A purchasing managers index (PMI) is a survey-based measure of changing business conditions, usually reported monthly. Purchasing and supply executives report whether measures such as output, orders or employment rose, stayed unchanged or fell relative to the previous month. For a quant researcher, it offers a timely activity signal while official output data are still arriving. The number summarises survey responses; it is not a percentage growth rate.
How responses become an index
Consider one fictional output question with 40% higher, 40% unchanged and 20% lower responses from equally weighted firms. This is an unadjusted diffusion item, not a reconstructed official headline PMI.
Let , and be the response shares expressed as numeric percentages, so . The scoring rule gives higher responses full weight, unchanged responses half weight and lower responses zero weight:
Fictional monthly example with equal firm weights and no seasonal adjustment. The score measures the balance of reported directions; it does not measure the size of output changes.
The same rule can be written as
Here the higher share exceeds the lower share by 20 percentage points, putting the index 10 points above 50. An unadjusted item above 50 has more higher than lower responses; below 50, the reverse holds. Exactly 50 can mean everyone is unchanged or equal higher and lower shares. It does not require every business to stand still.
A reading of 60 cannot establish 60% output growth. A tiny increase and a large increase both enter the higher category. Nor does a fall in an output diffusion index necessarily mean output fell: a reading that stays above 50 still signals expansion relative to the previous month, albeit at a weaker indicated pace. S&P Global explains the scoring and reference point.
Identify which PMI you are using
“PMI” is a family of surveys and series. Headline construction differs across publishers and sectors:
| Series | What is combined |
|---|---|
| S&P Global Manufacturing PMI | Five indices: new orders 30%, output 25%, employment 20%, suppliers’ delivery times 15% (inverted), and stocks of purchases 10%. |
| US ISM Manufacturing PMI | Equal weights on new orders, production, employment, supplier deliveries and inventories; its published methodology specifies the seasonal adjustments. |
| S&P Global Composite Output Index | Manufacturing output and services business activity, weighted by sector value added; it is distinct from the five-component manufacturing headline. |
The S&P Global manufacturing methodology, ISM calculation instructions and S&P Global output-index guide explain these distinctions.
Supplier deliveries deserve special care. Slower deliveries raise the delivery contribution to these manufacturing headlines. That can reflect strong demand, but a supply disruption can also lengthen deliveries while production suffers. Read output and new orders alongside the headline before calling the move stronger demand. This ambiguity is documented in S&P Global’s discussion of headline versus component signals.
Use the series as it was known
For nowcasting, record the publisher, jurisdiction, sector, exact index name, reference month, seasonal-adjustment basis and publication timestamp. Keep an S&P Global flash estimate separate from its later final reading. Retain historical vintages when seasonal factors change; ISM explicitly documents annual seasonal-factor updates. A backtest must use the reading actually available at its decision time, otherwise it can introduce look-ahead bias.
PMI points are neither real nor nominal currency amounts and are not annualised output growth. Mapping an output index into GDP growth requires a fitted relationship for the relevant economy and period, as the S&P Global output guide illustrates. The survey’s 50 reference point is not a universal GDP-growth or recession boundary.
Check the reference point
A different fictional unadjusted item has 25% higher, 50% unchanged and 25% lower responses. Does its score prove that every firm reported no change?
Answer. No. The score is , because higher and lower shares balance. Half the firms changed directionally. The same score would arise if all respondents were unchanged, so the headline alone cannot distinguish those response patterns.
Sources
- S&P Global — PMI data FAQ: survey questions, diffusion calculation, manufacturing weights and flash estimates.
- ISM — Seasonal Adjustment Factors: US manufacturing component calculation, equal aggregation and seasonal-factor revisions.
- S&P Global — How to interpret and use the PMI survey output index: output versus headline indices and GDP mapping.
- S&P Global — Headline PMI versus subindices: delivery inversion and supply-disruption ambiguity.