Japan Spent 7,975.6 Billion Yen in One Day. No Calendar We Read Carried the Disclosure.

The figures, from four readers, and the one page we could not open

The daily amounts, as printed by the news agency whose copy carries them to the yen:

• Thursday 30 July: 7,975.6 billion yen — the record single day.
• Friday 31 July: 5,552.3 billion yen.
• Monday 3 August: 1,871.4 billion yen.
• Three-day total: 15,399.3 billion yen.

We added the three days ourselves and they sum to the published total exactly, with no residual. That matters more than it sounds: it tells you the three disclosed days are the whole operation for that window rather than the three largest of a longer series. The previous record was 6,278.7 billion yen, bought on Thursday 30 April this year. The new one is 27.0 per cent larger — 1,696.9 billion yen more in one session — and that is our arithmetic on the two published figures.

Four independent pages carry this and they do not conflict on a single number. A state news agency gives 7.98 trillion and 15.4 trillion, rounded; a Japanese wire service gives all four figures to the hundred million yen; a Japanese public broadcaster gives the 30 July figure to the hundred million yen as a breaking item; a second file from the same state agency repeats the headline pair. The rounded and unrounded versions are consistent with each other. The data is the July–September quarter disclosure, which is why a July operation becomes news in November.

We could not open the Ministry of Finance’s own intervention pages. Four separate paths, English and Japanese, all failed outright. So every figure above is a secondary reading of the issuing agency, including the ones that agree to the hundred million yen, and the reconciliation we ran is a reconciliation of wire copy rather than of the issuer’s table. We would rather tell you that than imply we read the source.

Fifty billion dollars is the wrong number, and the same article says why

One of the four readers converts the record day to “about 50.4 billion U.S. dollars”. Divide 7,975.6 billion yen by 50.4 billion and you get 158.25 yen to the dollar. That is where the pair is trading this morning. It is not where it was trading on 30 July.

We know that because the same article tells us. Three paragraphs on, it says the yen had slid to the upper 163 range in late July, which it calls the weakest in around forty years. Convert the record day at 163.50 and it is 48.78 billion dollars; at 164.00 it is 48.63 billion. So the published dollar figure overstates the operation by roughly 1.6 billion dollars, about 3.3 per cent, because a July yen amount was converted at a November exchange rate. This desk has a standing category for a number that is arithmetically fine and attached to the wrong vintage, and this is a clean instance of it — made cleaner by the fact that the contradicting evidence is in the same piece, a few lines below the figure.

The practical form of this is dull and worth saying anyway. If anything you run stores intervention size in dollars, it is storing a quantity that depends on when somebody did the division. Store it in yen, which is the unit the Ministry publishes and the unit the operation was actually denominated in, and convert at the date of the operation if you need dollars at all. We are not going to publish a dollar figure of our own for the three-day total, because doing so would require us to pick a rate for three days that spanned a 163 handle and a 155 handle, and any single rate we chose would be the same error in a different direction.

What 15,399.3 billion yen bought, measured from here

The level history, from the one reader that gives it, is a sequence of ranges rather than figures, so treat these as approximations off range descriptions rather than marks: upper 163 in late July when the operation began, temporarily the lower 155 range on 3 August, briefly the 152 range on 8 September, and the 158 range recently. Dollar–yen was 157.994 at 07:48 UTC this morning on an independent publisher’s hour-stamped quote.

So the operation moved the pair roughly 830 pips over its own four days, and the yen went on to a best level around the 152 range five weeks later. From that level the pair has given back something like 550 pips, and from the 3 August level around 280 pips. The honest summary is that 15,399.3 billion yen was spent defending a level the market has since walked back through, and the yen is now closer to where the intervention started than to where it ended.

That is not an argument that the intervention failed. An operation that stops a disorderly forty-year low has done something even if the level does not hold, and we have no counterfactual. It is an argument about what the disclosure is worth to you now, which is a different question with a cleaner answer: it tells you the size of the facility and the willingness, not the level.

Your intervention threshold is probably 450 pips too low

Here is the part that should change something. One Tokyo publisher wrote this morning that above 159 yen, Japanese authorities’ intervention is a concern, and that this limits the upside. That is a reasonable-sounding sentence and it is the sentence a great many news filters are built on, because 159 is a round number near spot and it has been the ambient threshold in Tokyo commentary for weeks.

The disclosure says the authorities committed 15,399.3 billion yen with the pair in the upper 163 range. That is about 450 pips above the 159 threshold. Nothing in the data released today shows an operation anywhere near 159. On the contrary, the prior record of 6,278.7 billion yen is dated 30 April, and we have no level for that day, so the shallowest intervention point we can evidence from the issuer’s own disclosed days is the upper 163 area.

If you run an exposure rule that goes flat, halves size, or widens stops on approach to an intervention level, and that level is 159, you are paying for insurance against an event the disclosed record places four and a half yen higher. That is a real cost: 159 is inside the range the pair has traded all of the last three weeks, so the rule fires constantly. The reverse error is worse and also available — concluding from this that intervention cannot happen below 163 — so we will say plainly that the data constrains where it has happened and says nothing about where it will. What it does support is this: a threshold should be set from disclosed operations, and until today you had no disclosed operation for this cycle to set one from. Now you have three days of them.

One claim we are flagging rather than relying on. A single one of the four readers says the 31 July operation was a joint Japan–United States intervention, the first coordinated move in twenty-eight years. The other three do not mention the United States at all. If true it is the most important sentence published today, because a coordinated operation and a unilateral one are different objects with different credibility. It is single-sourced and we are not building anything on it.

What is still in front of you today

The rest of the day, in UTC, with the speaker times confirmed on two independent calendars: Bowman at 13:40, Goolsbee at 15:00, Cipollone of the European Central Bank at 16:00, the three- and six-month bill auctions at 16:30 against 4.040 and 4.150 per cent priors, Barkin at 17:15, Bailey of the Bank of England at 17:25, and the three-year note auction at 18:00 for 58 billion dollars. Goolsbee’s 15:00 is now on a third independent calendar, which closes an item this desk opened last night and corrected this morning.

Then at 23:50 UTC tonight the Bank of Japan publishes the Summary of Opinions from its 29–30 October meeting, and — this is new, and our own week-ahead table missed it — Japan’s September trade balance lands in the same minute. Two Japanese releases at 23:50, which is one chance to react and not two, twenty minutes before the Tokyo window opens.

The sequencing is the thing. A disclosure that the authorities will spend 15.4 trillion yen in three days, landing the same day as the first Bank of Japan minutes since the October meeting, with a December hike argument live in both directions, is a configuration where the yen’s risk is genuinely two-sided for the first time in a while. Tomorrow’s US consumer price index at 13:30 is still the week’s information event, and our companion piece this morning argued it will arrive without the field most people are waiting for. Today the yen got a different kind of information: not a forecast about policy, but an audited number for what the official sector actually did.

What this does not tell you

We did not read the Ministry of Finance’s own publication. Four attempts at its intervention pages failed, so the daily amounts, the total, the prior record and the quarter the data covers are all secondary readings, however well they agree. We have no settlement or value dates for the operations, no breakdown by counterparty or venue, and no statement from any named official — none of the four pages quotes one.

We cannot tell you the level at which the previous record of 30 April was executed, so the comparison in the fourth section is a comparison of sizes and not of levels. We cannot confirm the joint-intervention claim, and we cannot tell you whether there were intervention days in the quarter with amounts too small to make the wire summaries, because we did not see the table.

The level sequence — upper 163, lower 155, the 152 range — comes from one page and is given as ranges. Every pip distance we computed from it inherits that imprecision, which is why we wrote “roughly” and “something like” rather than giving you a number to two decimals. The 159 threshold we argue against is one publisher’s characterisation of market concern, not a measured level and not anybody’s official statement.

And we have no reaction. Nothing we read connects this morning’s dollar–yen move to the disclosure, and we are not going to assert a link: the pair was already lower on other grounds, and a quarterly data release about July is exactly the sort of thing a market can read and ignore. If you want the price action, our companion piece this slot has it with the option board.

Related

• Economic calendar and event windows
• Signals
• EA presets