2026-08-31 → 2026-09-04

The August ISM manufacturing survey is forecast at 55.2 against July's 55.6. July was the strongest reading since May 2022 and the seventh consecutive month of expansion, and the internals were better than the headline: production rose 6.3 points to 58.5, new orders edged up to 56.7, and the employment index crossed above 50 for the first time in 33 months, at 52.8. A consensus of 55.2 is not a call for deterioration. It is a call for that surge to hold.

The release matters more than a manufacturing survey usually does because of what the Fed chair said three days earlier. Warsh described himself as impressed with the economy's strength while insisting that underlying inflation has not improved — a framing that makes activity data a supporting argument for tightening rather than a constraint on it. ISM is the first clean read on the activity half of that sentence since the keynote, and the prices-paid component is the half that speaks to the other.

Above 55.6 would mean manufacturing accelerating into a Fed that has just signalled it may have work to do, and the front end would take it straight. Below 54 would be the first crack in the growth story, and it would matter far more for the 2-year than for the 10-year: it is the September question that is live, not the term premium. The employment sub-index is worth watching separately, since a second month above 50 there would raise the bar for Friday's payroll report to disappoint.

In isolation, one ISM print does not decide a meeting on 15-16 September, and it certainly does not move a chairman who has just told the market he does not pre-commit. Its function this week is positional: it sets how much the market is willing to lean on Friday, and after a 14bp single-session move in the 2-year, that leaning is where the risk sits.

Australia's June-quarter national accounts are forecast at 0.3% q/q, identical to the March quarter. That is the pace of an economy that is growing but not comfortably, and it arrives at the point in the cycle where the composition matters more than the headline.

The context is last Wednesday's July CPI, which put headline inflation at 3.5% year-on-year, down from 3.8%, but left the trimmed mean unchanged at 3.6% — still above the top of the RBA's 2-3% band on the measure the Bank actually watches. A board that has openly discussed tightening needs growth soft enough to open spare capacity and pull that underlying number down. Two consecutive quarters at 0.3% is not obviously that.

A print at or above 0.4%, with domestic demand rather than net exports doing the work, strengthens the hawkish case for the 29 September meeting: it says the economy is absorbing capacity while underlying inflation sits at 3.6%. Below 0.2% is the first genuine argument for patience, and would take tightening premium out of the front of the Australian curve and out of AUD. The middle case, a 0.3% built on an inventory drawdown offset by resource exports, is a number the RBA will read through rather than act on.

This is the last full national accounts release before the RBA decides, which gives a quarterly print the weight of a decision-point release. It is also, in a week dominated by the Fed, the sort of data that trades cleanly for two hours in the Asian session and is then overwritten by whatever happens in New York.

The RBNZ is expected to raise the Official Cash Rate to 2.75% from 2.50%, its second consecutive increase after July's hike. Consensus here is unusually tight, and the decision itself carries very little surprise premium.

The arithmetic behind it is stark. New Zealand's June-quarter CPI ran at 4.1% year-on-year, up from 3.1% in March, above the top of the 1-3% target band and the fastest rate since late 2023. Transport was the largest contributor as the energy shock fed through to petrol and fuel prices, with housing and utilities second. A central bank that was cutting a year ago is now tightening into an inflation overshoot it did not create.

Because the hike is so widely held, the risk is asymmetric. A hold would hit NZD hard and immediately, since nothing in the price allows for it. A hike as expected does almost nothing on its own — the tradeable content of this meeting sits in the accompanying track, not in the 25 basis points.

The wider significance is that New Zealand has become the clearest working example of this year's dominant macro theme: an energy-driven price impulse forcing central banks off an easing path and back into tightening. New Zealand is a small economy, but it is early in that sequence, and it is the template the market is now testing against the Fed, the Bank of England and the RBA.

The Monetary Policy Statement is the quarterly document that carries the RBNZ's published OCR track. When the decision itself is near-certain, the track is the entire event.

July's review left the market expecting further 25bp increments through the remainder of the year, with the debate among local forecasters centred on where the peak sits — a range that spans roughly 3% to 4% depending on the house. Where the new projection puts that peak, and how quickly it gets there, is what the Kiwi and the 2-year swap will actually trade on Wednesday morning.

A track that ratifies further hikes into 2027 and lifts the terminal rate is a straightforward hawkish repricing, and NZD should hold any gains. A track that delivers today's hike and then flattens — the dovish hike, tightening now to buy the option of stopping — would see the tightening premium come out quickly, because a great deal of it is already in the price.

There is an instructive contrast in the timing. This is one of the few statements in the world that still commits a central bank to a numerical path and holds it publicly accountable to that path — precisely the practice the new Fed chair spent Friday arguing against. Two banks on the same tightening side of the argument, with opposite philosophies about telling anyone. The week will offer a direct read on which approach produces more volatility.

The rate statement is the short-form record of the decision and the committee's reasoning, published alongside the fuller Monetary Policy Statement.

The line to read is how much of the 4.1% CPI overshoot the committee treats as an energy pass-through that will drop out of the annual rate, and how much it treats as a broadening of domestic price pressure. That distinction decides whether 2.75% is a waypoint or close to the destination, and it will be visible in the language long before it is visible in the forecasts.

Emphasis on medium-term uncertainty and the lingering effects of the supply shock reads as a committee tightening defensively and reluctantly, and caps the hawkish reaction. Emphasis on capacity pressure, wage settlements or inflation expectations reads as a committee that thinks it is behind the curve, and that is the version that extends a move in NZD rather than fading it.

Where the statement and the track conflict this week, the track will win the market's attention — a number beats an adjective. But the statement is where the committee's discomfort, if it has any, shows up in plain language, and discomfort is what tells you whether the published path is a plan or a placeholder.

The Governor's press conference, an hour after the release, is where the published track gets interpreted and occasionally undermined. On RBNZ days the Kiwi's largest intraday moves more often come in this hour than on the release itself.

The questions will press on the uncomfortable part of the decision: how a central bank tightening into an energy-driven inflation overshoot distinguishes that from demand-led pressure, and how much above-band inflation it is prepared to tolerate while the shock washes through. There is no clean answer, and the manner of the non-answer is the signal.

The familiar risk is a hawkish document walked back in the room — a Governor who stresses conditionality, data dependence and the possibility of pausing, against a track that says otherwise. The reverse is equally live: a neutral-looking projection delivered with conviction about the direction of travel. Either way, the tradeable reaction is usually the gap between the two, not the content of either.

The Bank of Canada is expected to hold the overnight rate at 2.25%, and with no Monetary Policy Report attached to this meeting, the statement carries the whole signal.

Friday's national accounts made the case for staying put considerably easier. Real GDP grew 3.3% annualised in the second quarter, comfortably above the Bank's own 2.5% call, led by exports, household spending and business investment; June was up 0.3% month-on-month, and the first quarter was revised up from flat to 0.1%. A bank that has been on hold since its last cut now has growth data that removes any residual case for easing.

The hold is not in question, so the variable is how the statement treats trade uncertainty against that growth beat. A statement that leans on the numbers and quietly drops any easing language is a hawkish hold, and it would firm the front of the Canadian curve. A statement that dwells on trade risk and on StatCan's flat July flash estimate keeps risk two-sided and does very little to CAD.

The strategic point is the policy spread. With the Fed newly hawkish, the Bank of Canada's willingness to sit still is what defines the gap, and the loonie trades that gap far more reliably than it trades Canadian data. A hawkish hold in Ottawa is worth less to CAD this week than a hawkish sentence from Washington is worth against it.

A sixth consecutive hold at 2.25% is both the consensus and the market price. The Bank cut to this level and stopped, and has now spent the better part of a year defending the decision to stop.

2.25% sits at the bottom of the 2.25-3.25% range the Bank treats as neutral. Holding there is neither stimulative nor restrictive; it is a central bank that has concluded its policy rate is roughly right and that the genuine uncertainty lies elsewhere — in trade policy, and in how much of the second quarter's strength was durable rather than pulled forward.

Any move on Wednesday would be a real shock, so the question worth asking is directional risk from here. After a 3.3% quarter with inflation running above target, the next change is more plausibly up than down. That is a reframing that has been building slowly rather than arriving suddenly, and it is not yet fully in the front end.

A fully priced hold is not tradeable on the day. What it does is concentrate everything onto Friday's employment report, which becomes the only Canadian data with the power to shift the Bank's stance before the October meeting.

The Governor's press conference follows 45 minutes after the statement, and it is where the hold has to be explained rather than merely asserted.

Two questions carry the content. First, whether the Bank frames the second quarter's 3.3% annualised growth as durable strength or as trade-distorted front-running that will reverse — StatCan's early estimate already has July flat. Second, how it talks about inflation running near the top of its comfort range through the second half of the year without conceding that the next move might be a hike.

An explicit statement that the bar for further easing is now high would be the hawkish outcome, and it is the one most consistent with what the data actually said on Friday. A Governor who stresses that the quarter flattered the underlying picture keeps two-sided risk alive and caps CAD, which is the more comfortable position for a bank that would rather not commit ahead of a US labour market print two days later.

Bailey speaks less than a fortnight before the 17 September MPC decision, with Bank Rate at 3.75% and a committee that split 6-3 in July — the three dissenters voting to raise, not to cut. That is an unusual configuration for a Governor to manage in public.

The backdrop is a Bank forecasting its own overshoot. UK CPI was 2.6% in June, and the Bank's central projection has inflation peaking near 3.2% in the fourth quarter with risks tilted to the upside. Bailey spent late August playing down second-round effects, which is the intellectual argument for sitting still while the headline rate climbs. It is coherent, and it is exactly what the three dissenters do not accept.

If he repeats that framing, the market reads a September hold as safe and the front of the gilt curve gets some relief. If he acknowledges the minority's case — that a 3.2% peak with an upside skew is not something a central bank simply looks through — the 2-year gilt reprices and sterling firms into the meeting.

Note the timing: 10:50 CEST puts him into a thin pre-payrolls tape, nearly four hours before the US employment report. Whatever he moves, the American number gets the chance to move it back. And the structural problem is unchanged — a 10-year gilt holding above 5%, more than 125bp over Bank Rate, is pricing fiscal and inflation risk rather than a policy path. Bailey can shift the front end. He has repeatedly demonstrated that he can do very little about the long one.

August employment is forecast at 15.8K after July's 75,000, a gain that was among the strongest of the year and genuinely broad: core-aged workers led it, with wholesale and retail trade, finance, professional services and construction all adding, against declines in public administration and agriculture.

Consensus is not calling for a reversal so much as normalisation — treating July as a single strong month inside a labour market that has otherwise been mediocre. Coming two days after a Bank of Canada hold, this is the first data with the power to change the October conversation, and it is the reason Wednesday's statement will probably avoid committing to anything.

Another print near 75K would make two consecutive strong months and, stacked on 3.3% annualised second-quarter growth, would start the market pricing the Bank's next move as a hike rather than a cut. That is a bigger repricing than it sounds, because almost none of it is currently in the curve. A negative print would be dismissed as payback for July unless the unemployment rate moved with it.

The practical difficulty is that it lands at the same minute as US payrolls. CAD's reaction on the first Friday of the month is almost always the residual of two labour reports, and the American one wins. Clean Canadian signals on this day are rare enough that positioning off them is usually a mistake.

Forecast unchanged at 6.4%, where July's report left it after a 0.1 percentage point decline alongside the 75,000 employment gain.

6.4% is not a tight labour market by Canadian standards, and that slack is precisely what gives the Bank of Canada cover to hold at the bottom of its neutral range while inflation sits above target. The unemployment rate is doing more work in the Bank's reaction function than the monthly employment count, because it is the variable that has stayed stable while the headline jobs number has swung.

A tick down to 6.3% alongside a decent employment gain is the combination that would turn the Bank's next move genuinely hawkish, since it removes the slack argument entirely. A tick up to 6.5% neutralises the second-quarter growth beat and quietly restores the easing option that Wednesday's statement will most likely have left open rather than closed.

Average hourly earnings are forecast at 0.3% after July's 0.1%. In a week when the Fed chair has just told the market that underlying inflation has not improved, this is the line he was talking about.

Warsh put headline PCE at 3.7% over twelve months and 4.1% over six on Friday — an inflation trend that is accelerating on the shorter window rather than fading. Core PCE for July, released two days earlier, came in at 0.2% month-on-month and 3.3% year-on-year. Wage growth is the mechanism by which an energy-driven price shock becomes a persistent one, and it is the variable that decides whether 'work to do' means one hike or a sequence.

0.4% or above would be the single most hawkish number of the week — more so than a strong payroll count, because it speaks to persistence rather than to activity, and persistence is the stated concern. A repeat of 0.1% is the strongest available argument that the labour market is cooling fast enough to do the Fed's work without a rate rise, and it would take the front end back toward where it sat on Thursday.

The September meeting carries a Summary of Economic Projections, so this print feeds a dot plot as well as a decision. Wage data landing at the margin of a forecast round has outsized influence on where the median dot ends up, and the median dot will outlive the meeting.

The August employment report, forecast at 58K after July's outright decline of 23,000, is the last payroll print before the FOMC meets on 15-16 September. There is no second look.

It is also a hard number to read cleanly, and that is new. A week ago the BLS published its preliminary benchmark revision for March 2026 at -79,000 for total nonfarm employment and -178,000 for private employment, confirming that the establishment survey has been over-counting. The level of employment is lower than the monthly prints implied. That makes the run rate more informative than the level, and the run rate — a negative July after a soft June — has been poor.

A print at or above consensus with the unemployment rate holding at 4.1% lets Warsh's characterisation stand: a labour market consistent with full employment, and a September decision that can be argued on inflation alone. That is the path of least resistance for the hawkish case, and it is roughly what the front end started pricing on Friday afternoon.

A second consecutive negative print puts the Fed in the genuinely uncomfortable position of contemplating tightening into visible labour-market contraction. That is where the market's conviction would break — but note the asymmetry, which is unusual. Normally a weak payroll rallies the front end hard. With a chair who has explicitly prioritised inflation and refused to pre-commit, a weak print buys less dovish repricing than it used to. Friday's 14bp move in the 2-year was the market saying exactly that: the reaction function has changed, and nobody has finished calibrating it.

Forecast unchanged at 4.1%. Warsh cited this level by name on Friday, calling the jobless rate low by historical standards and the labour market consistent with full employment. It is the statistic on which the entire case for tightening rests.

The household survey has been notably steadier than the establishment survey, which is part of why the unemployment rate has become the Fed's preferred anchor while payroll counts get revised away — a preference the -79,000 benchmark revision has just vindicated. A stable 4.1% alongside negative payroll prints is an awkward combination, and the Fed has resolved that tension in favour of the household number.

4.2% or higher, particularly on a rising participation rate rather than a falling one, would be the first evidence that the labour market is loosening rather than merely hiring less. That immediately complicates a September hike and would do more to the front end than a weak payroll count. A fall to 4.0% removes the last obstacle to the hawkish argument.

This is the number that has to break for the hawkish case to break. Everything else on Friday's calendar is corroboration, which is why a 0.1 percentage point move in a rounded statistic can matter more this week than the headline it accompanies.