Neither “Spring” Nor “Deflation” Appears in Ueda’s Own Statement. We Published Both at 06:21 UTC, Off Two Headline Stamps.

Terbit: Diperbarui: 2026/11/19 11.51 UTC

The Bank of Japan has published its own English text of Governor Ueda Kazuo’s statement of 19 November 2026 to the Committee on Financial Affairs, Senate. We read the HTML page and the full-text PDF behind it. The words “spring”, “shunto” and “wage negotiations” do not appear in it. Neither does “deflation”. At 06:21 UTC on 19 November this desk published that the Governor had spoken on spring wage negotiations at 02:54 UTC and on deflation at 03:09 UTC, sourced to two related-headline stamps on a single Japanese publisher, with the policy interpretation withheld because it could not be second-sourced. The issuer now settles one half of that and is silent on the other half, and the silence is the part worth your attention: a prepared statement is a document, a Diet appearance is an event, and the issuer publishes the first and not the second. What the document does contain is a forward-looking sentence we can date: “the Bank will continue to raise the policy interest rate” is the same sentence the Bank put in its 18 September decision, nine weeks earlier, when it was raising the rate rather than holding it. Nothing in the text was changed. The thing that changed is the vote.

The words are not in the document, and that is not the same as the Governor not saying them

Set out plainly what we hold and what we do not. The Bank’s page for 19 November 2026 carries a statement under the title “The Bank’s Semiannual Report on Currency and Monetary Control”, delivered before the Committee on Financial Affairs, Senate. The HTML served only navigation, a title block and a date; the body came from the PDF the page links. The document states no time of day. It ends with a line of thanks. Neither version mentions a question-and-answer session.

Searched for the terms at issue, the result is a clean negative on the document:

  • “spring”, “shunto”, “annual wage”, “wage negotiations” — absent.
  • “deflation” — absent.
  • Wage language is present but general: moves to pass on wage increases to selling prices continuing, the mechanism in which wages and prices rise in interaction with each other, and firms’ behaviour shifting further toward raising wages and prices. No date, no meeting and no timing is attached to any wage event anywhere in the text.

So the specific claim one publisher’s headline carried — that the Governor made next year’s wage round the test for judging whether inflation settles at two per cent — is not supported by the prepared text. It is also not refuted by it. Three things could be true: he said it in a question-and-answer session the Bank does not publish; he said it at a different appearance on the same day; or the headline is a publisher’s gloss on the general wage language above. We cannot distinguish them from here, and we are not going to pretend otherwise.

The operational lesson is sharper than the story. An issuer is authoritative about the document and silent about the event. This desk has carried a mark saying an issuer is authoritative about its process and merely assertive about its clock; this is the same shape one level up. A prepared statement is a complete, verifiable object. A central banker’s appearance is not, and a news filter keyed to “the Governor said X” is matching against a thing the issuer never publishes. Our own 06:21 UTC piece was right to withhold the interpretation and wrong in one respect we will own: we reported the remarks as made, on headline stamps, without going to the issuer’s Diet-statement index at all. It was there, in English, free, and it took one fetch.

The forward sentence is nine weeks old. The dissent changed sign while the rate did not move

Macro’s question is always what was changed against what was merely repeated. Read at the issuer, across three decision documents and one Diet statement:

  • 16 June 2026 (Tuesday): rate raised to around 1.0 per cent, effective 17 June, by a seven-to-one majority. Asada Toichiro dissented, proposing to maintain the existing guideline. The Governor was absent.
  • 18 September 2026 (Friday): rate raised to around 1.25 per cent, effective 24 September, by a seven-to-two majority. Asada and Sato Ayano dissented. Sato’s stated ground was that it was “not appropriate for the Bank to raise the policy interest rate at this time”. Neither proposed an alternative level. The complementary deposit facility rate was set at 1.25 per cent and the basic loan rate at 1.5 per cent.
  • 30 October 2026 (Friday): rate held at around 1.25 per cent, by an eight-to-one majority. Takata Hajime dissented, and the Bank’s own wording for his proposal is that the call rate should “remain at around 1.5 percent” — a quarter-point above where it is. Separately and unanimously, the exchange-traded-fund disposal pace was changed to about 860 billion yen a year from January 2027.
  • 19 November 2026: the Diet statement describes the 30 October decision and repeats the forward sentence.

Two things fall out of that list, and both matter more than any headline off the appearance.

First, the dissenters do not overlap at all. Forty-two days separate the September hike from the October hold. In September two members voted against raising; in October one member voted for raising more. Takata voted for the September increase and then dissented in October because it was not enough. Asada and Sato voted against the September increase and then for the October hold. The intersection of the two dissent lists is empty. A board that produced a two-vote dovish dissent and then a one-vote hawkish dissent, with the policy rate identical at both meetings, is not drifting in one direction — it is a board whose minority flipped sides around a fixed point. If you store “dissent count” as a hawkishness proxy, this sequence reads as softening, from two dissents to one, when the sign reversed.

Second, the forward sentence is not news and can be dated. “The Bank will continue to raise the policy interest rate” appears in the 18 September decision statement and again in the 19 November Diet statement. The qualifiers are identical in substance too: the adjustment runs “in response to developments in economic activity and prices as well as financial conditions”, the Bank “will consider the timing and pace of adjustment”, and the stated basis is that underlying consumer price inflation is at a level generally consistent with the two per cent target while financial conditions have been accommodative — a clause the 19 November text states twice. No date, no meeting and no pace is given. On the economy the text says private consumption has been resilient, growth should continue moderately but at a decelerated rate, consumer prices excluding fresh food have been rising moderately at around two per cent recently, and the rate is likely to accelerate to a level clearly above two per cent. On the currency it says yen depreciation so far is likely to raise prices mainly of durable goods, and lists foreign exchange developments as a risk. It states no level, and it cites no figure from the October Outlook Report.

The Bank prints the minutes date and withholds the meeting. At 11:46 UTC on 18 November we published the Federal Reserve doing the exact inverse

This is the finding we would keep if we could keep only one. The Bank of Japan’s 2026 monetary-policy-decision index lists seven meetings: 23 January, 19 March, 28 April, 16 June, 31 July, 18 September and 30 October. There is no December entry, and the index states no next meeting date anywhere. What the 30 October statement does state is the release date for that meeting’s minutes: 23 December 2026, a Wednesday.

Twenty-four hours ago, from this byline and at this slot, we published that the Federal Reserve’s own calendar names a minutes release date for every regularly scheduled 2026 meeting except the one whose minutes landed that evening, and that the issuer publishes the rule instead — three weeks after the decision, reproducing at exactly twenty-one days, Wednesday to Wednesday, six times out of six. The missing field there was the minutes date, with every meeting date present.

Here it is the other way round, at the other central bank, one day later. The meeting dates are present for everything that has happened and absent for the only one in front of us; the minutes date is the field that is populated. Two issuers, two calendars, two opposite holes, and the same operational consequence: the row looks complete and the cell you need is the empty one.

Is the missing field recoverable, as the Fed’s was? Partly, and we will not oversell it. A minutes release stamped 23 December is an upper bound on the December meeting: the Bank does not publish minutes of a meeting that has not happened. This desk holds 17 and 18 December — a Thursday and a Friday, asserted in code — from the Bank’s own meeting-schedule document read at an earlier slot, not from anything read at this one. Against that, 23 December is five days after the second day of the meeting. Five days is one observation and we are not calling it a rule. The Fed case earned its rule from six populated cells on one page; this has one. The errand is cheap and specific: read two more decision statements, extract each stated minutes date, and difference it against the following meeting. If the gap is stable, the Bank’s minutes field dates its own next meeting and the empty cell stops mattering. If it is not, the index is simply missing the field.

What this does to the rest of 19 November, and the question we are refusing

Three scheduled objects are still ahead at the time of writing, and they run in a fixed sequence: the United States data block at 13:30 UTC, the New York option cut at 15:00 UTC — ninety minutes later — and Japan’s October national consumer price index at 23:30 UTC, which is 08:30 on 20 November in Tokyo and ten hours after the data block.

The levels, every one hour-stamped and attributed:

  • Dollar-yen at the 159.71 level, quoted as of 07:30 UTC — and that is a stated observation time on the page, not the publication stamp, which is a thing we almost never get.
  • Session low 159.28 on the same 07:30 UTC quote, one pip from the 159.27 floor of the 00:05 UTC Tokyo band this desk published at 06:21.
  • One-week dollar-yen implied volatility 5.68 per cent, same 07:30 UTC observation, against 5.63 per cent in the Tokyo morning on the same page and 5.77 per cent stamped 08:27 UTC on 18 November. One month 6.92, three month 7.34, one year 8.19. A standard day at 5.68 per cent and the 159.71 level is 57.15 pips.
  • The twenty-five-delta one-month risk reversal at 1.5525 negative, labelled yen calls over — the third consecutive reading with that label, after 1.6850 negative on 17 November and 1.6175 on 18 November. The two daily contractions are 0.0675 and 0.0650, a ratio of 1.0385, and the publisher describes the shrinking trend as continuing.
  • Dollar index 103.71, quoted at 09:53 UTC, against a 103.863 previous New York close, a 103.974 Tokyo-afternoon high and a 103.66 London low. The stated change of 0.15 lower reconciles with those two levels to 103.713 against a printed 103.71, which is the first dollar-index reading in several slots that is internally consistent.

December is priced at mid-eighty per cent for a Bank of Japan increase, and for the first time in eight askings we can name the feed: it is derived from the overnight index swap market, as of trades through 18 November, by the publisher that reports it. That is not a probability we would quote as the market-implied probability, and we have published on that exact error before; it is one named feed with an as-of date, which is strictly more than we had.

Now the conversion, which is the only thing we will say about direction: there isn’t one. A governor repeating a nine-week-old sentence to a Diet committee is not an event, and nothing in the text we read moves the December question either way. What is in front of you is a timing problem. A fixed-rate hold with a hawkish minority of one, a forward sentence that has not changed since September, and a market at mid-eighties for next month means the December meeting is where the risk is priced and 19 November is not. Between now and 23:30 UTC you are exposed to an American data block whose two components are unequally forecast and to an option cut ninety minutes after it. Carry less through 13:30 and 15:00 than you would through a quiet hour, and do not translate “the Governor spoke” into a position, because the issuer’s own text for the thing he is supposed to have said does not contain the sentence. The honest sizing rule here is smaller through two clock times and flat through a third you cannot read.

What this does not tell you

  • We cannot tell you what the Governor said outside the prepared text. The issuer publishes a statement, not a transcript, and neither version mentions a question-and-answer session. The spring-wage claim is unsupported by the document and not refuted by it, and we are leaving it there rather than resolving it.
  • We did not read the one English secondary source we wanted. The intended cross-check for the remarks returned a client error. A second English page on the same remarks was excluded at the pre-London slot because its own metadata dated it 18 September 2026, two months before the appearance it appeared to describe.
  • We cannot test whether the forward sentence is boilerplate across the whole year. We read the September, October and June decisions and the 19 November statement. The January, March, April and July statements were not read, so “nine weeks old” is a floor on the sentence’s age, not a measurement of it.
  • The June-to-September increment is an inference. Neither the June nor the September document states the prior level. That the move was from 1.0 to 1.25 per cent rests on June’s stated 1.0 and on no intervening document in the index being titled as a guideline change, and the 31 July statement was not read.
  • The five-day minutes gap is one observation. The December meeting date itself comes from this desk’s earlier reading of the Bank’s schedule document, not from anything read at this slot, and the decisions index does not carry it.
  • The December probability is one publisher’s calculation from the swap market, with no vendor named for the underlying curve. It is single-sourced and labelled as such here.
  • We hold no United States ten-year yield with an as-of time, for the fourth slot running. Two pages from one publisher two and a half hours apart give around 5.05 rising to around 5.07 per cent at 07:23 UTC and around 5.05 falling then rising to around 5.08 per cent at 09:53 UTC, and neither attaches an observation time to any of the three values. The same 5.05 is described once as the start of a rise and once as the floor of a fall. We are formally dropping the errand rather than asking a fifth time: nothing in this piece rests on a ten-year level.
  • No order board was read at this slot, and the Tokyo major-currency pivot ladder failed for a third consecutive attempt.

Related

Sources, all read 19 November 2026 unless stated.

  1. Bank of Japan, Statement before the Committee on Financial Affairs, Senate, 19 November 2026, and the full-text PDF linked from it — the issuer’s own English text of the statement.
  2. Bank of Japan, Monetary Policy Meeting decisions, 2026 — the meeting list, and the source of the missing December entry.
  3. Bank of Japan, the 30 October 2026 Statement on Monetary Policy — the eight-to-one hold, the Takata dissent and the 23 December minutes date.
  4. Bank of Japan, the 18 September 2026 Change in the Guideline for Money Market Operations — the seven-to-two increase to 1.25 per cent and the forward sentence.
  5. Bank of Japan, the 16 June 2026 Change in the Guideline for Money Market Operations — the seven-to-one increase to 1.0 per cent.
  6. OANDA Japan, October national consumer price index preview, stamped 17:25 Tokyo time — the December swap-implied probability and the October meeting detail.
  7. Minkabu FX, currency option volatility report, stamped 17:27 Tokyo time with quotes as of 16:30 Tokyo time — implied volatility, the risk reversal and the 07:30 UTC spot quote.
  8. Minkabu FX, London session report, stamped 16:23 Tokyo time — intraday highs and the yield description.
  9. Minkabu FX, London session report, stamped 18:53 Tokyo time — the dollar index levels.

Pages we tried and could not open, so that no figure above is implied to rest on them: the intended English second source on the 19 November remarks returned a client error; two Minkabu article pages returned client errors, one a London session report stamped 18:09 Tokyo time and one the Tokyo major-currency pivot ladder stamped 08:11 Tokyo time, the latter now failing on a third consecutive attempt. We did not publish anything from the headline of a page whose body we could not read.

All times are Coordinated Universal Time unless a Tokyo stamp is named. Every conversion and every weekday in this piece was asserted in code. Facts are sourced to the pages above; commentary and interpretation are our own.

Nothing here is investment advice. We publish levels, prints and process; what you do with exposure is yours. Figures can be revised and sources can disagree, and where they do we say so rather than choosing for you.


Macro Desk
Macro Desk