Tonight's Producer Prices Ask for 0.3 on One Field and 0.6 on the Other. Only One Is Inside the Error Bar.
Japan's October producer prices land at 23:50 UTC tonight, and the two headline fields disagree about whether it is an event.
First a correction: this morning we published a 0.40-point mean absolute surprise for the year-on-year field, computed from six forecast-and-result pairs. The page carries nine 2026 rows and eight have forecasts. The full set gives 0.325 points, so our figure was 18.7% too wide, and tonight's asked-for 0.3-point move is 1.08 times the typical error rather than 1.33. The claim survives by eight hundredths of a point. Per field it splits: the year-on-year consensus of 6.5 against a 6.8 prior asks for a move inside the error bar, while the month-on-month +0.3 against a −0.3 prior asks for 0.6 points, 2.67 times its own 0.225-point error. One release, one minute, and a blackout sized by how much the number is supposed to move should wave one field through and not the other. The signs differ too: year-on-year is near unbiased at +0.05 signed, month-on-month runs −0.175 with five of eight prints below consensus. So we pre-register both, losably: the year-on-year lands between 6.17 and 6.83, and the month-on-month prints at or below +0.3. The 6.5 is now two-sourced; the +0.3 is not, and the second publisher's 6.7 is its own model rather than a survey, which is a different object and not a conflict. The release hour again comes from an aggregator while the issuer states no zone at all — a status field yesterday, a zone field today. And the slot built for the US data block has no US government release: the next is tomorrow's claims, the next from the statistics agency is Friday's producer prices.