Two purchasing managers’ indices cover the same US economy in the same month and regularly land on opposite sides of the line separating expansion from contraction. Neither is wrong. They survey different companies and treat the data differently, and knowing which is which resolves most of the confusion.
This article is for informational and educational purposes only. It is not financial advice, investment advice, or a recommendation to buy, sell, or hold any security, cryptocurrency, or financial product. Always verify data with official sources before making financial decisions.
Short answer: different panels, different methods
Short answer: The ISM indices come from the Institute for Supply Management and survey a panel weighted toward larger manufacturers, published on the first business day of the month for manufacturing and the third for services. The S&P Global PMI surveys a broader mix including more small and mid-sized firms, and publishes a flash estimate before month end followed by a final reading. Both use a 50 line to separate expansion from contraction; they frequently disagree about which side the economy is on.
| ISM | S&P Global | |
|---|---|---|
| Panel | Weighted toward large manufacturers | Broader size distribution |
| Manufacturing release | First business day of month | Flash mid-month, final at month start |
| Services release | Third business day | Flash mid-month, final early month |
| Seasonal adjustment | Applied to each subcomponent | Applied to each subcomponent |
| Headline construction | Weighted composite of five subindices | Weighted composite, different weights |
Sample composition — the main source of divergence
The ISM panel skews large. Big manufacturers have global supply chains, export exposure and the balance sheet capacity to weather downturns differently from smaller firms. The S&P Global panel includes proportionally more smaller companies, which are more sensitive to domestic demand and credit conditions.
When these two groups face different conditions, the surveys diverge — and that divergence is informative rather than a defect. A period in which large exporters struggle while domestically focused smaller firms hold up produces exactly the pattern of a weak ISM and a firmer S&P Global reading, and the gap tells you something the average would obscure.
The 50 line and what crossing it means
Both indices are diffusion indices. Respondents report whether conditions are better, the same, or worse than the prior month, and the index is constructed so that 50 means an even split.
This has an important implication that headline coverage routinely misses: the index measures breadth of change, not level of activity. A reading of 48 does not mean output is contracting by 2%. It means slightly more respondents reported deterioration than improvement. An economy can grow with a sub-50 reading if the firms reporting improvement are much larger than those reporting deterioration.
It also means the indices are more useful as direction-of-travel indicators than as magnitude estimates, and that month-to-month changes in the index carry more information than the level.
Timing: flash estimates vs the first-business-day release
S&P Global publishes flash estimates around the third week of the month based on a majority of responses, followed by a final reading. ISM publishes only a final figure, but publishes it before almost any other monthly indicator.
The result is a sequence: S&P Global flash arrives first for the current month, then ISM manufacturing on the first business day of the following month, then ISM services. Markets watch the ISM releases more closely, largely because the ISM manufacturing series has a much longer history and is embedded in economic models built decades ago.
Which one has tracked GDP more closely
The honest answer is that neither has a decisive advantage, and the ranking changes depending on the period tested and the specification used.
The ISM services index has generally tracked overall activity better than ISM manufacturing, simply because services are the larger share of the economy while manufacturing has historically received disproportionate attention. Studies comparing the two providers have found broadly comparable predictive content, with the composite of both frequently outperforming either alone — the usual result when two noisy measures of the same underlying variable are averaged.
The manufacturing indices in particular have a well-documented tendency to overstate weakness in an economy where manufacturing is a shrinking share of output. A manufacturing recession with services expansion has repeatedly produced sub-50 manufacturing readings without an economy-wide contraction.
Mini glossary
- Diffusion index. An index measuring the proportion of respondents reporting improvement, scaled so 50 represents no net change.
- Flash estimate. A preliminary reading based on a partial set of responses, published before the final figure.
- Subindices. Components such as new orders, production, employment, supplier deliveries and inventories, which are frequently more informative than the headline.
- Prices paid. A subindex tracking input cost pressure, watched as a leading inflation indicator.
What this article does not conclude
Neither index is authoritative. They are surveys of managerial sentiment with modest sample sizes, and both are revised as seasonal factors are updated. A single month from either carries limited information.
Both providers publish full methodology and subindex detail. The subindices — particularly new orders and prices paid — frequently contain more signal than the headline that gets reported.