At 10:01 UTC this morning the Ministry of Finance published how much it spent defending the yen between 27 August and 28 September. The number is zero. This desk told you twice in the last twelve hours that the release lands at 10:00 UTC and that on our reading September’s defence was words rather than money. Both readings were right, and we are going to be precise about how much that is worth — because the Ministry still does not publish a release time, and as of eleven fifty this morning its own English page still does not carry the number at all.
Zero, and three wires had it inside thirteen minutes
The figure for the 27 August to 28 September window is nil. Newsquawk carried it at 10:01 UTC. DZH’s indicator wire carried it at 19:02 JST, which is 10:02 UTC. Minkabu carried it at 19:14 JST, 10:14 UTC. Three readers, thirteen minutes apart, one number, no disagreement about the period or the value. After a fortnight in which this desk has published vendor disputes about a consensus, about a prior, about which field a figure belongs to and—yesterday morning—about what a printed actual even was, a release that every reader agrees on is worth naming as such.
It is also the easiest number in the world to agree on. Zero has no decimal places and no seasonal adjustment.
The 10:00 convention survives, and it is still a convention
This desk printed 10:00 UTC for this release four times before going to the Ministry to check it. We reported last night that the Ministry publishes no time — not on the page describing the disclosure schedule, not anywhere we could find — and that our 10:00 was a convention corroborated by a single calendar. That has not changed. What we now have is two independent timestamps on the wire at 10:01 and 10:02 UTC.
Treat that correctly. Two wires timestamping their own posts one minute apart is strong evidence that the release landed at or just before 10:01 UTC today. It is not evidence that the Ministry has a published 10:00 convention, because the Ministry has not said so and a wire timestamp is the wire’s, not the issuer’s. If you are building a news filter around this release once a month, the honest setting is a window that opens before 10:00 and stays open, not a spike at a minute the issuer never promised. One corroborated instance is one instance.
The Ministry’s own English page still does not have it
Here is the finding that matters more than the zero. This desk goes to the issuing agency whenever a figure is load-bearing; that is a standing rule here and we have written it into print repeatedly. Applied to this release, at 10:01 UTC this morning, it would have returned nothing.
At roughly 11:50 UTC — an hour and fifty minutes after three wires carried the number — the Ministry’s English monthly index still lists the 30 July to 26 August window as its most recent entry. The URL where the September report will eventually sit, following the Ministry’s own unbroken naming pattern of a publication date followed by the letter e, returns a 404. The Japanese landing page does carry the schedule and confirms the 30 September release date; it does not carry the figure.
So on this release, for at least the first hour and fifty minutes, the primary source was the last place the number existed. That is the exact inverse of the failure this desk published on 28 and 29 September, when a US print existed only at the issuing bank and no aggregator had it. Both failures are the same underlying problem wearing different clothes: there is no source that is reliably first, and a rule that says “always go to the issuer” is a rule about accuracy, not about speed. Use the issuer to settle a number. Do not use it to learn one.
The previous figure checks out exactly, to the hundred million
While the September page was missing, the August one was there, and it lets us audit ourselves. This desk published last night that Japan spent a record 15 trillion 399.3 billion yen between 30 July and 26 August. The Ministry’s own English report for that window gives 153,993 hundred millions of yen. That is the same number to the last digit.
We flag this because it is the first time in two weeks that a figure this desk carried from secondary readers has been settled at the issuing agency and come back exactly right. The recent record runs the other way: a Canadian print where the issuer was unreachable, a Japanese industrial production year-on-year where four publishers gave two values 5.6 points apart, a consumer confidence prior revised out from under a published finding. Getting one clean confirmation does not retire the caution. It does tell you the secondary chain on Japanese intervention data is tight, which is useful, because that chain is the only one you will have on the day.
The day-by-day data is five weeks away, and that is the number you actually want
The zero settles September. It does not settle August, because a single monthly aggregate cannot tell you which days a record was spent on. The Ministry’s schedule page states when it will: the daily-basis data for the July to September quarter publishes between 2 and 9 November. That is primary-sourced and it is a date almost nobody has on a calendar.
Put the two facts together and the shape of what you are working with becomes clear. For the 15.4 trillion yen already spent, you will not know the daily distribution until early November — roughly fourteen weeks after the first of those days. For 29 and 30 September, the two sessions in which officials have talked hardest, you have nothing at all until the next monthly release at the end of October. The zero published this morning is the most authoritative flow number available to you today and it stops covering the market two days before you read it.
That is not a complaint about the Ministry, which publishes on a schedule it states and meets. It is a statement about what you can size on. If a rule in your system is conditioned on official intervention, it is conditioned on data that arrives a month late in aggregate and a quarter late in detail. There is no configuration of that rule that makes it timely.
What this does not tell you
It does not tell you the Ministry will not intervene. A zero for the month to 28 September is a fact about the past and the Minister and the currency diplomat have both been talking this week; a defence that costs nothing is compatible with one that costs a great deal the following day.
It does not tell you the release time is 10:00 UTC. It tells you that today, two wires timestamped it at 10:01 and 10:02.
It does not tell you why the English page lagged. We observed the lag; we did not diagnose it, and it may well have been filled within minutes of our looking. We have said when we looked so you can judge the claim.
And it does not tell you where spot goes into the New York data block. We have not put a directional view in this article and you should be suspicious of one built on a zero.
