How do PMI surveys predict GDP changes?

The short answer

PMI surveys are conducted monthly by the Institute for Supply Management in the US (ISM) and by S&P Global for most other major economies. They survey hundreds of purchasing managers about month-on-month changes in orders, production, employment, supplier deliveries, prices and inventories.

The headline number is a diffusion index: a reading of 50 means as many respondents reported improvement as deterioration; above 50 indicates net expansion, below 50 net contraction. Sustained readings well below 50 have historically been associated with recessions.

PMI predictive accuracy is regime-dependent. In the eurozone, the PMI-GDP correlation was around 76% pre-pandemic and dropped to 56% post-2020. In the US, the manufacturing PMI now covers only about 11% of GDP, limiting its representativeness.

→ New to macro indicators? Financial education hub

What the data shows

Research published by S&P Global Market Intelligence in October 2025 documents the post-pandemic erosion of PMI-GDP correlations across major economies.

The documented evolution (S&P Global, ECB Economic Bulletin, ISM, 1995-2026):

• Eurozone PMI-GDP correlation against first-estimate GDP: 76% pre-pandemic, 56% since 2020.
• Eurozone PMI-GDP correlation against latest-estimate GDP: 74% pre-pandemic, 53% since 2020.
• US PMI correlation post-pandemic improves to 68% if GDP coverage is restricted to PMI sectors (manufacturing plus private services).
• The ECB documents that PMI-based forecasting accuracy improved in 2022-2023 after deteriorating in 2020-2021, returning toward pre-pandemic norms.

The exception that nuances the picture: the manufacturing PMI has been below 50 for extended periods (2015-2016, 2022-2024) without triggering NBER recessions, illustrating that a goods-sector survey carries limited information about a services-dominated economy.

→ Dataset: ISM Manufacturing PMI

Why it happens — the macro mechanism

PMIs are useful because they capture forward-looking decisions before they show up in hard data — but their information value depends on how well purchasing managers’ decisions map to aggregate GDP.

Channel 1 — purchasing decisions precede production and consumption. When managers cut order volumes, the effects flow through to industrial production with a lag of several weeks. This timing makes PMIs leading rather than coincident.

Channel 2 — sentiment and expectations are embedded in survey responses. The angle that distinguishes PMIs from hard data: managers respond based on what they expect, not just what they observe. During shocks (COVID lockdowns, tariff announcements), sentiment-driven responses can amplify the actual underlying change in activity.

This was visible in 2020-2021: PMIs swung wildly while the underlying economy was shaped more by lockdown timing than by manager-driven decisions.

Channel 3 — services PMI now matters more than manufacturing PMI. ISM and S&P Global both publish services PMIs that cover the larger share of modern economies. Reading manufacturing PMI alone in a service-dominated economy systematically under-represents the broader cycle.

Synthesis by regime: in the pre-2020 regime, with stable trade flows and globalized supply chains, PMI-GDP correlation reached 74-76% in the eurozone and similar levels in the US; in the 2020-2022 lockdown-and-reopening regime, PMIs swung sharply in ways that GDP did not fully follow, breaking the correlation; in the 2023+ recovery regime, ECB research documents partial restoration of forecasting accuracy as the unusual COVID dynamics faded.

The PMI did not become less informative — the economy became less manufacturing-dominated, and the survey design did not catch up.

→ Framework: Economic cycle phases

What it means for different economic actors

Industrial firms use PMIs to anticipate demand changes and adjust inventory and capacity. The new orders sub-component is particularly valuable because it leads the headline by one to three months.

Investors read PMIs as one of several leading signals. Cross-validating manufacturing PMI with services PMI and the LEI provides more robustness than relying on the manufacturing headline alone — see leading indicators reliability.

Central banks use PMIs in nowcasting models. The ECB explicitly documents the predictive performance of its PMI-based short-term forecasting models in its Economic Bulletin, including their recent regime-dependent performance.

A common error is to assume a PMI below 50 always means recession. The 2015-2016 manufacturing weakness, the 2022-2024 manufacturing contraction, and the eurozone’s persistent sub-50 manufacturing PMI without broad recession all show that the threshold is not a binary recession trigger.

Practical observation

Go deeper

Related questions

• What is the ISM Manufacturing New Orders index signaling?
• Conference Board LEI false signals
• Leading economic indicators reliability
• Freight indexes for forecasting

Frequently asked questions

What is the difference between ISM PMI and S&P Global PMI?

ISM is the original US PMI, surveying hundreds of US purchasing managers. S&P Global publishes PMIs for most major economies including the US, with somewhat different coverage and methodology. The two US PMIs sometimes diverge slightly because of methodology differences. Most US analysts cite ISM as the reference; S&P Global is widely cited internationally.

Which PMI sub-component matters most for forecasting?

New orders is generally considered the most leading sub-component, as it captures demand before production materializes. Backlogs of orders signals capacity tightness when high. Supplier deliveries can be misleading: slow deliveries can reflect either strong demand (positive) or supply disruptions (negative), as illustrated during 2021-2022 supply chain disruptions.

Has the post-pandemic PMI weakness been recession-relevant?

In the manufacturing-only sense, yes — the goods sector did contract during 2022-2023. In the broader GDP sense, no — services-led GDP continued to grow. The angle that matters: PMI signals are now sector-specific rather than economy-wide indicators. Reading the manufacturing PMI as a recession signal in a services-dominated economy creates the kind of false alarms documented over 2022-2024.

Last updated — 15 May 2026