Japan just showed you how a reserve event gets managed
Deep dive · Wednesday, 5 August 2026
The interesting thing about 31 July is not the level of the yen.
It is that two treasuries demonstrated, in public, a way to defend a currency that does not require the defending country to liquidate its holdings of another country's debt — and a way for the United States to support a partner currency without saying anything at all about the dollar.
If you run a long-horizon portfolio, that mechanism is worth more of your attention than the exchange rate it produced.
What actually happened, and who did it
On Friday 31 July, in U.S. Eastern time, Japan's Ministry of Finance purchased yen in coordination with the U.S. Department of the Treasury. The statement came from Minister of Finance Katayama Satsuki on 3 August, and it is worth quoting because the wording is careful: Japan "will not hesitate to conduct further joint intervention," and it "plans to utilize the Federal Reserve's Foreign and International Monetary Authorities (FIMA) Repo Facility in the future." (Japan Ministry of Finance, 3 August 2026)
Two things about that sentence get reported wrongly almost every time, and both matter.
First, it was the Ministry of Finance, not the Bank of Japan. Under Article 7, Section 3 of the Foreign Exchange and Foreign Trade Act, the Finance Minister is the authorising officer. The Bank of Japan's own published outline is explicit: the Bank, "as the agent of the Minister of Finance, executes foreign exchange interventions in accordance with the directions of the Minister of Finance by using the Foreign Exchange Fund Special Account," and "the Ministry of Finance officially announces the total amount of interventions." (Bank of Japan) There is no such thing as a "Bank of Japan intervention volume." The money is the government's, the decision is the Finance Minister's, and the disclosure is the Ministry's.
Second, nobody knows how big it was. The Ministry's most recent published intervention report covers 29 June to 29 July and records a total of ¥0. The 30–31 July operations fall in the next reporting window — 30 July to 27 August — which is not published until the end of this month. (MOF intervention report, 31 July 2026) The figure in circulation, roughly ¥8.45tn or about $53bn, is Bloomberg's inference from differencing the Bank of Japan's daily current-account projection against money-broker forecasts. It is a reasonable technique and it is not a disclosure. Anyone printing it as an official number is printing a reporter's arithmetic.
There is one further caveat that the market has quietly skipped over. The euro-funding leg — the detail that made this interesting to strategists, because it implies Washington sold euros rather than creating dollars — is reported, not confirmed. The Ministry's statement does not mention euros. Treasury issued no press release. Secretary Bessent confirmed only that "Friday's coordinated foreign exchange actions countered disorderly yen movements." The currency composition traces to Financial Times reporting. Treat it as credible reporting and label it as such.
Rarity is the signal
Joint U.S.–Japan intervention has happened three times in twenty-eight years.
On 17 June 1998, U.S. monetary authorities sold $833m against the yen — buying yen — split evenly between the Exchange Stabilization Fund and the Federal Reserve System, "in cooperation with the Japanese monetary authorities." On 18 March 2011, in the opposite direction, the U.S. purchased $1bn against the yen as part of a G7-coordinated operation to stop the post-Tohoku appreciation. And now 31 July 2026. (Federal Reserve Bank of New York, Treasury and Federal Reserve Foreign Exchange Operations, 1998 Q2 · 2011 Q1)
Every quarterly report between those dates says the same thing: the Federal Reserve and U.S. Treasury did not intervene. Japan acted alone in 2022 and again in 2024; Washington stayed out both times.
Twice in twenty-eight years the object was to rescue a yen that had fallen too fast. Once it was to stop one that had risen. That is the base rate you should hold in mind when you hear that intervention is "back": it is an instrument of last resort, and it has just been used at a level — around ¥164 in late July — that no post-war Japanese government has previously had to defend.
The plumbing item, which is the real story
Read the Ministry's statement again and notice what it flags for the future: the Fed's FIMA repo facility. Secretary Bessent has publicly urged that it be "upsized."
FIMA lets a foreign central bank raise dollars against its holdings of U.S. Treasuries, overnight, rather than selling them into the market. If you own long-dated U.S. government debt — and every large pension and sovereign fund does — this is the difference between a Japanese currency defence that is invisible to you and one that makes Japan a forced seller of your asset class.
Japan is the largest foreign holder of U.S. Treasuries. The historic fear around a yen defence has always been the second-order effect: Tokyo needs dollars, Tokyo sells Treasuries, U.S. long rates back up, and a currency problem becomes a global duration problem. FIMA is the pipe built specifically to stop that transmission. Japan naming it, and the U.S. Treasury Secretary calling for it to be larger, is the two governments telling the bond market in advance that this defence will not be funded by liquidating the bond market.
That is a more consequential disclosure than the size of the trade.
The disclosure gap
There is an irony worth sitting with. In their Joint Statement of 11 September 2025, the two finance ministries committed to disclosing any FX intervention "on at least a monthly basis," and reserved intervention for "combatting excess volatility and disorderly movements" — language that explicitly covers "excessively volatile or disorderly depreciation or appreciation."
Monthly disclosure sounded like transparency when it was written. In practice it means that a market-moving operation conducted on 31 July will not have an official number attached to it until the end of August. For four weeks, every position taken on the scale of the intervention is a position taken on inference.
Meanwhile the Bank of Japan is holding its policy rate at 1.00% — the highest since 1995 — while projecting core inflation "clearly above" 2% from the second half of this fiscal year. (Bank of Japan) A currency defended by intervention rather than by rates leaves the rate risk entirely intact. What intervention buys is time. What it does not buy is a resolution.
What breaks it
Both finance ministries have now pre-committed to repeat. That is deliberate — the deterrent only works if it is believed — but it has a cost. An intervention that works becomes an expectation, and an expectation becomes a floor that speculators will eventually test, because testing it is how you find out whether it is real. The 1998 and 2011 operations both worked. Neither prevented the yen from going where the rate differential eventually took it.
The observable that would tell you the regime is failing is not the exchange rate. It is a second operation inside a short window, or a widening gap between the estimated size of an operation and the market's willingness to respect it.
>
You hold this event three times over, and the three legs do not net.
>
(a) Japanese duration. The Bank of Japan is on hold at 1.00% with inflation projected clearly
above target. A currency defended by intervention rather than by rates leaves that rate risk sitting
in your JGB and global-aggregate sleeves untouched.
>
(b) U.S. Treasuries. The reason FIMA matters to you is that it is the difference between Japan
raising dollars against your bond and Japan selling it. If you hold long U.S. duration, the size of
that facility is a more relevant number than USD/JPY.
>
(c) Unhedged international equity. A yen that travels from 164 to 156 and back toward 158 inside
a week is a translation shock inside your MSCI ex-US sleeve that has nothing whatsoever to do with
the underlying businesses you thought you were buying.
>
The practical tilt. The question to put to the investment team this week is not "where is
USD/JPY." It is: **what is our hedge-ratio policy actually triggered by, and does it fire on
realised volatility?** An intervention regime *suppresses realised volatility while raising the
tail* — that is precisely what it is designed to do. A hedging policy calibrated on trailing
realised vol will therefore reduce hedges exactly as the tail is growing. You do not need a view on
the yen for that to be worth checking. You need only to know which input your policy reads.
>
⚠️ Embedded-exposure flag. If today's Danube story has you looking again at regulated utilities
or listed global infrastructure, check the gas weight in the index before you act. Listed
infrastructure and many "climate solutions" benchmarks carry meaningful embedded natural-gas
exposure through midstream and gas-fired generation constituents. A water-constrained thermal and
nuclear fleet is an argument about cooling, not an argument for whatever the index happens to hold.
Source ledger
- Coordinated purchase, MOF statement, and FIMA intention — Japan Ministry of Finance, 3 August 2026.
- MOF/BOJ division of authority; monthly disclosure convention — Bank of Japan, *Outline of the Bank
- ¥0 recorded for 29 June – 29 July; next window 30 July – 27 August — MOF intervention report,
- 1998 and 2011 joint operations, sizes and directions — Federal Reserve Bank of New York, *Treasury
- Intervention purpose language and monthly-disclosure commitment — U.S.–Japan Finance Ministers'
- Policy rate at 1.00% and the inflation projection — Bank of Japan, July 2026 statement.
- The ~¥8.45tn figure is a Bloomberg estimate derived from BOJ current-account data, not an official
of Japan's Foreign Exchange Intervention Operations*.
31 July 2026.
and Federal Reserve Foreign Exchange Operations*, 1998 Q2 and 2011 Q1.
Joint Statement, 11 September 2025.
disclosure. The euro-funding leg is Financial Times reporting, not a Treasury confirmation. Both are labelled as such above.
Primary or first-party sources where available; press reporting is identified as reporting. This is editorial analysis for institutional readers and is not investment advice.