2026-09-07 → 2026-09-11

The main refinancing rate is forecast at 2.65% against 2.40% now, which is a 25bp hike and the second of a cycle that began with June's surprise. On 11 June the Governing Council raised the deposit facility, main refinancing and marginal lending rates to 2.25%, 2.40% and 2.65% respectively, the first increase in three years, and then held all three on 23 July with an explicit commitment to a data-dependent, meeting-by-meeting approach. A move to 2.65% on the MRO carries the corridor with it: deposit to 2.50%, marginal lending to 2.90%.

The case for hiking was made by Eurostat on 1 September. Euro-area annual inflation jumped to 3.3% in August from 2.9% in July, and the driver is unambiguous — energy at 14.3% year-on-year against 10.3% the month before. That is the Middle East energy shock the ECB flagged in July as standing close to the June projection baseline and well above pre-conflict levels, with the full inflationary impact still to play out. Two consecutive upside months on headline, with the second more than a percentage point above target, is the kind of sequence that costs a central bank its patience.

The case against is in the same release. Services inflation actually fell to 3.0% from 3.3%, which is the domestically generated component and the one monetary policy can reach. Non-energy industrial goods rose to 1.2% from 0.9% and food, alcohol and tobacco held at 1.2% — neither is a wage-price spiral. A Council that hikes here is tightening into an energy shock it cannot influence while the underlying measure it can influence is improving, and some members will say so.

A hike delivered as expected is largely in the price and the reaction will sit in the statement language rather than the number; EUR should firm modestly and the front of the German curve cheapen. A hold would be the genuine surprise and would take the euro down hard, because the market would read it as the Council conceding that a supply shock is not its problem. The tail risk almost nobody is positioned for is 50bp, which would require the new projections to have moved the 2027 inflation path materially — and if the Council were going there, Thursday's projection round is exactly the meeting where it would.

The statement is where the June-hike-then-July-hold sequence has to be reconciled into a stance, and the wording that matters is whatever replaces 'data-dependent and meeting-by-meeting'. That formula was how the Council bought itself optionality in July. Keeping it verbatim after a second hike says the Council is reacting to prints, one at a time, with no path. Adding anything directional — a reference to further steps, or to rates being at or near restrictive levels — converts Thursday from a data point into a signal about October and December.

September is a projection meeting, and the projections carry more weight than usual because June's baseline was built around an energy path that has since delivered 14.3% annual energy inflation. Watch the 2027 and 2028 headline numbers rather than 2026. A 2026 revision higher is mechanical — the energy is already in the index. A 2027 revision higher is the Council saying the shock is propagating into wages and services, and that is the number that justifies a hiking cycle rather than a two-step adjustment.

The second-round-effects language is the specific thing to read. In July the Council's position was that the energy shock's full inflationary impact had yet to play out. If Thursday's statement now says second-round effects are materialising, the market will price a terminal rate meaningfully above 2.50% on the deposit facility. If it says they remain contained while headline is temporarily elevated, this hike looks like insurance and the euro curve should flatten.

For a market whose dominant question is American, the statement's practical function is relative. The ECB hiking into 3.3% headline while the Fed is still deciding whether to move at all narrows the policy gap from the European side — which is the mechanism that has kept EUR bid despite a US front end pricing 65% odds of a hike. A hawkish ECB statement and a soft US CPI on Friday is the combination that produces the largest move in EURUSD this week, and the two are only twenty-four hours apart.

Core PPI is forecast at 0.3% month-on-month after July's 0.2%. July undershot: consensus had been 0.3% and the print came in a tenth below it, part of a report in which headline final demand was outright unchanged against an expected 0.2% rise. A forecast of 0.3% is the market assuming July was the anomaly rather than the trend.

The detail worth carrying into Thursday is that July's core ex trade services rose 0.4% — double the ex-food-and-energy figure. Trade services is the margin line, and a soft reading there while the underlying goods-and-services core runs at 0.4% means wholesalers and retailers absorbed cost increases rather than passing them on. That is a compression that reverses eventually, and when it does it shows up in core PPI first and core CPI a month or two later.

The reason a wholesale price index gets attention on this particular Thursday is that it is the last inflation read the FOMC receives before Friday's CPI, and PPI's health-care and portfolio-management components feed directly into the PCE deflator that Warsh cited at 3.7% over twelve months and 4.1% over six. A core print at 0.4% or above would push the front end toward pricing a September hike as the base case rather than a coin flip, before CPI is even published.

A print at 0.1% or below does the opposite and gives the doves on the Committee their first real piece of evidence in a month. The asymmetry is worth naming: after Friday's payroll beat, the burden of proof has shifted. Soft inflation data no longer merely supports a hold — it has to actively unwind a repricing that two consecutive surprises have already put in place.

Headline PPI is forecast at 0.4% against July's 0.0%. That is a four-fold-plus jump in the monthly rate off a flat base, and the composition of July explains why the forecast is so much higher: final demand goods fell 0.7% on the month, offsetting a 0.2% rise in services and a 2.2% advance in construction. A goods drop of that size is an energy and commodity effect, and it is not the sort of thing that repeats two months running.

The annual number is the one that gets lost. Final demand prices rose 4.7% over the twelve months to July on an unadjusted basis, which is well above anything consistent with 2% consumer inflation once margins normalise. The monthly series has been volatile enough this year to be nearly uninformative on its own; the annual rate says the pipeline is still carrying pressure regardless of which month the energy line happens to fall in.

If the print lands at 0.4% as forecast, the read-across to Friday is neutral-to-hawkish and mostly mechanical — energy back in, goods no longer subtracting. If it lands materially above, say 0.6%, the market will not wait for CPI: the 2-year would cheapen through 4.40% and the dollar would extend, because the combination of 162,000 payrolls and accelerating producer prices is the exact condition Warsh set out as leaving the Committee with work to do.

In isolation this is the least important of the week's four American numbers, and it is published at the same minute as core PPI, which is the one desks actually trade. Its function is as a tell for Friday. Two hawkish surprises in two days would leave the 15-16 September meeting effectively decided before the blackout begins.

Lagarde's press conference thirty minutes after the decision is where a 25bp hike becomes either the second step of a cycle or the last step of an adjustment, and the distinction is worth more to markets than the rate itself. The question she will be asked repeatedly is why the Council is tightening against an energy shock while services inflation is falling, and the shape of that answer sets the October pricing.

The specific formulation to listen for is whether she frames 3.3% headline as a level the Council must respond to or as a path it must look through. In July the line was that the outlook for energy prices, while highly volatile, sat close to the June baseline and well above pre-conflict levels with the full impact yet to play out. Repeating that while hiking is coherent only if the Council now believes the shock is propagating. Pressed on it, she either confirms that belief — hawkish, and the front of the German curve cheapens further — or she does not, in which case the hike reads as insurance and the euro gives back its gains.

The second thing to watch is dissent. June's hike was a surprise and July was a hold; a Council that arrives at two hikes in three meetings during an energy shock is not going to be unanimous, and Lagarde's characterisation of the discussion — broad consensus, large majority, or the careful silence that means neither — is the cleanest available signal on whether October is live.

The timing is the practical risk. The press conference runs from 14:45 CEST, which is fifteen minutes after US PPI prints at 14:30. European desks will be trading a US inflation number and a live ECB press conference simultaneously, which historically produces the widest spreads and the least reliable price discovery of the week. The move that holds into Friday is rarely the move made in that half hour.

July monthly GDP is forecast at 0.0% after June's 0.3%. The monthly path this year has been genuinely weak: April fell 0.1%, May was flat and revised down from an initially reported 0.1% gain, and June's 0.3% was the first solid month in a quarter. A flat July would mean the June rebound bought the economy one month rather than a trend.

The Bank of England is the reason this matters more than a monthly activity print usually does. Bank Rate has been 3.75% since before the summer and the MPC held there on 30 July by six votes to three, with the three dissenters voting to hike 25bp to 4.00%. The next decision is 17 September, six days after this release, and the market is fully pricing a hike by year-end with another by March 2027 — a path driven by inflation persistence and by concerns about UK fiscal sustainability rather than by any strength in output.

That gives Friday's number an unusual asymmetry. A stronger print — 0.2% or better — removes the last argument the six holders have for waiting and pulls the September meeting genuinely live; gilts cheapen and sterling firms. A flat or negative print does not remove the hike from the curve, because the curve is pricing an inflation and credibility problem rather than an overheating one. It simply makes the hike harder to justify, which is a worse position for a central bank than either a clear yes or a clear no.

The market context is a 10-year gilt yield that has been trading above 5% — the highest of the major sovereigns by a wide margin, some 39bp above US Treasuries and 180bp above Bunds. At that level the UK curve is no longer primarily a growth or policy instrument; it is a fiscal one. A weak GDP print that raises the deficit path is capable of cheapening gilts and weakening sterling at the same time, which is not how a normal currency responds to bad growth news and is exactly how this one has been responding.

Core CPI is forecast at 0.2% month-on-month, matching July. Two-tenths a month is a 2.4% annualised pace, which is close enough to target that on this measure alone there would be no policy debate at all. The debate exists because of everything the measure excludes.

July's internals are the reason the forecast is credible. Shelter rose just 0.1% and accounted for roughly two-thirds of the entire headline increase, food rose 0.1%, and the print landed exactly in line with consensus. Shelter is the largest core component and the slowest-moving; at 0.1% it is running below its own multi-year average and mechanically caps how high core can print.

The hawkish scenario is 0.3% or above, and it would matter disproportionately because it would be the first month this quarter in which the core measure confirmed rather than contradicted the headline. That is the print that makes a September hike near-certain: the Committee could then argue that the energy shock has propagated into underlying prices, which is precisely the condition Warsh named. A 0.1% print is the dovish tail and would put real pressure on the 65% odds now in the curve, though after a 162,000 payroll beat it would take more than one soft core month to unwind the repricing.

The honest reading is that this single number is the most policy-relevant figure of the week and the least likely to surprise. Core CPI has printed 0.2% repeatedly; the distribution around it is tight. Which means the risk on Friday sits in the headline and in the annual figures, not here — and a market positioned for a core beat is positioned against the base rate.

Core CPI is forecast at 2.4% year-on-year, down from 2.5% in July, which itself was a tenth below June. A third consecutive decline would put underlying consumer inflation within four-tenths of target and falling — on the narrowest reading of the mandate, an argument for doing nothing on 15-16 September.

Set it against the number the chair actually cited and the problem appears. Warsh's Jackson Hole framing rested on 12-month PCE at 3.7% and the six-month change at 4.1%. A core CPI of 2.4% and a headline PCE of 3.7% are not contradictory — different baskets, different weights, one excludes energy and the other does not — but they support opposite conclusions, and which one the Committee leads with on 16 September is the entire meeting.

The forecast also widens the headline-to-core gap to a full percentage point, from 0.9pp in July, with headline held at 3.4%. A gap that wide and still widening is the signature of a supply shock rather than a demand problem, and supply shocks are the textbook case for looking through. The counter-argument, and the one the hawks will make, is that a shock is only worth looking through if it does not last; energy has now been running above 14% annually for months, which is long enough for expectations to start moving.

A print at 2.5% or higher — no decline at all — would be the genuinely hawkish outcome, because it would remove the disinflation trend that is currently the strongest single argument for patience. At 2.3% or below the doves have their number, and the front end would rally hard into the blackout. The forecast itself, 2.4%, resolves nothing, which is the most likely outcome of the week.

Headline CPI is forecast at 0.4% month-on-month against July's 0.1%, and that four-fold acceleration is the largest forecast gap on this week's calendar. July's 0.1% was held down by energy falling 1.5% on the month; a 0.4% forecast implies the market expects that subtraction to become an addition, which is consistent with the same energy line that pushed euro-area headline inflation to 3.3% in August.

The arithmetic worth doing by hand is that a 0.4% monthly print alongside an unchanged 3.4% annual rate requires the August 2025 base month to have contributed roughly the same amount. So the year-on-year figure holds still while the monthly rate quadruples. That is a warning about how to read Friday: the annual number will look calm and the monthly number will not, and both will be describing the same data.

For the September meeting the monthly figure is what the hawks will use. A 0.4% print is a 4.9% annualised pace on headline consumer prices, and a Committee whose chair has publicly anchored on a 4.1% six-month PCE rate will not find it easy to describe that as progress. Above 0.5% the front end prices a hike as the base case and the dollar extends; below 0.2% would be a genuine shock in the other direction and would put the September meeting back to a coin flip.

The wider point is that this is the last data the FOMC sees. Whatever prints at 14:30 CEST on Friday sits unchallenged for four days until the Committee convenes, with no further releases and no communication permitted. That gives an ordinary monthly inflation number unusual staying power — there is nothing scheduled behind it to correct an outlier in either direction.

Annual headline CPI is forecast unchanged at 3.4%. July's 3.4% was itself a tenth below June, so a flat forecast describes disinflation that has stalled rather than reversed — and stalled a full 1.4 percentage points above target, in a year in which the Federal Reserve has not yet tightened.

The composition is what keeps this from being straightforwardly hawkish. Energy rose 14.7% over the twelve months to July even as it fell 1.5% on the month, which means the annual headline figure is substantially a Middle East supply story rather than an American demand story. That is the same shock now running at 14.3% annually in the euro area — where the ECB is nonetheless expected to hike on Thursday, one day before this print, which is a precedent the hawks on the FOMC will not have to be reminded of.

The scenarios are asymmetric in an unusual way. A print at 3.5% or above breaks the disinflation narrative outright and, coming two days after a PPI forecast at 0.4% and a week after 162,000 payrolls, would make a September hike very difficult to argue against; the 2-year would cheapen through 4.40% and the curve would bear-flatten further. A print at 3.2% or below is the only outcome on this calendar capable of taking the hike out of the curve on its own, because it would be the first evidence the energy pass-through is genuinely fading.

In isolation an annual headline rate is the least informative of the four CPI figures published at this timestamp — it is a twelve-month average dominated by base effects and by a component the Fed cannot influence. Its importance on Friday is entirely about what it is next to: a 4.1% six-month PCE rate, an unemployment rate at 4.1% the chair has called consistent with full employment, and a Committee that has been told it has work to do unless underlying inflation is clearly converging. Three of those four facts point the same way.