g17/current/table0.htm), and the [BLS marks the latest hours observations preliminary](https://www.bls.gov/news.release/empsit.t20.htm).

[BLS calculates aggregate hours from employment and average weekly hours](https://www.bls.gov/news.release/empsit.t20.htm). That makes the measure useful for asking whether factories used more employee time, but it does not directly measure the goods produced during that time.

Output and hours can diverge because productivityproductivityProductivity is the amount of output produced per unit of labor input, so output can rise without a matching increase in hours., industry mix, measurement error, and later revisions can change the relationship. Between the second- and third-quarter averages of 2022, the current data show [manufacturing output falling 0.45 percent](https://fred.stlouisfed.org/data/IPMAN) while [aggregate manufacturing hours rose 0.61 percent](https://fred.stlouisfed.org/data/CES3000000016).

That episode is an illustration, not a backtest. It was selected after the fact and uses revised data, so it cannot tell us how often the PMI correctly predicts official output or what will happen in July 2026.

The first follow-up arrives with July aggregate hours on [August 7](https://www.bls.gov/ces/publications/news-release-schedule.htm). The first direct government estimate of July manufacturing output follows on [August 18](https://www.federalreserve.gov/Releases/G17/release_dates.htm).

[ISM is scheduled to publish its August survey on September 1](https://www.ismworld.org/supply-management-news-and-reports/reports/ism-pmi-reports/pmi/july/). August hours and output then arrive on September 4 and September 18, followed by September hours and output on October 2 and October 16.

One positive July output estimate would be encouraging but still preliminary. A stronger confirmation would combine rising output with persistence into August or September and revisions that do not erase the initial gain.

The best-supported answer is narrow. Production responses supplied most of the July PMI increase recoverable from ISM's rounded components, while slower deliveries supplied a much smaller share. July's headline was therefore not mainly a delivery-delay artifact.

That is a finding about the construction of a survey index, not yet a finding about physical factory output. The [August 18 manufacturing-output release](https://www.federalreserve.gov/Releases/G17/release_dates.htm) is the first scheduled official test of whether July's broader production responses coincided with a real increase in goods produced.