Empire State 7.6 vs 14.8 Est — Demand Stalls as Input Costs Hit a 38-Month High
Fundamentals · 2026-09-15
The Empire State general business conditions index fell 13.0 points to 7.6 in September, against a 14.8 consensus — roughly half what the street expected, and down from a 56-month high of 20.6. The composition is worse than the headline: new orders 2.0 (from 17.3) and shipments -3.2 (from 11.7), the first negative shipments read since March. What went the other way was cost and labor — prices paid 63.1, the highest since July 2022; expected prices paid 67.3, the highest since June 2022; average workweek 17.0, the highest since November 2021; employment 10.6, an eighth straight positive month. Inventories swung from -5.2 to 8.9, delivery times stayed long at 18.8, supply availability -11.9. Forward expectations softened broadly — future conditions 29.0 from 32.1, with 8 of 12 forward indices lower. 101 responses, collected 2-10 September.
What It Changes
- It is a miss on growth and an escalation on prices, and the second half is what reaches the Fed. Input costs at a 38-month high and six-month cost expectations at a 51-month high, printed the morning the FOMC sits down with a 25bp hike priced at 84% to 91%.
- The headline understates how weak demand was. New orders fell 15.3 points and shipments turned negative. The headline held above zero on delivery times, inventories and hours — supply-side and labor lines, not demand.
- This is the growth scare the market was watching for, arriving with the wrong companion. A clean growth scare would argue for the Fed to wait. A growth miss alongside a 38-month cost high argues for nothing comfortable at all.
- The survey closed before the week's two biggest price events. Responses were collected 2-10 September — before Friday's hot core CPI, before Monday's Saudi pipeline revision to five or six weeks, and before diesel's record. The cost readings are a floor, not a peak.
- It does not change the decision. One regional survey of 101 firms, released the day before a decision the market has already priced twice over, does not move a committee. It changes what the committee is looking at afterward.
Impact
- USD — Mixed, lean bullish — the growth miss is dollar-negative on its own, but the price internals push the same direction as the hike that is already priced, and the dollar has been trading the rate path rather than the activity data all week.
- This morning's Open put DXY at 99.61, up 0.22%, pressing the 99.80 confluence of the 38.2% retracement and the 100-day average, with hike odds repriced from 69% before Friday's CPI. A soft regional survey is a thin reason to unwind that.
- The honest caveat: any move on this print is small and short-lived. Empire is a 101-response regional survey on the eve of an FOMC decision, and positioning dominates.
- US Indices (ES / NQ / YM / RTY) — Bearish — an orders-and-shipments miss removes the growth leg without delivering the rate relief that would normally offset it.
- RTY carries it most directly. Small caps are the domestic-cyclical expression, and a New York factory sector where orders stalled and shipments went negative is exactly that exposure. ES and NQ were already down 0.3% to 0.4% into the print on pre-FOMC de-risking.
- The margin story is the equity story: 63.7% of firms reported higher input costs and 0.6% reported lower, while only 30.6% got higher selling prices through. That gap is a cost squeeze that shows up in fourth-quarter guidance, not in this week's tape.
- Gold (GC) — Mixed — a growth miss is normally a gold bid, but gold is trading as a rates instrument right now, and the price internals reinforce the hike rather than argue with it.
- The Open had gold at $4,322.00, down 0.69% and a third straight session lower, falling alongside a strengthening dollar with a live Middle East escalation on the tape. Nothing in this release breaks that mechanism.
Inside The Number
The general business conditions index came in at 7.6, down 13.0 points from August's 20.6. Consensus was 14.8, so the print landed at roughly half of what was expected — a miss of about 7 points on an index that routinely swings 15.
Take the headline on its own terms first, because it is less alarming than the drop implies. 7.6 is still positive, still the sixth consecutive positive month, and still above the survey's full-history average of 6.4 across 303 months since July 2001. August's 20.6 was the highest reading since December 2021 — a 56-month high — so this is a fall from an unusually good month, not a collapse into contraction. The release itself frames it as growth continuing "modestly."
The composition is where it stops being reassuring. New orders fell 15.3 points to 2.0. Shipments fell 14.9 points to -3.2, the first negative shipments reading since March. Averaged together, the two demand lines went from +14.5 in August to -0.6 in September. Demand did not decelerate; it stopped.
So what held the headline above zero? The lines that are not demand. Delivery times stayed long at 18.8, unfilled orders remained positive at 5.9, inventories swung 14.1 points from -5.2 to 8.9, and supply availability was still negative at -11.9 — worsening for another month. Longer lead times and rising stock read as activity in a diffusion index. They are equally consistent with a factory sector that cannot get parts. Only five months in 303 have combined a positive headline with negative shipments: August 2010, May and June 2013, February 2026 and now.
The price lines are the part that travels. Prices paid rose 4.5 points to 63.1, the highest since July 2022 and the 24th highest reading of 303 months — above the 92nd percentile of the survey's entire history. 63.7% of respondents reported paying more, and 0.6% reported paying less. Prices received rose 5.4 points to 28.1, so pass-through is picking up too, but the paid-minus-received spread sits at 35.0 points against a 12-month average of 26.9. Costs are rising faster than firms can pass them on.
Forward, it is worse. Expected prices paid jumped 9.6 points to 67.3, the highest since June 2022 — a 51-month high, and the single largest month-on-month move anywhere in this release. Expected prices received *fell* 6.4 points to 42.3. Firms are simultaneously raising their cost forecasts and lowering their pricing-power forecasts, which is a margin warning with a six-month fuse.
There is a historical pattern here, and it is not a comfortable one. Prices paid above 60 with new orders below 5 has happened 11 times in 303 months. Nine of those eleven fall in two clusters: May to August 2008 and February to July 2022. The other two are June 2026 and today. Weak demand with this much cost pressure is rare, and the survey's only precedents are the two worst inflation episodes of its lifetime. That is a small-sample observation about a regional survey, not a forecast — but it is the specific combination the Fed spent 2022 trying to break.
Labor, meanwhile, got stronger. The number of employees index rose to 10.6, an eighth consecutive positive month. The average workweek index rose 10.1 points to 17.0 — the highest since November 2021, 58 months ago, and the 13th highest reading in the survey's history. Firms with stalled orders lengthened hours anyway. That is not what a factory sector does when it expects demand to stay weak, and it lines up with this morning's ADP weekly brief, where private hiring accelerated to 16,250 a week. Two labor reads today, both up.
One timing point that changes how to weight all of it. Responses were collected 2-10 September. That window closed before Friday's core CPI print, before Monday's Reuters report that the Saudi East-West pipeline needs five to six weeks rather than days, and before diesel set a record on Monday. Every cost figure in this release predates the week's energy escalation. Whatever 63.1 says about input prices, it does not yet contain the part traders spent this week repricing.
Against the running narrative, this fits rather than breaks it. ISM Prices printed 71.1 in the latest reading, a 23rd straight month of rising input costs. Headline PCE is running 4.1% over six months against 3.7% over twelve. Warsh told Jackson Hole the Fed's "predominant focus right now should be on prices." A regional survey showing costs at a three-year high and orders at zero does not argue with any of that — it just adds the demand side of the squeeze.
The Internals
Current conditions, every index the survey publishes:
Index · September · August · Change · Read
General business conditions · 7.6 · 20.6 · -13.0 · Positive for a 6th month, but half of consensus
New orders · 2.0 · 17.3 · -15.3 · Demand essentially flat
Shipments · -3.2 · 11.7 · -14.9 · First negative since March
Unfilled orders · 5.9 · 15.5 · -9.6 · Backlogs still building, more slowly
Delivery times · 18.8 · 20.6 · -1.8 · Lead times still lengthening substantially
Inventories · 8.9 · -5.2 · +14.1 · Largest swing in the release
Prices paid · 63.1 · 58.6 · +4.5 · Highest since July 2022
Prices received · 28.1 · 22.7 · +5.4 · Pass-through accelerating, not keeping up
Number of employees · 10.6 · 9.3 · +1.3 · 8th straight positive month
Average workweek · 17.0 · 6.9 · +10.1 · Highest since November 2021
Supply availability · -11.9 · -13.4 · +1.5 · Still worsening, marginally less so
Expectations six months ahead:
Index · September · August · Change · Read
Future business conditions · 29.0 · 32.1 · -3.1 · Optimistic, and slightly less so
Future new orders · 25.3 · 37.1 · -11.8 · The largest forward downgrade
Future shipments · 25.5 · 33.7 · -8.2 · Demand expectations cut with it
Future unfilled orders · 5.9 · 19.6 · -13.7 · Backlog confidence collapsing
Future delivery times · 7.9 · 7.2 · +0.7 · Still expected to lengthen
Future inventories · 9.9 · 7.2 · +2.7 · Stock-building expected to continue
Future prices paid · 67.3 · 57.7 · +9.6 · Highest since June 2022
Future prices received · 42.3 · 48.7 · -6.4 · Pricing power expected to weaken
Future employees · 20.0 · 28.2 · -8.2 · Still growth, materially less of it
Future workweek · -4.0 · 1.0 · -5.0 · Hours expected to shorten
Future capital expenditures · 14.9 · 16.5 · -1.6 · Capex plans stay modest
Future supply availability · -8.9 · -9.3 · +0.4 · Expected to keep worsening
Where this month's standout readings sit in 303 months of history:
Reading · Value · Rank in history · Last time higher or lower
Prices paid · 63.1 · 24th highest of 303 · July 2022, at 65.7
Future prices paid · 67.3 · 16th highest of 303 · June 2022, at 69.2
Average workweek · 17.0 · 13th highest of 303 · November 2021, at 19.8
General business conditions · 7.6 · Above the 6.4 full-history average · August 2026, at 20.6
Shipments · -3.2 · Negative in 9 of the last 24 months · March 2026, at -6.9
Prior month, for scale · 20.6 · Highest since December 2021 · December 2021, at 34.4
The split that defines the release:
Group · Direction · Detail
Demand indices · All lower · New orders -15.3, shipments -14.9, unfilled orders -9.6
Cost indices · All higher · Prices paid +4.5, prices received +5.4, expected prices paid +9.6
Labor indices · All higher · Employment +1.3, average workweek +10.1
Current indices overall · 5 fell, 6 rose · The rises were cost, labor, inventories and supply
Forward indices overall · 8 fell, 4 rose · The rises were delivery times, inventories, costs and supply
What The Firms Actually Said
There were no special questions this month, so the response distributions are the granular layer — and they say something the index values do not.
The headline drop was firms moving to neutral, not to negative. In August, 43.8% reported better conditions and 23.1% reported worse. In September, 33.9% reported better and 26.3% reported worse — so the share seeing improvement fell nearly 10 points while the share seeing deterioration rose barely 3. The residual went to "no change," which climbed from 33.1% to 39.8%. That is a stall, not a downturn, and it is why the release describes activity as still increasing "modestly."
On input costs there is no disagreement at all. 63.7% of respondents paid more in September, up from 59.4%. 0.6% paid less. Across 101 firms, that is fewer than one company in the entire panel reporting a lower input bill. Looking six months out, 69.3% expect to pay more and 2.0% expect to pay less.
Pricing power is the asymmetry. 30.6% raised selling prices against 2.6% who cut them — real pass-through, and up from 26.6% in August. But more than twice as many firms are absorbing a cost increase as are passing one on.
Hours moved sharply and in one direction. 23.8% of firms lengthened the workweek against 6.8% who shortened it, from 17.7% and 10.8% a month ago. Supply availability stayed a problem: 15.8% reported worse availability against 4.0% reporting better.
The forward distribution is widening rather than darkening. More firms expect conditions to improve than last month — 50.5% against 47.0% — but the share expecting deterioration rose faster, from 14.9% to 21.5%. About half the panel still expects a better six months; the disagreement inside the panel is growing.
The New York Fed's own summary, from Richard Deitz, Economic Research Advisor: "On the heels of strong growth in August, New York State manufacturing activity continued to pick up modestly in September. Employment grew at a solid pace, while pricing pressures intensified." The release adds that supply availability "continued to worsen," that delivery times lengthened "substantially," and that firms "maintained an optimistic outlook for business activity."
Against This Morning's Open
- The Open called this the session's swing factor and set the exact bar this print had to clear. It wrote that Empire State was "the only live print before the Fed" at 14.8 against 20.6 prior, and that "a number that undershoots 14.8 meaningfully would add a growth-scare angle to a week that has so far been purely about inflation and rate path" — The Open.
- It undershot by about 7 points, so that conditional fired. The growth-scare angle is now on the table, and it is the first activity-side miss in a week that had been entirely about CPI and the rate path.
- Where the Open's framing needs an amendment: this is not a clean growth scare. The Open's conditional implicitly treated a miss as the dovish case. It arrived with input costs at a 38-month high and six-month cost expectations at a 51-month high, which points the other way.
- The Open was right that an upside surprise would be ignored. It is worth noting the downside surprise probably gets discounted too — one regional survey, 101 responses, the day before a decision priced at 84% to 91%.
What This Sets Up
- Next Empire State — Thursday 15 October, covering October, with responses collected in the first half of the month. That survey window will be the first to fully contain Friday's CPI, the Saudi pipeline outage and this week's Fed decision.
- Whether prices paid keeps climbing is the number to watch, not the headline. 63.1 was recorded before the week's energy escalation, so it is a floor. A move above 65.7 would be the highest since June 2022 and would make the 2008 and 2022 comparisons harder to wave away.
- Whether orders recover or shipments stay negative. Two consecutive months of negative shipments would take this from a stall to a contraction, and would make the delivery-times-and-inventories support under the headline look like exactly what it might be.
- Whether the hours surge survives. A workweek index at 17.0 with orders at 2.0 is an unstable pair. Either orders come back and the hours were a leading signal, or hours give back the 10 points they just gained.
- The immediate cross-check is Thursday's Philadelphia Fed survey, which covers the same month with a different panel and will show whether the New York cost spike is regional or general.
What Is This?
- What it is: The Empire State Manufacturing Survey is a monthly poll of senior executives at manufacturers across New York State, run by the Federal Reserve Bank of New York and published around the 15th of each month — the first regional manufacturing survey of the cycle and typically the earliest read on the month just started. This month drew 101 responses, collected 2-10 September. Each question asks whether a measure is higher, lower or unchanged versus the prior month, and the published index is the share answering higher minus the share answering lower, seasonally adjusted. The survey has run since July 2001, giving 303 monthly observations. It also asks the same questions about expectations six months ahead, producing a parallel set of forward indices.
- Why it matters: It lands first, so it frames how the market reads the regional surveys that follow — Philadelphia, Richmond, Dallas, Kansas City — and it arrives weeks before the ISM manufacturing print for the same month. Right now its price indices matter more than its headline: with the Fed explicitly focused on prices and ISM Prices showing rising input costs for 23 straight months, a survey that shows what firms are actually paying, and what they expect to pay, is a live input to the inflation debate rather than a growth curiosity.
- How to read it: Four things, and they all cut toward humility. It is a diffusion index, so zero is the dividing line between expansion and contraction and the number says how *widely* something is happening, not by how much — an index of 63.1 on prices paid does not mean prices rose 63.1%. It is one state and about 100 firms, so month-to-month swings of 10 to 15 points are ordinary and a single print is close to meaningless in isolation. The headline is its own question, not a weighted composite of the sub-indices, which is why it can stay positive while shipments go negative — so read the sub-indices rather than inferring them from the top line. And the collection window matters: responses close in the first third of the month, so anything that happened after roughly the 10th is simply not in these numbers.
Want the full explainer? Economic News Events, Explained breaks down this release and every other one we cover, in plain English.
_For informational purposes only. Not investment advice._