2026-09-14 → 2026-09-18

Statistics Canada publishes August CPI, and consensus looks for a 0.1% monthly fall after July's 0.5% rise. That would be the first negative monthly print since the spring and it would come almost entirely from one line. July's acceleration to 3.0% year-on-year from 2.8% in June was driven by gasoline, which rose 25.7% annually against 20.5% the month before, the pass-through from the Strait of Hormuz blockade and the partial closure of Red Sea shipping routes in late July. Strip gasoline out and July inflation was 2.2%.

The forecast, then, is a bet that the energy shock has stopped getting worse rather than a bet that it is reversing. That distinction is what the Bank of Canada cares about. On 2 September the Bank held the overnight rate at 2.25%, where it has sat since the October 2025 cut, but it explicitly warned that upside risks to inflation have increased and flagged the risk of spillover from high oil prices into the prices of other goods and services. A monthly print at or below consensus supports the hold; a positive surprise driven by anything other than gasoline does not.

For the currency the asymmetry is unusual. The Canadian dollar is trading against a US dollar whose central bank is two days from hiking, and the Bank of Canada is on hold with a policy rate roughly 140bp below the midpoint of the Fed's current target range. A soft Canadian print widens that gap and does the loonie no favours. A hot print is more interesting, because it would force the market to price a Bank of Canada that has been describing upside inflation risk without acting on it, and the front end of the Canadian curve is not positioned for that conversation.

In isolation the headline monthly number matters less than the two core measures released alongside it, which is the right way to read this release. A -0.1% headline that comes with stable core is the Bank's base case arriving on schedule. A -0.1% headline that comes with core drifting up is the spillover the Bank warned about, and it is the version of this release that moves rates.

CPI-median is one of the Bank of Canada's two preferred core measures and consensus expects it unchanged at 2.0%. That is the number the Bank has been leaning on. In its 2 September statement it noted that core measures remained close to 2% in July even as headline hovered around 3%, which is the whole basis for holding the policy rate while headline inflation runs a full point above target.

The measure works by taking the price change at the 50th percentile of the CPI basket weighted by expenditure, which makes it structurally resistant to precisely the kind of shock Canada is absorbing. A 25.7% annual move in gasoline is one component; the median does not care how large the outlier is, only which side of the middle it sits on. That is a feature when the shock is genuinely confined and a liability when it starts to broaden, because the median is the last measure to register a broadening.

So the informative outcome is not 2.0%. It is 2.1% or higher. A median that starts to lift while headline is being dragged down by energy would tell you the second-round effects the Bank flagged are arriving, and it would do so in the one series the Bank has committed to watching. That combination — soft headline, firming median — is the configuration that would take a October Bank of Canada move from unpriced to live.

Absent that, an on-consensus 2.0% is a non-event for the Canadian dollar and for the front end. The Bank has told the market what it needs to see, and an unchanged median is the Bank getting what it asked for.

CPI-trim is the Bank of Canada's second preferred measure, excluding the 20% of the basket with the most extreme price changes at either end, and consensus has it unchanged at 1.9%. Read alongside the median at 2.0%, the two measures bracket the target from just below, and they have done so consistently while headline inflation has run near 3%.

The gap between trim at 1.9% and headline at 3.0% is the cleanest single statistic in Canadian macro right now: 110bp of inflation that the Bank has judged to be outliers rather than trend. The trimmed measure is more aggressive than the median in discarding tails, so it will be slower still to pick up an energy shock — but it is also the measure that moves first if the shock stops being a tail and becomes the distribution, because trim removes a fixed share of the basket rather than a fixed list of components.

Watch the two core measures together rather than separately. Both unchanged validates the Bank's characterisation and the hold at 2.25%. Both up a tenth is noise. Trim up with median flat is the technically interesting case, because it implies the price increases are spreading across more components while the middle of the distribution has not yet moved — the earliest signature of broadening you can get from monthly data.

For positioning, this release completes a set that either confirms or breaks the Bank of Canada's story, and it lands two days before the Fed. A Canadian core complex that stays at 1.9-2.0% leaves the loonie carrying the full weight of a widening policy differential on Wednesday, with no domestic offset.

The ONS labour market release lands two days before the Bank of England decides, and consensus looks for the claimant count to rise by 8.3 thousand after a fall of 11.0 thousand the month before. The swing matters more than the level: a 19.3 thousand month-on-month turnaround in the direction of the flow, from improving to deteriorating, in the one high-frequency labour series the MPC can actually rely on.

That reliability caveat is doing real work. The ONS has said Labour Force Survey estimates are less reliable than usual, partly because of a data collection error in May and June, and has asked users to read them alongside other sources. That guidance has pushed the claimant count and the HMRC payrolled-employees series to the front of the MPC's dashboard by default. On the August bulletin the unemployment rate was 4.9% for April to June, up 0.2 percentage points on the year; payrolled employees fell 94,000 over the year on July's provisional estimate; vacancies slipped 6,000 to 707,000 in May to July, below pre-pandemic levels.

Set against that, wage growth is the part that has not cooperated. Regular pay grew 3.5% and total pay 4.1% in the three months to June, which is above what the Committee considers consistent with 2% inflation once you allow for productivity. This is the central tension in the UK data and it is why the MPC split 6-3 in July: three members looking at 4.9% unemployment and falling payrolls and voting to hold, three looking at pay growth and services inflation at 3.4% and voting to hike to 4%.

A claimant count near or above the 8.3 thousand forecast, on top of falling payrolls, arms the doves for Thursday and should take some tightening premium out of the front of the gilt curve. A negative surprise — another fall — is the one that would genuinely unsettle the consensus for a hold, because it removes the labour-market argument from the three members currently using it and leaves them facing Wednesday's inflation print with nothing to offset it.

August CPI is forecast at 3.1% year-on-year against 2.9% in July, and it arrives twenty-eight hours before the MPC announces. July's 2.9% was itself an acceleration from 2.6% in June, the first increase in the annual rate since March, with core steady at 2.6% and services inflation easing from 3.6% to 3.4%. A 3.1% print would be the second consecutive rise and would put headline inflation more than a full percentage point above the 2% target.

The composition will decide how the Committee reads it. The UK is absorbing the same energy shock as the euro area and Canada, and imported energy inflation is, in the MPC's standard framing, something to look through. Services inflation is not. Services at 3.4% with regular pay growing 3.5% is a domestically generated inflation rate that has proven durable, and the three members who voted to hike in July — Greene, Pill and Mann — have consistently cited it. If headline rises to 3.1% while services falls again, the doves keep the argument. If services turns back up, they do not.

The market has Bank Rate on hold at 3.75% on Thursday with a vote expected to repeat July's split, and sterling's reaction function into this print is therefore about the meeting after, not this one. An upside surprise on services pulls forward the first hike and steepens the front end; a downside surprise on services pushes the whole tightening question into November's Monetary Policy Report round, when the Committee will have a fresh forecast to hang a decision on.

The wider context is that Britain is now the only major economy in this week's calendar whose central bank is not expected to move. The Fed hikes on Wednesday evening, the BoJ is expected to hike on Friday, the ECB hiked last Thursday and the RBNZ hiked a fortnight ago. A 3.1% CPI print puts the Bank of England's inaction into sharper relief than it has been at any point this year, and that is a sterling story independent of what the Committee actually does on Thursday.

The FOMC is expected to raise the target range to 3.75-4.00% from 3.50-3.75%, where it has sat since December 2025. It would be the first increase of this cycle and the resolution of a repricing that has run for nineteen days. On 28 August, before Chair Warsh spoke at Jackson Hole, market-implied odds of a September move were roughly 30%. After the keynote they were a coin flip; after the 4 September payroll report at 162,000 they were around two-thirds; and after last Friday's CPI they were about 90%, up from roughly 70% the day before.

The data did the work. August CPI rose 0.4% on the month and 3.4% on the year, with core at 0.3% monthly — a tenth above consensus — and 2.4% annually, which is the lowest core rate since March 2021. That split is the entire argument of this meeting in two numbers. The headline is being pushed by energy, up 2.1% on the month and 16.3% on the year with gasoline at 27.4% annually; the core is decelerating, with shelter easing to 3.0% from 3.2% and food to 2.7% from 3.0%. Thursday's PPI told the same story from the other end of the pipeline: final demand up 0.4% monthly and 5.4% annually, with goods up 1.1% against services up 0.1%, and more than a third of the goods increase traceable to diesel at 24.1%.

Warsh's stated condition at Jackson Hole was that the Committee must be confident underlying inflation is moving to target clearly and at sufficient speed, and that absent that confidence it has work to do. A core rate at a five-year low is evidence for the defence; a headline rate at 3.4% with twelve-month PCE running at 3.7% and the six-month change at 4.1% as of his speech is evidence for the prosecution. The Committee is choosing to hike into an energy shock while its preferred underlying measure improves, which is a defensible but genuinely contestable call, and the dissent count will be the tell.

Because the move is close to fully priced, the rate decision itself is not the trade. A hike lands on a 2-year already at 4.63% after cheapening 24bp last week; it is in the price. The asymmetry runs the other way entirely: a hold, at 90% implied odds, would be the single largest rates surprise of the year and would take the front end down hard. Short of that, the reaction function on Wednesday evening belongs to the projections and the press conference, not to the number itself.

The Summary of Economic Projections, and the dot plot inside it, is the release that carries this meeting. With a 25bp hike at roughly 90% implied probability, the marginal information in the decision is close to zero and essentially all of it sits in where the median participant puts the funds rate at end-2026 and end-2027. This is the first full projection round since Warsh reoriented the Committee's guidance, and the first since the energy shock became the dominant feature of the inflation data.

Three things need to be read together. The 2026 median dot tells you whether Wednesday is one hike or the first of several — a median implying one further move this year makes October or December live, a median implying none frames this as a single insurance adjustment. The inflation projections tell you whether the Committee is treating the energy shock as transitory: a headline PCE forecast well above core for 2026 that converges in 2027 is the look-through framing, while upward revisions to the 2027 core path would say the Committee now expects second-round effects. And the longer-run dot tells you whether any of this has changed the Committee's view of neutral.

The comparison that will be drawn immediately is with the ECB, which published its own projection round last Thursday alongside a 25bp hike: headline inflation averaging 3.0% in 2026, 2.5% in 2027 and 2.1% in 2028, core at 2.5%, 2.6% and 2.3%, and growth revised up to 0.9% for 2026 and 1.4% for 2027. That is a central bank forecasting inflation back at target only in 2028 and hiking to get there. If the Fed's numbers describe a faster convergence on a shallower path, the dollar-euro rate differential story narrows from the American side.

For positioning, the dots are where a hawkish surprise can still be delivered on a day when the rate move cannot be. A 2026 median showing a second hike, against a market that has largely priced one and then a pause, flattens 2s10s further from a spread already at +33bp and is the most likely route to a stronger dollar out of this meeting. A median showing this hike and done steepens the curve and takes the front end back down, and given how much tightening the 2-year absorbed last week, that is the larger of the two available moves.

The statement is short and this month its job is narrow: to explain a hike delivered while the Committee's preferred underlying inflation measure is improving. The characterisation of inflation is the sentence to read. Language that attributes the elevated headline rate specifically to energy, while noting progress in underlying measures, frames Wednesday as an insurance move against second-round effects. Language that describes inflation as elevated without that attribution frames it as the start of a sequence.

The second thing to read is what remains of forward guidance. Under Warsh the Committee has pared back explicit commitments and pushed the market towards the incoming data, which is why individual releases have moved pricing so violently this cycle — the 20-point jump in hike odds on Friday's CPI being the clearest case. A statement that reintroduces any conditional language about the path would be a meaningful change in the Fed's communication strategy and would matter well beyond this meeting.

Third, the dissents. A hike into an energy shock with core at a five-year low is exactly the decision that produces them, and they can come from either side: a participant who thinks the improvement in core argues for patience, or one who thinks a 3.4% headline and 3.7% twelve-month PCE argue for more than 25bp. Two or more dissents in either direction would tell the market that the Committee's centre is thinner than the vote count suggests, and would raise the weight placed on every speech between now and October.

In isolation the statement rarely moves markets when the decision is priced and the projections publish simultaneously. Its function this month is corroborative: does the prose match the dots. A hawkish dot plot paired with a statement that leans on the energy attribution is an internally inconsistent package, and the market will resolve the inconsistency in Warsh's direction thirty minutes later.

Warsh takes questions half an hour after the decision, and on a day when the rate is priced and the dots are published, the press conference is where the week's largest single-asset moves have historically originated. His Jackson Hole keynote on 28 August moved September hike odds from roughly 30% to a coin flip in one session and bear-flattened 2s10s by 8bp; the market has learned to treat his unscripted remarks as policy-relevant in a way it had stopped doing with his predecessor.

The question he cannot avoid is the one the data poses: if core CPI at 2.4% is the lowest since March 2021, why hike now. The available answers point in different directions. If he argues that the energy shock risks contaminating expectations and that the Committee is acting pre-emptively, this is a one-move insurance hike and the front end should richen. If he argues that twelve-month PCE at 3.7% is the relevant measure and that core CPI understates the underlying trend, the market will price a sequence, and the 2-year has room to cheapen further from 4.63%.

The second question is about the reaction function itself. Warsh has pared back forward guidance and told markets to follow the data; markets have obliged by repricing 60 points of probability in nineteen days. He will be asked whether that volatility is a feature or a problem. Any indication that the Committee is uncomfortable with the sensitivity — any step back towards signalling — would compress rates volatility broadly and is the most underpriced possible outcome of this meeting.

The third is Canada's question in American form: what would make the Committee stop. A named condition — a core threshold, a number of months, a specific measure — is worth more to the curve than anything in the statement, because it converts the Fed's path from a judgement call into something the market can price against incoming data. His record suggests he will decline to give one, and the absence of a condition is itself the answer that keeps every CPI print between now and December as market-moving as Friday's was.

Stats NZ publishes June quarter GDP, and this is the most genuinely two-sided number on the week's calendar. Consensus sits at 0.1% quarter-on-quarter after 0.8% in the March quarter, but the dispersion around it is wide and, unusually, every forecaster sits at or above the RBNZ. The Bank's own projection is flat at 0%; ASB looks for 0.3%, Westpac and BNZ for 0.2%, Kiwibank and ANZ for 0.1%. A print anywhere in that band beats the central bank's forecast.

The context is a Reserve Bank that has already restarted tightening. On 2 September the Monetary Policy Committee raised the Official Cash Rate to 2.75% by consensus, the second consecutive hike, with inflation at 4.1% in the June quarter driven largely by fuel prices connected to the Middle East conflict. The published OCR track signals a pause in October and a further 25bp to 3.00% in December, and the Bank expects inflation to stay above 3% for the rest of 2026 before returning inside the 1-3% band next year.

That track is what this release tests. The RBNZ is tightening into an economy it forecasts to have stalled outright in the June quarter — a defensible position when the inflation overshoot is 210bp and fuel-driven, but a fragile one. A GDP print at the top of the forecast range, 0.3%, tells the Bank the economy absorbed the fuel shock better than it assumed and makes the December hike look under-priced; a print at or below zero validates the flat forecast and puts the October pause under pressure to become a longer one.

For the currency the setup is unusually clean because the RBNZ's own number is the lowest on the board. The asymmetry favours an upside surprise moving the New Zealand dollar more than a downside surprise, since a miss merely confirms what the Bank has already published and priced, while a beat forces a revision. The caveat is that the June quarter is now nearly three months stale and the market will be reading it primarily for what it implies about the September quarter, which the RBNZ has at 0.5%.

The Monetary Policy Summary and minutes publish at noon London time, and with Bank Rate expected unchanged the prose is the release. The Committee held at 3.75% on 29 July by a majority of 6-3, with three members voting to raise to 4% — Catherine Mann having joined Megan Greene and Huw Pill. The summary's job on Thursday is to explain a second consecutive hold against an inflation rate that will, if Wednesday's forecast is right, have risen to 3.1%.

The specific language to watch concerns how the Committee characterises the energy contribution. The UK is absorbing the same fuel shock that pushed euro-area inflation to 3.3% in August on a 14.3% annual energy rate and Canadian gasoline to 25.7% annually. Every central bank in the week's calendar has had to decide whether to look through it. The ECB decided not to and hiked; the RBNZ decided not to and hiked; the Fed is expected not to and hike. A summary that leans heavily on the imported, temporary nature of the shock is the Bank of England explaining why it is the outlier.

The second thing is the treatment of services and pay. Services inflation eased to 3.4% in July from 3.6%, regular pay growth is 3.5%, and the labour market data available to the Committee — 4.9% unemployment, payrolled employees down 94,000 on the year, vacancies at 707,000 — points to loosening. If the summary frames the pay-services complex as clearly on a downward path, November stays live but not urgent. If it describes progress as having stalled, the market will price the November Monetary Policy Report round as a hike meeting.

The third is any guidance on the pace of the balance sheet run-off and on the Committee's reaction function to a further energy-driven overshoot. With gilt yields having risen 8bp in the first three sessions of last week to 5.192% at the 10-year — the highest of the window — the Bank's tolerance for an inflation overshoot it cannot control is being tested in the long end whether or not it moves Bank Rate.

Consensus expects the vote to repeat July's split exactly: three for a hike, none for a cut, six to hold. That configuration has been stable for one meeting, having moved from 7-2 to 6-3 in July when Mann joined Greene and Pill on the hawkish side. The market's central case is that the Committee stands still while the direction of travel within it does not.

Vote counts on a hold are a forward-looking instrument and this one is unusually informative because of where the marginal member sits. Six votes to hold with three to hike means two further defections take the Committee to a hiking majority. One defection takes it to 4-5, which is a hold in name and a hike in expectation, and the front end of the gilt curve would price the November meeting accordingly within minutes. That is the single highest-beta outcome available on Thursday.

The reverse is worth stating because it is not priced at all. A vote of 2-0-7 — one hawk stepping back after Tuesday's labour data and a benign services reading on Wednesday — would be read as the tightening question closing rather than merely being deferred, and would take sterling lower against a dollar whose central bank hiked the evening before. Neither of the individual hawks has shown any inclination to retreat, which is why the market assigns this little weight, but a claimant count rising 8.3 thousand on top of 94,000 fewer payrolled employees is a real argument.

The sequencing this week is what makes the vote count matter more than usual. The Committee will have seen Tuesday's labour market release and Wednesday's CPI before voting, so the split is a direct, same-week read on how nine policymakers weighed a rising headline inflation rate against a visibly loosening labour market. There are very few moments in the calendar where you get that clean an answer.

Bank Rate is expected to remain at 3.75%, where it has been held since late 2025. If that is the outcome, the Bank of England will be the only central bank in this week's calendar not to have moved, and the only one in the G10 holding a policy rate steady through an inflation rate forecast at 3.1% and rising.

The case for holding is genuinely strong on the real economy. Unemployment at 4.9% is up 0.2 percentage points on the year, payrolled employees have fallen 94,000 over twelve months on July's provisional estimate, and vacancies at 707,000 are below pre-pandemic levels. Against a backdrop like that, tightening into an imported energy shock risks the standard policy error of the last two cycles: responding to a price level shift with an instrument that only works on demand.

The case against is that the UK's inflation problem was never purely imported. Services at 3.4% and regular pay at 3.5% are domestic, and both have been sticky enough that three of nine Committee members have been voting to hike through it. A second consecutive hold with headline inflation accelerating for a third month — 2.6% in June, 2.9% in July, 3.1% forecast for August — puts real weight on the Bank's credibility that its 2% target still binds.

For sterling, the decision itself carries less information than the vote split and the summary, both of which publish at the same moment, because the hold is expected with near-unanimity. The tradable question is the policy gap. The Fed hiked on Wednesday, the BoJ is expected to hike on Friday and the ECB hiked last Thursday; a Bank of England that holds and signals no urgency leaves sterling as the only major currency this week whose central bank is standing still while its inflation rate rises, and the gilt market's 8bp cheapening in the first three sessions of last week suggests the long end has already started to price what that means.

The Bank of Japan is expected to raise its policy rate to 1.25% from 1.00% at the end of its 17-18 September meeting, which would take Japanese short rates to a level last seen in 1995. It would be the second 25bp increase of 2026, following June, and it is about as heavily anticipated as a central bank decision gets: Reuters reported on 11 September that the Bank is set to move, a Bloomberg survey published the same day found all 52 BOJ watchers forecasting a hike, and swap contracts implied roughly a 97% probability.

Governor Ueda built that consensus deliberately. On 2 September he said the Bank would consider a rate hike at every policy meeting including the September one, and board member remarks on 10 September reinforced that the Bank intends to keep raising the benchmark to cap the price trend at 2%. The supporting data has cooperated: core consumer prices are approaching the 2% target and producer price inflation reached a three-and-a-half-year high in August, the latter being the pipeline measure the Bank has cited when arguing that the inflation it faces is no longer purely imported.

The thirty-one-year framing is not just colour. A 1.25% policy rate is the level at which the carry economics of the yen funding trade change materially rather than marginally, and it comes in the same week the Fed is expected to take its own range to 3.75-4.00%. The nominal differential still overwhelmingly favours the dollar, but the direction of travel on both sides is now the same, which is a new condition — for most of the last two years the trade has been a Fed on hold or easing against a BoJ inching up.

Because the move is ~97% priced, the yen's reaction will be driven by the accompanying signal rather than the rate. Reuters reported the Bank could signal faster future hikes given the risk of an inflationary overshoot, while also noting Ueda is expected to avoid committing to a timeframe and that there is no internal consensus on how far or how fast rates should rise. The Bloomberg survey found 93% expect another move by January and around a third expect December. A hike delivered with language that validates a December follow-up is the yen-positive outcome; a hike delivered with explicit emphasis on data-dependence and no pace signal is the one that sells the yen on the fact.

The statement accompanying the decision is where the Bank either does or does not open the door to a December move. Reuters' reporting frames the choice precisely: the Bank could signal readiness to speed up rate hikes amid the risk of an inflationary overshoot, but has no pre-set view on the terminal rate or on the timing of further increases. Those two positions are compatible in a statement and incompatible in a market's interpretation, which is why the specific wording will matter more than usual.

The substantive question underneath is whether the Bank has changed its view of where policy needs to end up. A 1.25% policy rate is still deeply accommodative in real terms with core prices near 2%, and the distance to anything resembling neutral is large. If the statement introduces or sharpens language about the appropriate level of rates — as opposed to the pace of getting there — it would be the most significant communication shift from the Bank in this cycle, and the JGB curve would reprice well beyond the front end.

The second thing is the treatment of the energy and import channel. Japan is a net energy importer absorbing the same shock that has pushed euro-area, UK and Canadian headline inflation above target, and the yen's level amplifies it. Producer prices at a three-and-a-half-year high in August are partly that. A statement that distinguishes clearly between imported cost pressure and domestically generated inflation would tell the market which of the two the Bank is actually tightening against, and therefore what data to watch between now and December.

For the 10-year JGB, which closed last week at 2.987% after rising 5bp, the statement is the more relevant document of the two published on Friday. The front end is pricing the hike; the long end is pricing a path, and it has cheapened 7bp in the last two sessions of the week alone on the expectation that the path steepens. Confirmation extends that move; a statement that reads as one-and-done reverses a good deal of it.

Ueda takes questions three hours after the decision and closes the week. The single question that matters is the one Reuters has already flagged he will decline to answer directly: how fast, and how far. Around 93% of surveyed watchers expect another hike by January and roughly a third expect December, so the market has a specific hypothesis for him to confirm or deflect, and the yen will trade off which he does.

His record is one of deliberate ambiguity, and it has generally served the Bank well. Saying on 2 September that a hike would be considered at every meeting was enough to build a 97% consensus for this one without committing to anything. The same technique applied on Friday — every meeting is live, no pre-set path — would leave December priced roughly where it already is, which is a modest yen-negative because the market has been positioning for an acceleration signal rather than a repetition.

The more consequential line of questioning concerns the internal split. Reuters reports there is no clear consensus within the Bank on how much rates should rise or at what pace, and the vote and any dissents published with the decision will give that shape. If Ueda is pressed on divergence within the board and confirms it exists, it caps how much a hawkish statement can be worth, because it tells the market the pace is contested rather than settled.

The timing is the last point worth making. This press conference happens roughly thirty-six hours after the FOMC's, in the same week the Fed is expected to have hiked and published fresh dots. Ueda will be asked about the differential and about the yen, and whatever he says will be read against a US 2-year that finished last week at 4.63%. Both central banks tightening at once is a genuinely new configuration for this pair, and Friday morning in Tokyo is the first time the market gets to price it with both decisions in hand.