From 68% to 48.4% (3 September 2026): Waller Said He Could Wait One Meeting — and Friday's Jobs Report Now Lands Into a Coin Flip

The August Employment Situation still lands at 8:30 a.m. Eastern on Friday 4 September 2026, and the Dow Jones consensus is still 53,000 jobs after July's outright loss of 23,000. What changed on Thursday is the thing the print will be measured against. Governor Christopher Waller said he would be inclined to support holding the federal funds target where it is if disinflation continues — and the market-implied probability of a September hike fell from the 68% that stood on 1 September to 48.4%, the two-year Treasury yield shed more than five basis points to 4.328%, and the dollar fell against every G10 peer. Friday's payroll number now arrives into a coin flip rather than into a market leaning hard one way, which changes which direction has room to move — and the reason a soft print still may not move it much is buried in the household survey, where the labour force has shrunk by 2.4 million people since December.

What actually happened on Thursday, a day before the print

The August payroll figure is still unpublished. The number it will be judged against is not.

At 8:30 a.m. Eastern on Thursday 3 September, in remarks prepared for a Reuters interview, Federal Reserve Governor Christopher Waller conceded that inflation remains "meaningfully above" the Committee's 2 percent goal but said recent trends "suggest we are finally seeing some signs of disinflation." Then the sentence that moved the curve: "If this continues in the data due over the next two weeks, I would be inclined to support holding the target for the federal funds rate at its current setting." He put it less formally a moment later. "I'm going to paraphrase John Lennon here: Give disinflation a chance. We can wait one meeting," Waller said. "What's the cost of waiting one meeting? Hiking 25 basis points, one meeting right now, is not going to bring the CPI down to 2%."

He supplied the arithmetic behind the view as well. The three-month rate of the Fed's preferred inflation gauge has fallen from 4.76% in February to 3.05% now — "a considerable improvement, and the speed of this downward trajectory is encouraging" — while the annual figures, headline 3.7% and core 3.3% in July, "are not the best guide for where inflation is today," with some of the excess sitting in nonmarket services prices that are estimated rather than observed. The remarks were reported by CNBC.

Read the caveat too, because it is why this is a repricing and not a reversal. "I judge that policy is currently only slightly restricting aggregate demand, and it may not take much acceleration in inflation to nudge me into supporting tighter policy," Waller said. "If there is evidence that progress toward 2% inflation reversed in August, a small adjustment in our stance would help ensure that it resumes." A governor leaning toward a hold, conditional on two inflation prints, has not ruled out a hike. He has named a trigger.

The market took the first half. CNBC put the CME FedWatch probability of a September hike at 50.4% shortly after the remarks and 48.4% later in the morning, against 63.2% on Wednesday and the 68% that stood on 1 September. By late morning the two-year Treasury note yield was more than five basis points lower at 4.328%, the ten-year down more than three at 4.756% and the thirty-year off more than two at 5.241%, per CNBC's bond report. The dollar fell against all of its G10 peers.

The hiring data softened and the price data did not

Two labour readings landed either side of that, and both pointed the same way.

ADP reported on 2 September that private employers added 38,000 jobs in August, below the Dow Jones consensus of 47,000 and down from an upwardly revised 46,000 in July — the smallest monthly gain since January. The composition is the part worth keeping. Education and health services added 45,000, leisure and hospitality 16,000 and construction 12,000; manufacturing lost 17,000, professional and business services 16,000, and natural resources and mining and trade, transportation and utilities shed 5,000 each. Firms employing 500 or more workers accounted for 34,000 of the 38,000 net gain, and companies with fewer than fifty employees for 3,000. Pay growth did not move: base pay up 3.0% year over year for job-stayers and 3.2% for all private-sector workers, both unchanged from July, as CNBC reported.

The ISM Services report said the same thing about hiring while shouting the opposite about prices. The headline Services PMI registered 55.4% in August, up 1.3 points from July's 54.1% and above the Dow Jones forecast of 54.1% — a twenty-sixth consecutive month of expansion. Business Activity jumped 2.6 points to 61.7%, its highest since November 2022. New Orders rose 3.7 points to 60.9%, the highest since February 2023. And the Employment Index registered 47.8%: contraction for a second straight month, and below 50 in thirteen of the last eighteen months.

The Prices Index went to 72.6%, up 2.3 points on the month, above 70 for the fifth time in six months and the highest reading since August 2022. Its twelve-month average rose to 68.5%, the highest since April 2023. Of the commodities panellists reported on, exactly one was down in price — fuel — which was also reported up in price for a seventh consecutive month, both being true within a single month. Graphics processing units and steel joined the short-supply list, and tariffs and the Middle East conflict were the most-cited supply-chain issues.

Meanwhile the firing side stayed quiet. Initial jobless claims for the week ended 29 August came in at 206,000 against roughly 205,000 expected — a historically low level, published weekly by the Department of Labor.

And here is the tell. On the same morning that the front end rallied on the prospect of a hold, WTI futures for October delivery rose almost 1% to above $91 a barrel and Brent traded near $96, after Iran launched missile and drone strikes against Kuwait. Oil higher, a services prices index at a four-year high, and the two-year yield down five basis points anyway. That is what it looks like when a curve reprices on the reaction function rather than on the data — traders were not marking down inflation, they were marking down the Committee's willingness to answer it this month.

The dollar's fall was broad but not uniform, and one leg of it had nothing to do with Waller. The yen touched 156.15 per dollar in Thursday's session, its strongest since 3 August, on rising bets for a Bank of Japan hike at the 18 September meeting and renewed talk of intervention after Japan spent a record 15.4 trillion yen between 30 July and 26 August — CNBC reported that move hours before the Waller remarks landed. Two currencies can strengthen against the dollar in the same session for entirely unrelated reasons, which is the argument for reading the eight majors factor by factor rather than through an index. The yen's own machinery was traced in the note on the rate gap and intervention.

What drops on Friday, and what the consensus actually says

Two consensus numbers are circulating, and the gap between them is smaller than the gap between either and reality has been lately. The Dow Jones estimate is 53,000 jobs, reported by CNBC on 1 September. A survey compiled by Proactive and published on Yahoo Finance on 31 August put it at 58,000, with the unemployment rate holding at 4.1% and average hourly earnings up 0.3% on the month; inside that survey Deutsche Bank and UBS both looked for 65,000 and Wells Fargo for 80,000, with UBS the outlier on unemployment at 4.2%.

Event contracts are pricing something slightly gloomier. On Kalshi the contract asking whether the economy added more than 50,000 jobs traded at roughly even money, and the equivalent Polymarket contract at 48%. The interesting part is the shape rather than the centre: CNBC reported about one-in-four odds on a contract implying the United States lost jobs in August, and a similar probability on a print above 80,000. That is a market that has been burned. Both economists and event traders overestimated payrolls in each of the last two months.

The denominator: 2.4 million fewer workers since December

Here is the fact that makes this release different from the twelve before it. On BLS household-survey data the civilian labour force stood at 171.495 million in December 2025 and 169.094 million in July 2026. That is 2.401 million people out of the labour force in seven months, an average of about 343,000 a month. The participation rate fell from 62.4% to 61.4% over the same window, and the employment-population ratio from 59.7% to 58.9%.

The unemployment rate fell from 4.4% to 4.1% while that happened. Both statements are true at once, and they are not in tension — they are the same statement. Unemployment is a ratio, and the labour force is its denominator.

The two surveys disagree about the level, as they often do, and the disagreement is unusually wide right now. The establishment survey has total nonfarm payrolls up 426,000 between December and July, roughly 61,000 a month. The household survey has employment down 1.815 million over the same seven months. They are different samples answering different questions — payrolls count jobs at businesses, the household survey counts employed people — and a persistent gap of this size is itself information about how much confidence any single monthly figure deserves.

Why the FOMC wrote "job gains have kept pace with the workforce"

That clause is doing real work. In its 29 July statement the Committee decided to hold the target range for the federal funds rate at 3-1/2 to 3-3/4 percent, and described conditions this way: "Economic activity is expanding at a solid pace despite elevated uncertainty that owes, in part, to the conflict in the Middle East. Productivity growth and capital investment are strong. Job gains have kept pace with the workforce, and the unemployment rate has changed little. Inflation remains elevated relative to the Committee's 2 percent goal, in part reflecting supply shocks that have driven price increases in certain sectors, including energy."

Read the labour sentence as a benchmark rather than a description. If the standard is keeping pace with the workforce, and the workforce is contracting by roughly 343,000 a month, then a great many payroll outcomes clear the bar — including several that a headline writer would call weak. And the vote records where that leaves the balance of opinion: the hold passed 9-3, with Beth Hammack, Neel Kashkari and Lorie Logan all preferring to raise the target range by a quarter point at that meeting.

Chair Kevin Warsh then spent Jackson Hole on the other half of the mandate. "While this summer's PCE and CPI readings were better than expected, they do not tell me that underlying trends have meaningfully improved," he said, per CNBC's report of the 28 August speech, adding: "We must be confident that underlying inflation is moving to our objective, clearly and at sufficient speed. Otherwise, we have work to do." Hike odds moved from 35.4% to 57.5% in a single session on the strength of it, and to 68% by 1 September. The speech was previewed here in the Jackson Hole note.

Six days later a governor took the other side of that in public, reading the same summer prints as the beginning of a trend rather than noise. Neither statement is a forecast of the vote. Both together are evidence that the Committee arrives at 15-16 September genuinely split — which is the condition under which a single data release carries the most weight, and the one under which the market has the least idea what it is meant to be pricing.

What is already priced, in basis points rather than adjectives

The cleanest read on positioning is the Treasury's own par yield curve. Through 2 September it was unambiguous at the front: the three-month bill yielded 3.92% and the two-year note 4.39% — a forty-seven basis point gap in the wrong direction for an economy widely described as slowing, and about seventy-six basis points above the midpoint of the current target range. Curves do not price hikes by accident. Thursday took roughly six of those basis points back out of the two-year, which leaves the front end still priced for tightening, just less of it.

Source: US Treasury daily yield curve. Twenty basis points at the two-year in four sessions, thirteen at the ten-year, and then a flat 2 September before Thursday's reversal. The scale is worth holding on to: the move Waller triggered at the two-year is under a third of what the Jackson Hole repricing added, which is a reasonable description of the difference between a chair and a governor.

And the reason those yields rose in the first place was mostly not the labour market. WTI closed 1 September up 5.2% at $90.22 a barrel and Brent up 4.6% at $94.65 after US Central Command said American forces were striking Islamic Revolutionary Guard Corps targets in Iran; the Dow fell 419.02 points to 52,766.88, the S&P 500 0.71% to 7,631.47 and the Nasdaq Composite 1.03% to 26,099.77. An energy shock raises headline inflation and lowers real income at the same time, which is precisely the configuration in which a central bank's two objectives point in opposite directions — and the July statement said plainly which one it is answering: "The Committee will deliver price stability." That the same barrel was a dollar higher on Thursday while the curve rallied anyway is the measure of how much of the front end is now about the reaction function rather than the shock.

Three scenarios, and the channel each one runs through

None of these is a prediction, and none carries a level. The point of a scenario map is to know in advance which mechanism you are watching — and Thursday changed the map without changing the release, because a market at 48.4% has room to move in both directions where a market at 68% mostly had room to move in one.

A beat above 80,000 — roughly one-in-four on Kalshi pricing — is the cleanest case, and it now has more to work with than it did on Wednesday. It corroborates the Chair's characterisation of the labour market as broadly consistent with full employment, removes the strongest argument for waiting, and would do so into a market that has just given up thirteen points of hike probability. Expect the work to be done at the front of the curve: the two-year has room to price a full quarter point and more, and that is the leg of the curve carry actually claims. Equity duration is the transmission on the other side, which is why an index weighted toward long-dated cash flows tends to wear a front-end repricing worse than a broad one does.

An in-line print of 40,000 to 70,000 does very little on its own, and it now hands the decision somewhere specific rather than nowhere. Waller named his condition out loud: the data due over the next two weeks. The only major inflation releases in that window are next week's consumer and producer price indexes, which means an in-line payroll figure effectively transfers the whole September question to CPI. In that case the number worth reading on Friday is not payrolls at all but average hourly earnings, and after that the participation rate — because a stabilising labour force would begin to rebuild the slack the last seven months have removed from the statistics.

A miss, or a second negative month — also around one-in-four — is where the usual reflex misleads, and Thursday made that more true rather than less. The textbook response is a lower two-year and a softer dollar, and that is a reasonable expectation for the first few minutes. But some of that trade has now been done in advance: the front end has already priced a governor leaning toward a hold, so a soft print is partly confirming a move that has happened rather than triggering a new one. The amount of easing the front end can price while Brent trades near $96 is limited in any case, and a weak payroll number arriving alongside a shrinking workforce and an ISM services prices index at 72.6% does not produce the disinflationary slack that would justify much more. Watch whether the two-year extends below 4.32% or bounces; the currency will follow that, not the headline.

For the mechanics of why the dollar responds to some yield moves and ignores others, the distinction between the front end and the long end was traced in detail through August's long-end selloff, and the factor-by-factor read on the currency itself sits on the USD page.

The revision problem: the first print is the least reliable number in the release

The July report cut May by 66,000 and June by 37,000 — a combined 103,000 — taking the three-month average payroll gain to roughly 20,000 a month against a prior-twelve-month average of 34,000. June's first print of 57,000, itself a shock at the time as covered when it landed, now stands at 20,000. July's −23,000 will be revised on Friday too, and so will June again.

This matters more than it usually does, because the policy question is currently being settled on the level of labour demand rather than its direction, and the level keeps moving after publication. Anyone reacting to the headline at 8:30 is reacting to the least-settled figure in the document. The same pattern ran through the July release, and the full account of that report and its revisions is in the July note.

What would change the picture

Five things now, in rough order of how much they would move the front end.

Next week's CPI and PPI first, which is new. Waller made his lean toward a hold explicitly conditional on the inflation data due before the meeting, and those two releases are the only major ones in the window. That is an unusually clean piece of information: a policymaker has published his own reaction function and named the trigger, so the market can price the releases rather than guess at the man. It also means Friday's payroll figure is no longer the last word on September — a fortnight ago it looked like it might be.

Average hourly earnings second. July's gain was two cents, from $37.60 to $37.62 — 0.05% on the month, an annual rate of 3.2%. Consensus wants 0.3% in August. Even if it arrives, the annual rate falls to roughly 3.1%, because August 2025 was itself a 0.41% month and the base effect does the arithmetic. A 0.5% monthly print, by contrast, would be the first genuine wage-side argument for tightening in months.

Participation third: a labour force that stops shrinking would begin restoring the slack that has quietly vanished from the household survey, and would make the unemployment rate an informative statistic again. Fourth, the breadth of hiring, since a payroll gain concentrated in one or two sectors reads very differently from a broad one — the broader measure of underemployment, U-6, stood at 7.9% in July against 8.4% in December, on the same shrinking base, and the ISM services employment index at 47.8% says the private hiring impulse has been narrow for months rather than for one. And fifth, oil, which is currently setting the inflation expectations that set the front end; the energy channel has been running the rates story for a fortnight, as the collapse in Hormuz transits showed.

The Committee's next move is not knowable from here, and this post has deliberately not guessed it. What Thursday established is narrower and more useful: the September question is no longer mostly a labour-market question. One governor has said in public what would make him wait, the market has moved thirteen points toward believing him, and the test he named arrives next week rather than tomorrow. Friday's payroll figure still matters — it is the last employment reading the Committee sees, and revisions to June and July come with it — but it now sets the backdrop to the inflation prints rather than the other way round. What is knowable is which number carries the information, and where to look at 8:31 a.m. More about how this site reads data releases is on the about page.

Educational macro context only — not investment advice.