The ISM Services PMI reports monthly business conditions across surveyed US service industries. Its headline and components can alter expectations for growth, inflation and interest rates. This article explains the 50-point threshold, what the component indexes measure and how to place the release in an economic calendar.

What the ISM Services index actually is

ISM surveys purchasing and supply executives about their US operations and changes from the previous month. The coverage is broader than everyday use of “services”: it also includes industries such as construction and mining. The headline combines four specified components; it is not a direct measure of all service output or a fixed percentage of national GDP.

Older material uses Non-Manufacturing and NMI. ISM’s July 6, 2020 roundup announced the Services PMI name for publications beginning that August. The naming history does not make all old data vintages interchangeable: check methodology, seasonal revisions and the edition being compared.

Why services weigh more than manufacturing

Services account for a large part of US economic activity, so their business conditions matter alongside manufacturing. The ISM survey is one source of information about those conditions; its headline is not a direct measurement of services’ share of output or employment. Use official GDP and employment data when comparing sector sizes.

Services and manufacturing surveys can move in different directions because they cover different industries. Neither has a deciding vote on whether the whole economy is slowing. Read both alongside other activity data and the market’s expectations. The ISM manufacturing index provides the companion survey.

How to read the fifty-point line

A raw component diffusion index adds the positive share to half the unchanged share. Hypothetical Business Activity shares 30% higher, 48% unchanged and 22% lower give 54. That is not 4% sales growth and does not mean most firms increased activity: the higher share is 30%. The four-component headline is also not a count of all firms improving. Keep raw components, seasonal adjustments and the composite separate.

  • Above 50 generally indicates expansion in the covered sector. It does not establish a particular percentage growth rate or make 55 a universal threshold for healthy growth.
  • Below 50 generally indicates contraction in the covered sector, not an automatic recession designation for the whole economy.
  • At 50 use the conventional sector threshold. A 48–52 “neutral zone” is not the official definition used here; inspect components and comparisons rather than assigning an invented band.

Level, monthly change and surprise answer different questions. For a hypothetical headline 54.0, previous 53.1 and pre-release consensus 54.8, monthly change is +0.9 index points and surprise −0.8. The reading is above 50 and improves, yet misses the expectation. Save the consensus source and timestamp; none of these differences is a currency return.

A blank response does not mean stable activity

Our fictional survey has 100 valid Business Activity responses and 20 blank forms. Adding the blanks to “unchanged” invents information we never received.

Scroll sideways to see the full diagram

Blank forms and the diffusion indexValid responses are 36 higher, 40 unchanged and 24 lower. Raw DI is 56. Treating 20 blanks as unchanged produces an incorrect 55. Nonresponse does not identify a direction.NO RESPONSE ≠ UNCHANGEDModel: 100 valid responses + 20 blanks36 higherweight 140 unchangedweight ½24 lowerweight 020 blanks: direction unknown100 valid: (36 + ½ × 40) / 100 × 100DI = 56Error: add 20 blanks to “unchanged”(36 + ½ × 60) / 120 × 100 = 55

All numbers are fictional. The model equally weights only 100 valid responses, without seasonal adjustment or missing-data estimation. It does not describe ISM’s actual treatment of nonresponding firms. The result is a raw Business Activity DI, not the four-component Services PMI. ISM: Services PMI methodology confirms that Business Activity is a separate component, the report presents response directions and DI, and the headline combines four components.

Compare the calculation under our assumptions with a deliberately incorrect reassignment.

In your own survey or spreadsheet, record valid responses and missing observations separately. Do not code a blank as “unchanged” without a justified rule. For a real report, check the provider’s methodology instead of attributing this exercise’s procedure to it.

Four components worth watching

The full report includes additional indicators beyond the headline. Read each definition and distinguish its response direction, seasonal-adjustment status and role in the composite.

Business Activity, New Orders, Employment and Supplier Deliveries each receive 25% in Services PMI. Prices is separate. Under the published method, the first three headline inputs are seasonally adjusted; deliveries are not. Prices is also adjusted but does not enter the average.

Hypothetical calculation: factors and values are invented
ComponentRaw indexFactorInput after rounding
Business Activity54.01.02052.9
New Orders56.01.00056.0
Employment48.00.96050.0
Supplier Deliveries57.1No adjustment57.1

Divide each relevant raw input by its factor and round to one decimal place. The average (52.9 + 56.0 + 50.0 + 57.1)/4 is 54.0; averaging the raw inputs gives 53.775, rounded 53.8. The latter is a different calculation. Supplier Deliveries uses slower responses: hypothetical 30.2% slower, 53.8% unchanged and 16% faster give 57.1. A higher delivery index means slower deliveries, not a known demand increase.

When the report comes out

The usual schedule is the third business day of the next month at 10:00 Eastern Time, with exceptions in ISM’s actual calendar. An early-March report normally covers February. Convert the dated publication to your own zone rather than assuming a fixed local hour or adding two calendar days to manufacturing.

ISM ordinarily publishes manufacturing on the first business day and services on the third business day, with holiday exceptions listed in its calendar. These are business-day intervals rather than two calendar days. Check the actual date and local release time in your economic calendar, and allow for potentially wider spreads or abrupt prices around the event.

Which way it moves the dollar

Headline expansion does not establish employment expansion. In a hypothetical adjusted-input model, Business Activity 65, New Orders 55, Employment 40 and Supplier Deliveries 56 average 54. Swapping only activity to 45 and employment to 60 leaves the same 54. Employment sits on opposite sides of 50 while the composite is identical. The 20-point employment difference is not 20,000 payroll jobs, and the arithmetic does not prove firms substituted production for workers.

No fixed rule establishes that crossing 50 gives the strongest dollar signal or requires rate cuts. Separate the data, your policy hypothesis and the observed pair movement, including the other currency and simultaneous news. The how growth data move the dollar article explains another measure with different units; the fundamental analysis section in our library places surveys in a wider framework.

What now

  1. Flag the release time. Check the next ISM Services date in the official release calendar and convert its 10:00 a.m. Eastern release time to your local zone. Set a reminder before the publication; that time is the data release, not the New York stock-market opening.
  2. Save the expectation before publication. Record provider, timestamp, period and aggregation method. Then calculate the surprise separately from monthly change and the sector threshold. If no consensus is available, record the gap in your evidence rather than inventing a forecast after release.
  3. Check prices and employment. Read their levels and month-to-month changes in the full report. Compare the survey’s staffing measure with official payroll data for the same month, while noting the different coverage and publication dates.
  4. Compare both ISM surveys. Find manufacturing and services for the same reference month. Record where they agree and diverge, then compare them with other activity data; sector size alone is not a rule for assigning trading weight.