MacroDataHub

Reading the regional Fed factory surveys

Five Reserve Banks publish a manufacturing survey weeks before the national data. They are diffusion indexes centred on zero, cover different districts, and are routinely over-read as forecasts of the ISM.

At a glance

MeasureLatestPeriodPublished byFrequency
Empire State general business conditions7.6Sep 2026Federal Reserve Bank of New YorkMonthly
Philadelphia Fed general activity37.8Sep 2026Federal Reserve Bank of PhiladelphiaMonthly
Dallas Fed general business activity11.6Aug 2026Federal Reserve Bank of DallasMonthly
Kansas City Fed composite index14Sep 2026Federal Reserve Bank of Kansas CityMonthly
Richmond Fed manufacturing index−2Sep 2026Federal Reserve Bank of RichmondMonthly
ISM Manufacturing PMI (national)54.6Aug 2026Institute for Supply ManagementMonthly

How each one is built

Empire State general business conditions
Method: Diffusion index: % reporting an increase minus % reporting a decrease
Covers: New York State manufacturers, about 200 responses
Released: Around the 15th, first of the month
Philadelphia Fed general activity
Method: Diffusion index, the longest-running of the five (from 1968)
Covers: Third District: eastern Pennsylvania, southern New Jersey, Delaware
Released: Third Thursday
Dallas Fed general business activity
Method: Diffusion index; the district is heavily weighted to energy-linked manufacturing
Covers: Texas
Released: Last Monday
Kansas City Fed composite index
Method: Composite of production, new orders, employment, supplier delivery time and inventories
Covers: Tenth District: Colorado, Kansas, Nebraska, Oklahoma, Wyoming and parts of Missouri and New Mexico
Released: Fourth Thursday
Richmond Fed manufacturing index
Method: Composite of shipments, new orders and employment
Covers: Fifth District: Maryland, Virginia, the Carolinas, most of West Virginia
Released: Fourth Tuesday
ISM Manufacturing PMI (national)
Method: Diffusion index centred on 50, not zero
Covers: National, about 400 purchasing managers
Released: First business day of the following month

Zero, not fifty

The five regional surveys are centred on ZERO: a reading of +5 means slightly more firms grew than shrank. The ISM PMI is centred on FIFTY. Comparing a Philadelphia reading of 10 with an ISM of 48 without adjusting for that is the single most common mistake made with these series.

They also measure breadth, not size. A diffusion index counts firms, so one very large plant cutting output heavily moves industrial production but barely moves the survey. That is why the regional indexes can look far weaker than actual factory output in a given month.

Which one to use

For an early national read, the Empire State survey lands first, around the 15th of the month it describes. Philadelphia follows and has the longest history, which makes it the best of the five for comparisons across cycles.

Do not treat any single district as a national proxy. Dallas is dominated by energy-linked manufacturing and Kansas City by agriculture and equipment, so each can diverge from the country for reasons that have nothing to do with the cycle. Reading all five together is far more informative than reading any one.

For the national picture itself, the ISM PMI remains the reference, and industrial production is the measure of actual output rather than of sentiment about it.

Frequently asked questions

What does a diffusion index of zero mean?
It means the share of firms reporting an increase equals the share reporting a decrease. Above zero, more firms expanded than contracted; below zero, the reverse.
Do the regional Fed surveys predict the ISM?
Only loosely. Each covers one district with its own industry mix, and the relationship with the national ISM varies by period, so a single strong or weak district reading is a poor predictor.
Which regional survey is released first?
The Empire State survey from the New York Fed, published around the 15th of the month it covers — about two and a half weeks before the ISM.
Why is the ISM centred on 50 and the others on zero?
The ISM converts responses to a 0–100 scale where 50 means no change, while the Reserve Banks report the raw balance of positive minus negative responses.
MacroDataHub on Facebook