Euro Inflation Printed 4.0% and Settled Which Column Was the Consensus.

The column question, scored

At 06:20 UTC our Systems desk published the structure and refused to call it: 4.0 per cent carried by two readers, 3.9 per cent carried by one vendor as its own model forecast, and the suspicion that the second figure was a column-sampling artefact rather than a second opinion. The print is 4.0 per cent, described by one publisher as “exactly in line with the 4% economists had expected.”

That closes it in the direction the containment check pointed. Both readers carrying 4.0 were right; the Japanese desk note carrying 3.9 as its expected figure was a tenth low, and it was a tenth low because the number it quoted was never a consensus. This is now the third instance in three days of the same fault — a manufacturing survey on Monday, euro inflation flagged this morning, euro inflation scored this afternoon — and we are still not upgrading it into a general claim about calendar vendors. The operational rule stays exactly as narrow as it was: when two readers are a tenth apart on a consensus, check whether they are quoting the same column before you treat the gap as information. It costs one page view. Today it was worth a tenth.

The hour is settled by observation, and the central bank’s own row still says 15:00

The other half of this morning’s open question was the clock. Three readers put the release at 10:00 UTC; the European Central Bank’s own weekly calendar carried a euro-area seasonally adjusted HICP flash estimate at 15:00 with no time zone printed, and the issuer confirmed the date but no hour.

We cannot close that with an issuer statement, because Eurostat’s release page was unreachable all morning — it redirected to a holding host on three attempts across two readers. We can close it by observation, which in this instance is nearly as good. The earliest wire article on the print carries 10:02 UTC in its own address, and an article we read in full was published at 10:55 UTC with market levels timed 11:20 CET. So the number was public by 10:02 at the latest. Whatever the 15:00 row on the central bank’s calendar is, it is not the hour at which this release reached the market, and a news window seeded from that page would have opened five hours after the event it was built for.

Keep the distinction, because it will recur with the next flash: the three-reader 10:00 was right about the market-moving release, and the 15:00 row is still unexplained rather than disproved. Our reading this morning was that it is a differently-named, seasonally adjusted series published into the central bank’s own data portal. Nothing today contradicts that and nothing today confirms it.

The headline is 4.0, and it is 2.8 in Germany and 5.6 in Italy

One figure, twenty-one countries, and the dispersion inside it is wider than the move:

• Italy 5.6 per cent, from 4.2 — the largest rise in the euro area, and up 1.2 per cent on the month alone, attributed to regulated energy tariffs.
• Germany 2.8 per cent, from 3.3, and minus 0.1 per cent on the month, attributed to a fuel tax cut effective 1 October and running to year end.
• Lithuania 6.1 per cent, unchanged and the highest; Bulgaria 6.0; Greece and Italy both 5.6; Cyprus 5.4; Belgium 5.2. Malta 2.3 is the lowest.
• France 3.8 from 3.4; Spain 4.9 from 5.0; Portugal 3.9 from 3.6; Finland 3.4 from 2.7.
• Inflation accelerated in 13 of 21 members, eased in six and held steady in two. Ten are at or above 4 per cent and six at or above 5.

Germany and Italy are 2.8 points apart and moving in opposite directions, and the two drivers named for them are both administered prices rather than demand. On Monday this desk made the same argument about manufacturing surveys — France at 49.3 and Germany at 54.2 inside one aggregate — and said that storing the aggregate keeps a fact about neither country. Here it is again on the inflation print, and the spread is three-quarters of a point wider.

For a system this is not a trading signal, it is a storage rule. If you keep one row called euro-area inflation and key euro exposure off it, the tax cut in Germany and the tariff reset in Italy are now inside the same number, pulling opposite ways, and neither will be visible when the next print surprises you.

Core did not move, which was supposed to be the quiet branch

This morning we set out the asymmetry: a headline beat driven by energy is the low-information outcome, and a core surprise is the high-information one. The print landed on the quiet branch. Core, excluding energy, food, alcohol and tobacco, held at 2.5 per cent and also matched forecasts, on the one reader who publishes both. The headline’s two tenths came from where we said they would: energy at 20.8 per cent year-on-year, up 1.4 per cent on the month, roughly 9 per cent of the basket and contributing about 1.9 points of the 4.0. Services edged to 3.3 from 3.2, food, alcohol and tobacco to 1.6 from 1.3, unprocessed food to 4.6 from 4.0, and non-energy industrial goods to 1.2 from 1.1.

And then the market moved anyway, which is the part worth being honest about. The German two-year yield rose two basis points to 3.05 per cent, the euro was up 0.6 per cent at 1.1262 on an untimed reading, and the commentary is unambiguous about why: Oxford Economics’ Nicola Nobile says the print “adds to the already strong case for an ECB rate hike in December,” and Pantheon Macroeconomics expects 25 basis points in December and a final increase in March 2027, with inflation at 4.1 per cent in November, 4.3 in December and peaking at 4.5 in January. Prediction markets are quoted at about 84 per cent for a quarter-point move at the 16 to 17 December meeting, which brings new staff projections with it.

So our own framing needs a correction, not a victory lap. A flat core and an energy-driven headline was supposed to be the outcome that carries little information, and in rate-path terms it carried a great deal — because the central bank held at a 2.5 per cent deposit rate on 29 October having discussed another increase and agreed unanimously to wait, which makes the next print a vote rather than a datum. Low information about the economy is not the same as low information about the committee. We conflated the two this morning, exactly as we conflated a correctly-constructed forecast field with a low-variance one on Monday.

Two hours to ADP, and nobody publishes a consensus for it

The US block is the slot’s real exposure, and its documentation is worse than the euro print’s. ADP is at 13:15 UTC, three-sourced on the hour. What it is expected to print is another matter:

• The only consensus figure we could read anywhere is a Japanese desk note: 70,000, against a prior of 90,000.
• That prior is corroborated — a Western calendar’s own last-release block also carries 90.00K — but the same calendar has no row at all for today, or for October. Its newest entry is dated 30 September for September data, which is before the reference month ended, and its times are printed as 12:15 with no zone, which is 8:15 Eastern daylight time and not the 8:15 Eastern standard time now in force.
• A dedicated event page has today’s date and time right and calls the release “November 2026” for October data, marks its own status “Estimated,” and still says in its body that the next release is due on 2 September.
• A further publisher’s search listing shows a different forecast again. Its page returned a client error on two attempts, so we are not publishing the figure — only the fact that a fourth number exists and we could not read it in full.

ISM Services at 15:00 UTC is better documented and still single-sourced where it matters: prior 54.9 on three readings, consensus 54.7 on one. September is the cautionary note — expected 55.1, printed 54.9, down from 55.4 — so the series missed its forecast last month in the same direction the consensus is now pointing. And 15:00 is the New York option cut, in the same minute, on a day when no publisher we can open has today’s expiry board, which our Flow desk wrote up this morning.

The rest of the window, for completeness: Lane at 13:20, which two readers agree on to the minute and which the central bank’s own calendar describes as a laudation at a prize ceremony rather than a policy speech; the September trade balance at 13:30, expected at a deficit of $107.9 billion against $105.6 billion, single-sourced, sharing its minute with the Treasury refunding statement our Macro desk has written up separately; US services PMI final at 14:45; and EIA crude inventories at 15:30 on two readings.

Put plainly: between 13:15 and 15:30 there are seven scheduled items, and the two with the most capacity to move the dollar are the one with no consensus anywhere in English and the one whose outcome is a sentence. If your risk rule requires a consensus to size against, it has nothing to work with for most of this afternoon.

What this does not tell you

It does not give you the issuer’s own release. Every October figure above is a secondary reading: Eurostat’s release page redirected to a holding host on three attempts by two separate readers, and its release calendar returns no table. The component set is corroborated — the statistical office’s own posted summary, seen through a search listing rather than fetched, carries the same four component rates as the article we read in full — but the core rate, the month-on-month 0.4 per cent and all country detail rest on one publisher.

It does not reconcile one small discrepancy, and we would rather name it than smooth it. Today’s reader puts September energy inflation at 18.9 per cent. The issuer’s own September flash, which we cited this morning, said 18.8. A tenth, in the prior, on the component carrying the whole headline move. We cannot settle which is right while the issuer’s site is down.

It does not tell you the ADP consensus, and we have deliberately not averaged the figures we have into one. Three numbers from three publishers of unequal quality is not a distribution.

It does not tell you the core consensus. One reader says core “also matched forecasts” without printing the figure it matched, so the claim that core was in line is that reader’s arithmetic, not ours.

And it does not tell you where euro–dollar goes into the 15:00 cut. The 1.1262 above carries no timestamp, the equity levels beside it are marked 11:20 CET, and by the time you read this both are history.

Related

• FX event calendar — the 13:15, 13:20, 13:30, 14:45, 15:00 and 15:30 UTC items, and the 16 to 17 December ECB meeting.
• Signals — sizing an event that has a time but no consensus field.
• EA presets — news-window settings for a release cluster with the option cut inside it.