ADP Weekly Hiring 16,250 (from 12,250) — Nearly Double the July Trough, on FOMC Eve
Fundamentals · 2026-09-15
Private employers added an average of 16,250 jobs per week in the four weeks ending 29 August, up from a revised 12,250 — a 32.7% jump and the strongest reading since 4 July, which it matches exactly. The series has nearly doubled off its 8,250 trough the week ending 25 July, rising in four of the five weeks since, and the monthly equivalent is roughly 70,400 against about 50,900 in our last published brief. It is still 33.0% below the 24,250 at the top of today's table and 47.2% below the 30,750 peak of 6 June — a figure that rolled off the twelve-week window today, which makes every "down from the peak" comparison easier from here without anything improving. One revision: 22 August moved up 250, from 12,000 to 12,250; the other ten overlapping weeks are unchanged to the job. Preliminary, seasonally adjusted, two-week lag, no published consensus. Next release 22 September.
What It Changes
- It removes the last dovish objection on the morning the FOMC sits down. Hiring accelerating by a third in a week is not a labor market asking for patience, and the decision lands tomorrow at 14:00 ET with roughly 84% to 91% already priced for a 25bp hike.
- It is small, lagged and untradeable, and that is the honest framing. No consensus is published for this series, it carries a deliberate two-week lag, and it has never moved a market. It is information, not an event.
- The level still says frozen. At 16,250 a week the pace is 47.2% below where it was in early June. A near-doubling off a trough is arithmetic on a very low base, not a hiring recovery.
- The window rolled, and the optics improved for free. The 30,750 peak of 6 June left the table today. Nothing changed in the economy; the comparison just got kinder.
Impact
- USD — Slight bullish — a labor print that accelerates on the eve of a live decision supports the hawkish case, though almost nothing trades on this release.
- The dollar has already done the work: DXY at 99.61 this morning and closing on the 99.80 confluence, with hike odds repriced from 69% before Friday's CPI to 84-91% now.
- US Indices (ES / NQ / YM / RTY) — Mixed, lean bearish — better hiring is good growth news that arrives at the worst possible moment for a market hoping the committee finds a reason to wait.
- The divergence worth holding: private hiring is accelerating while consumer sentiment printed 47.8 on Friday, its second-lowest reading since 1952, with the expectations component down 11.1%. Households are describing an economy their employers are not.
- Gold (GC) — Slight bearish — one more datapoint removing the case for patience, and gold is still trading as a rates instrument rather than a haven.
Inside The Number
The NER Pulse reported 16,250 jobs per week for the four weeks ending 29 August, against 12,250 for the four weeks ending 22 August. That is a 32.7% increase in a single weekly reading, and the strongest number the series has printed since 4 July — which it matches to the job, at 16,250.
Read that second fact carefully, because it is the whole shape of the summer. The series went 16,250 on 4 July, fell for three straight weeks to a trough of 8,250 on 25 July, and has now climbed back to exactly where it started. Eight weeks, a round trip, no net progress.
The recovery off the trough is real on its own terms. From 8,250 the readings run 9,500, 11,750, 10,000, 12,250, 16,250 — up in four of five weeks, and 97.0% above the low. In monthly terms that is roughly 70,400 jobs a month, against about 50,900 in the last brief this desk published on the series, covering the four weeks to 8 August.
But the level is the part that has not changed. 16,250 is 33.0% below the 24,250 sitting at the top of today's table, and 47.2% below the 30,750 the series printed for the week ending 6 June. The pace is still less than half what it was a quarter ago.
And one structural point that matters more than it looks. Today's table shows twelve weeks, from 13 June to 29 August. Last week's showed twelve weeks from 6 June to 22 August. The 30,750 peak rolled off the window today. Anyone reading only this release now sees a series whose worst-to-best range is 8,250 to 24,250, rather than 8,250 to 30,750. Nothing in the labor market changed. The denominator did. Every week from here, the peak comparison gets easier for purely mechanical reasons, and a brief that quotes only what is on the page will drift toward a rosier read without noticing.
The revision is small and it is in the right place. Comparing today's table with the one published on 9 September, ten of the eleven overlapping weeks are identical to the job. The exception is the most recent one: 22 August moved from 12,000 to 12,250, up 250. That is exactly the behavior a two-week-lagged series should show — the newest week is the least complete, so it is the one that moves when more payroll records land. Last time this desk checked, all ten overlapping weeks were unrevised, which was itself worth noting. One week moving by 250 is not a signal; a pattern of upward revisions to the newest week would be, and it is now worth tracking.
On the reference period, the two ADP products still agree. The four-week window ending 15 August averaged 10,000 a week, or roughly 43,300 a month, and that is the window that brackets the August survey reference period — the pay period including the 12th. ADP's own monthly National Employment Report put August private payrolls at 38,000. Those are close. The Bureau of Labor Statistics put the same month at 127,000, roughly 89,000 higher, and that gap has not closed.
Today's window, ending 29 August, sits almost entirely after the reference period. It is not a read on August at all — it is the first read on where September started, and it says September started better than August finished.
Two things should temper that. The first is the benchmark revision published on 28 August, which found private employment through March overstated by 178,000, leaning the argument toward ADP's lower numbers rather than the BLS ones. The second is that a 4-week moving average with a two-week lag cannot tell you about the last fortnight, and the last fortnight contained a hot core CPI print, a repricing to a near-certain hike, and crude above $100.
The contrast with the rest of the data is the genuinely interesting part. Thursday's jobless claims came in at 206,000 with the four-week average turning down for the first time in five weeks — firing is not happening. This series says hiring picked up sharply. And on Friday, consumer sentiment printed 47.8, the second-lowest reading in 677 months, with the forward-looking expectations component down 11.1%. Employers are hiring a little faster and firing no more than before, while households report the bleakest outlook in seventy-four years bar one month. Both cannot be describing the same September.
The Internals
This week against the reference points that matter:
Comparison · Reading · Gap · Read
This week, four weeks to 29 August · 16,250 per week · Not applicable · The headline
Prior week, four weeks to 22 August · 12,250 per week · Up 32.7% · A large single-week jump
Trough, four weeks to 25 July · 8,250 per week · Up 97.0% · Nearly doubled in five weeks
Top of today's table, 13 June · 24,250 per week · Down 33.0% · Still a third below
Peak of 6 June, off the table today · 30,750 per week · Down 47.2% · Less than half the June pace
Matches 4 July exactly · 16,250 per week · Unchanged in eight weeks · A complete round trip
Monthly equivalents, at 4.33 weeks to the month:
Window · Per week · Monthly equivalent · Note
Four weeks to 29 August · 16,250 · About 70,400 · Today's release
Four weeks to 22 August · 12,250 · About 53,100 · Revised up from 12,000
Four weeks to 8 August · 11,750 · About 50,900 · The last brief this desk published
Four weeks to 15 August · 10,000 · About 43,300 · Brackets the August reference period
Trough, four weeks to 25 July · 8,250 · About 35,800 · The low
The two ADP products against the official one:
Measure · August 2026 · Source
NER Pulse, window bracketing the reference period · About 43,300 per month · This series, four weeks to 15 August
ADP National Employment Report, monthly · 38,000 · ADP's own monthly release
BLS private payrolls · 127,000 · Employment Situation, 4 September
Gap between ADP monthly and BLS · About 89,000 · Unresolved
Preliminary benchmark revision, private, through March · Overstated by 178,000 · BLS, 28 August
The Twelve-Week Table
Every week ADP publishes, with what it showed six days ago:
Week ending · Today's release · 9 September release · Change
29 August 2026 · 16,250 · Not yet published · New
22 August 2026 · 12,250 · 12,000 · Up 250
15 August 2026 · 10,000 · 10,000 · Unrevised
8 August 2026 · 11,750 · 11,750 · Unrevised
1 August 2026 · 9,500 · 9,500 · Unrevised
25 July 2026 · 8,250 · 8,250 · Unrevised
18 July 2026 · 11,000 · 11,000 · Unrevised
11 July 2026 · 14,500 · 14,500 · Unrevised
4 July 2026 · 16,250 · 16,250 · Unrevised
27 June 2026 · 19,750 · 19,750 · Unrevised
20 June 2026 · 21,000 · 21,000 · Unrevised
13 June 2026 · 24,250 · 24,250 · Unrevised
6 June 2026 · Rolled off the window · 30,750 · No longer shown
What that table says, in order of how much it matters:
Observation · Why
Ten of eleven overlapping weeks unrevised · The series is stable once a week is a fortnight old
Only the newest overlapping week moved · Exactly what a two-week lag should produce
The 6 June peak has rolled off · The visible range narrows from 8,250-30,750 to 8,250-24,250
The trough and the peak are both still in view · 8,250 on 25 July, 24,250 on 13 June
Four of the last five weeks rose · The direction is genuine, the level is not recovered
Against This Morning's Open
- This release is not on The Open's calendar, and the calendar makes a claim this release contradicts. It said "Empire State is the only live print before the Fed" — The Open. The NER Pulse publishes every Tuesday and published today.
- To be fair to the framing, the claim is right about what *moves* anything. This series has no consensus and no market history. But the calendar is meant to be a complete list of what prints, and a recurring Tuesday release should be on it.
- What The Open got right is the repricing this lands into. It put hike odds at 84% to 91% against 69% before Friday's CPI, and DXY at 99.61 closing on the 99.80 confluence. A labor print that accelerates does nothing to argue with that.
What This Sets Up
- Next NER Pulse — Tuesday, 22 September, covering the four weeks ending 5 September. That is the first window sitting entirely after the August reference period, and the first that can speak to September.
- Whether 16,250 holds or the round trip repeats. The series hit this exact level on 4 July and gave it all back within three weeks. A second reading above 16,000 would make this a trend rather than a bounce.
- Whether the newest week keeps revising up. 22 August moved up 250 today after ten weeks of nothing moving. One is noise. A run of upward revisions to the freshest week would mean the series is systematically understating hiring in real time.
- Whether the household and employer pictures reconcile. Hiring accelerating against the second-lowest consumer sentiment reading since 1952 is not a stable pair, and one of them is going to move.
What Is This?
- What it is: The NER Pulse, a weekly preliminary estimate drawn from ADP's payroll records and published as a high-frequency companion to the monthly ADP National Employment Report. It reports the average weekly change in US private employment across a rolling four-week window, seasonally adjusted, with a deliberate two-week lag so the underlying payroll data is more complete when it is published. ADP made it publicly available on a weekly cadence in late 2025. Today's release covers the four weeks ending 29 August and shows a twelve-week table.
- Why it matters: Hiring, not firing, is the soft half of this labor market, and this is the fastest read on it that exists. Claims tell you nobody is being let go; this series tells you whether anybody is being taken on. It also provides an independent check on the official payroll count, which the preliminary benchmark revision of 28 August marked down by 178,000 in the private sector through March.
- How to read it: Four cautions, and they all point the same way. It is a four-week moving average, so a single week's move is one-quarter new information and three-quarters carried forward. It has a two-week lag, so it cannot describe the last fortnight. No consensus is published, so there is no beat or miss — only the direction and the level. And the table is a rolling twelve-week window, so the high and low it displays change without any data changing: read the level against a fixed reference, not against whatever happens to be on the page.
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._