ISM Services and Composite Measures

The US economy is approximately 70% services. That single number explains why the ISM Services PMI (formally the ISM Services Index) often tells you more about economic direction than the manufacturing survey that gets most of the attention. When services contract, recession risk is elevated. When services expand even as manufacturing stumbles, the economy typically keeps growing. Understanding the services survey—and how composite measures combine both sectors—gives you a more complete read on where the economy is actually heading.

TL;DR: The ISM Services PMI covers 70% of US economic activity and is a stronger GDP signal than manufacturing alone. Composite PMIs merge both sectors into a single reading, giving you the most reliable macro snapshot available from survey data.

Why Services PMI Deserves More Attention Than Manufacturing

Manufacturing PMI has historical prestige (it dates to 1948, decades before the services survey launched in 1998). It also moves global trade narratives and commodity prices. But the structural shift in the US economy means services employment, services output, and services inflation now dominate GDP, payrolls, and the Fed's policy calculus.

Services sectors covered include:

• Healthcare
• Finance and insurance
• Retail trade
• Transportation and warehousing
• Professional and technical services
• Accommodation and food services
• Information technology

The point is: Manufacturing can contract for months (even quarters) without triggering recession—as long as services hold. The reverse is not true. When services break down, the broader economy follows.

The 2023–2024 period demonstrated this clearly. Manufacturing PMI stayed below 50 for much of that stretch, yet GDP growth remained positive because services PMI held firmly above 52. The economy grew despite manufacturing weakness because services carried the load (as it usually does in a modern services-dominant economy).

How the ISM Services Index Works (Structure and Components)

The ISM Services PMI uses a diffusion index methodology similar to the manufacturing survey. Each month, purchasing managers across service industries answer whether business conditions are better, the same, or worse than the prior month. The percentage reporting "better" plus half the percentage reporting "same" produces each component's diffusion reading.

The headline index is an equal-weighted average of four components:

The calculation: ISM Services PMI = (Business Activity × 0.25) + (New Orders × 0.25) + (Employment × 0.25) + (Supplier Deliveries × 0.25)

Why inventories are excluded: Unlike manufacturing, many service businesses do not hold physical inventory. A law firm, a hospital, and a consulting practice all produce output without warehousing goods. The headline formula reflects this structural difference (and it's a detail worth remembering when comparing services and manufacturing readings directly).

Why this matters: Because the four components are equally weighted, a sharp move in any single component shifts the headline by up to 25% of that move. A 4-point drop in employment alone would pull the headline down a full point—enough to change the narrative from "solid expansion" to "slowing momentum."

Key Thresholds (What the Numbers Actually Signal)

The 50-level dividing line between expansion and contraction applies to services just as it does to manufacturing. But the practical thresholds differ because services PMI runs structurally hotter than manufacturing.

Historical context that matters:

• Services PMI has spent less than 10% of its history below 50
• Long-run average: approximately 55 (compared to roughly 52 for manufacturing)
• 2008–2009 recession low: 37.8 (a reading that extreme signals genuine economic crisis)
• Pandemic low (April 2020): 41.8
• Typical revision size: ±0.5 to 1.0 points between flash and final readings (smaller than manufacturing revisions because services output is less volatile)

The point is: A services reading of 52 might sound fine in isolation, but relative to the long-run average of 55, it signals meaningful deceleration. Context against the historical average matters more than the raw distance from 50.

Business Activity vs. New Orders (The Leading-vs.-Coincident Split)

These two components deserve separate attention because they answer different questions.

Business Activity measures current output levels—what is happening right now across service firms. It is a coincident indicator (confirming what GDP data will eventually show).

New Orders measures incoming demand—what clients and customers are requesting. It is a leading indicator (previewing where business activity is headed in coming months).

The practical insight: The spread between new orders and business activity tells you about acceleration or deceleration before it shows up in the headline.

• Rising new orders with stable business activity → acceleration ahead. Firms will need to ramp up output to meet incoming demand.
• Falling new orders with elevated business activity → deceleration ahead. Current activity is running on backlog, but the pipeline is thinning.
• Both falling together → contraction risk is real. This is the signal that warrants defensive portfolio positioning.

Why this matters: By the time business activity itself drops, the deceleration is already underway. New orders give you a one-to-two-month head start on recognizing the shift.

Employment Component and What It Tells You About Payrolls

The services employment subindex correlates with service-sector payroll gains in the monthly Bureau of Labor Statistics employment report—but the correlation is imperfect and worth understanding clearly.

The caveat (and it's an important one): The employment subindex measures whether more firms are hiring than firing—not the magnitude of hiring. A reading of 53 could mean many firms are adding one employee each, or a few firms are adding hundreds. The diffusion index captures breadth, not depth.

The practical use: Pair the employment subindex with ADP private payroll data and weekly jobless claims. When all three point in the same direction, you have a high-confidence signal. When they diverge, the ISM employment reading alone is not sufficient to call the labor market direction.

Prices Paid (The Inflation Signal You Shouldn't Ignore)

The services prices paid subindex captures input cost pressures across labor-intensive sectors. This component is not part of the headline PMI calculation, but it may be the most important number in the entire report for investors focused on Fed policy.

Why it matters for inflation: Services inflation is structurally stickier than goods inflation. Goods prices can fall quickly when supply chains normalize or demand drops. Services prices—driven heavily by wages and rent—adjust slowly. Elevated services prices paid often precedes persistent core inflation readings by two to three months.

Threshold interpretation:

• Above 60: Significant cost pressures (inflationary, likely keeping the Fed hawkish)
• 50–60: Moderate increases (manageable, unlikely to shift policy)
• Below 50: Cost pressures easing (disinflationary signal)

The point is: If you're trying to anticipate Fed rate decisions, the services prices paid subindex deserves as much attention as CPI itself. The Bureau of Labor Statistics publishes the inflation data after the fact. The ISM prices paid subindex gives you a real-time read on where cost pressures are building.

Composite PMI (Combining Manufacturing and Services)

S&P Global publishes a Composite PMI that merges manufacturing and services into a single reading, weighted by each sector's share of GDP.

The calculation: Composite PMI = (Manufacturing PMI × Manufacturing GDP Share) + (Services PMI × Services GDP Share)

For the US: Services weight is approximately 70%, manufacturing 30%. This means services PMI dominates the composite reading—a 2-point move in services shifts the composite roughly 1.4 points, while a 2-point move in manufacturing shifts it only 0.6 points.

Why composites matter:

• Single number for overall economy: Instead of reconciling two surveys, you get one GDP-correlated signal
• Better GDP correlation than either sector alone: Bureau of Economic Analysis GDP estimates align more closely with composite PMI than with manufacturing or services individually
• Cross-country comparisons: Eurozone, UK, Japan, and other major economies publish composites using the same methodology, making global growth comparisons straightforward

The caveat on cross-country comparisons: Sector weights differ by country. Germany's manufacturing share is closer to 40% (versus 30% for the US), so a German composite reacts more strongly to manufacturing swings. Comparing headline composites directly without adjusting for sector composition can mislead you.

Worked Example: Reading the October 2024 Services Report

Here's how to process an actual release systematically.

The data:

• Business Activity: 57.2
• New Orders: 57.4
• Employment: 53.0
• Supplier Deliveries: 56.4
• Prices Paid: 58.1 (not in headline, but critical)

Headline calculation: (57.2 × 0.25) + (57.4 × 0.25) + (53.0 × 0.25) + (56.4 × 0.25) = 56.0

Interpretation, component by component:

1. Business Activity at 57.2: Well above 55—current output is strong. No signs of services-sector stress.
2. New Orders at 57.4: Slightly above business activity. The pipeline is healthy—no deceleration signal. If anything, mild acceleration.
3. Employment at 53.0: Positive but notably weaker than the other components. Firms are hiring (more reporting increases than decreases), but not aggressively. This suggests payroll gains will be moderate, not strong.
4. Supplier Deliveries at 56.4: Above 50 means slower deliveries (suppliers are busier). This is consistent with healthy demand but also hints at capacity constraints.
5. Prices Paid at 58.1: Below 60 but firmly above 50. Cost pressures are present but not alarming. Worth monitoring for trend direction over the next two months.

The synthesis: Solid expansion with no immediate recession risk. The employment component is the weakest link—if it drifts toward 50, that would signal labor market softening worth watching. Prices paid are elevated enough to keep the Fed cautious about cutting rates aggressively.

The revision context: Flash-to-final revisions for services PMI typically run ±0.5 to 1.0 points. A headline of 56.0 could finalize anywhere from 55.0 to 57.0 without changing the fundamental interpretation (solid expansion either way).

Manufacturing vs. Services Divergence (The Pattern That Matters Most)

When the two surveys disagree, the divergence itself is the signal.

The core principle: The scariest combination is manufacturing expanding while services contract. It's rare (services almost never contract while manufacturing grows), but when it happens, it typically means consumer and business spending are pulling back simultaneously across the dominant sector of the economy. The 2001 recession saw a version of this pattern.

The more common divergence is manufacturing contracting while services expand. This happened through much of 2023–2024 and did not produce recession. The economy grew because services-sector employment, output, and spending more than offset manufacturing weakness.

Common Pitfalls (And How to Avoid Them)

Focusing only on manufacturing PMI. The manufacturing survey gets released two days before services. Media coverage peaks on manufacturing day. By the time services data arrives, attention has moved on. Fix: Treat the services release as the more important data point for US economic direction.

Ignoring component details. A headline of 54 could mask new orders at 48 (deceleration coming) or employment at 58 (strong hiring). The headline is a summary, not the full story. Fix: Always check new orders and employment individually.

Comparing US and Eurozone composites directly. Different sector weights mean the same headline number implies different underlying conditions. Fix: Compare sector-level readings, not just headline composites.

Treating single-month readings as trends. PMI data is noisy month-to-month. A drop from 56 to 53 could be statistical noise. Fix: Use three-month moving averages for trend identification. Only act on sustained directional moves.

Ignoring prices paid. The headline doesn't include it, so many investors skip it. But for rate-sensitive portfolios (bonds, REITs, growth stocks), the prices paid subindex is arguably the most actionable number in the report. Fix: Add prices paid to your release-day checklist.

Release-Day Checklist (Tiered)

Essential (do these every month)

These steps prevent the most common misreads:

• Note consensus expectation for the headline before release (third business day of month)
• Compare to manufacturing PMI released earlier in the week—check for divergence
• Read new orders and employment subindexes individually (do not rely on headline alone)
• Check prices paid for inflation and Fed policy implications

High-Impact (for active macro investors)

For investors making allocation decisions around data releases:

• Calculate the manufacturing-to-services spread and compare to the three-month average
• Cross-reference employment subindex with weekly jobless claims and ADP data
• Review ISM's industry-level commentary for sector-specific signals (healthcare resilience, retail demand, finance credit conditions)
• Track the composite PMI trend alongside Bureau of Economic Analysis GDP nowcasts

Optional (for deep macro analysis)

If you're building a systematic macro framework:

• Monitor supplier deliveries for capacity constraint signals
• Compare US composite PMI to Eurozone and UK composites for global growth context
• Track the new orders minus business activity spread as an acceleration/deceleration indicator

Next Step

For the next quarter, track both ISM Manufacturing and ISM Services PMI side by side on release days. Calculate the spread between them each month. When the spread widens significantly (services stronger than manufacturing by 5+ points), investigate which goods-producing sectors are weak and whether services-sector strength is broad-based or concentrated in a few industries. That spread, tracked over time, gives you one of the simplest and most reliable signals for whether the economy is in a rotation (sector-specific weakness) or a genuine slowdown (broad deterioration).