The simple version
One monthly survey of firms in a single state produces more than twenty separate lines, and the one the news quotes is a single question among them.
It is not a summary of the others. Nothing underneath it is averaged together to produce it.
So the quoted figure can drop hard in a month when several of the lines beneath it are rising. That is not an error in the data. It is what reporting one answer out of twenty-three looks like.
The numbers
- The survey is the Empire State Manufacturing Survey, run monthly by the Federal Reserve Bank of New York among manufacturers in New York State. The overview page labels the series a diffusion index, seasonally adjusted. Read on November 16, 2026.
- The figures below are the seasonally adjusted diffusion readings from the New York Fed's own data file, esms_seasonallyadjusted_diffusion.csv, which carries 23 series and 305 monthly rows running back to July 2001.
- November 2026 against October 2026. General business conditions, the quoted headline: 7.8 against 25.5. The release describes it as remaining positive but falling eighteen points.
- On this survey zero is the dividing line, not fifty. The release reports new orders at 3.0 as having increased and unfilled orders at 10.3 as having risen, while supply availability at minus 9.2 is reported as having worsened. A figure above zero means more firms reported an increase than a decrease, however small the figure looks.
- Number of employees: 18.1 against 10.3. Average workweek: 13.5 against 9.1. General business conditions six months ahead: 28.9 against 21.0. All three readings rose.
- New orders: 3.0 against 23.2. Shipments: 8.8 against 13.1. Unfilled orders: 10.3 against 14.0. Prices paid: 45.9 against 59.5. All four readings fell, and all four stayed above zero, so the release describes each of those things as still increasing. A reading falling and the thing it measures falling are two different statements.
- The release states that employment expanded at the fastest pace in more than four years. Checking that against every one of the 305 rows in the file rather than a sample: the last month the employment reading stood at or above 18.1 was June 2022.
- Survey responses for the November report were collected between November 2 and November 9. The report was published on November 16.
- For what a diffusion reading is and how a balance of answers is turned into one number, see our piece on builder sentiment. Note that the builder index pivots on fifty while this survey pivots on zero, so the dividing lines are not the same.
Why the quoted figure does not move with the ones beneath it
Each one comes from its own question put to the same firms. The broad one asks about conditions in general, and the rest ask about narrower things, one apiece.
Nothing in the arithmetic ties them together. A firm can answer the broad question one way and the narrow ones another way, and both answers are counted exactly as given.
That is the whole mechanism, and it is worth knowing because some other well-known surveys do work the other way. On those, the quoted figure is worked out from the ones beneath it, so the two cannot come apart.
The actual math on one month of readings
Take the month above and take one away from the other. The quoted figure went from 25.5 to 7.8, a fall of 17.7, which the release rounds to eighteen points.
Over the same month the employment line went from 10.3 to 18.1, a rise of 7.8. The working week went from 9.1 to 13.5, a rise of 4.4. The half-year outlook went from 21.0 to 28.9, a rise of 7.9.
Those four sums are ours. Every figure going into them is the survey's own, and so is the eighteen-point description we are checking them against.
What this means
When a survey figure moves a long way in a month, the useful next step is to open the file behind it rather than reason from the one number. The file is public and it is a spreadsheet.
And it is worth noting when the answers were collected. A reading published in the middle of a month can describe a window that closed a week earlier, before anything that happened in between.
This article explains how one regional survey is structured and reports one month of its readings. It is general education, not investment advice, and it is not a forecast of any future reading.
What this is NOT
This article explains that the Empire State Manufacturing Survey reports more than twenty separate readings rather than one composite, using the November 2026 report and the New York Fed's own seasonally adjusted diffusion data file as its example, both read on November 16, 2026. It is not a forecast of any future reading, not a claim about the direction of the national economy, and not a view on whether manufacturing in New York State is strong or weak: the readings are reported as the file gives them. It is not advice about any security, fund or sector, and it does not suggest acting on a survey figure. It is not a statement about any other regional survey or about national surveys, whose structures differ and which were not read here; where this article says some surveys compute their headline from the readings beneath it, that is a general statement about survey design, and no survey methodology other than this one was read. Ours are the four sums in the actual math section and the check of all 305 rows that located June 2022 as the last month employment stood at or above 18.1; every reading itself is the New York Fed's. One survey and one data file were read. This is not a political endorsement or criticism of anyone.
Sources
- Federal Reserve Bank of New York, Empire State Manufacturing Survey overview page, read November 16, 2026, for the November summary, the diffusion index label, and the November 2 to November 9 collection window.
- Federal Reserve Bank of New York, Empire State Manufacturing Survey seasonally adjusted diffusion data file, esms_seasonallyadjusted_diffusion.csv, read November 16, 2026, for every reading quoted here, the 23 series, the 305 monthly rows and the June 2022 employment comparison.
Receipts1 receipt: the source's own words, with a link to eachShowHide
Employment expanded at the fastest pace in more than four years, and the average workweek increased
Federal Reserve Bank of New York, Empire State Manufacturing Survey, November 2026 overviewRead Nov 16, 2026Open the source (opens in a new tab)
Found this useful?