The BOJ does not own the reserves and cannot decide to intervene. The dollars sit in the Ministry of Finance's Foreign Exchange Fund Special Account (外国為替資金特別会計, FEFSA). Article 7(3) of the Foreign Exchange and Foreign Trade Act vests intervention authority in the Minister of Finance. Article 40(2) of the Bank of Japan Act then forces any FX transaction aimed at exchange-rate stability to be conducted by the BOJ as agent under Article 36(1). The BOJ is the trading desk and the settlement agent. It is not the principal.

The BOJ does hold its own foreign currency assets — ¥11.98trn (≈$76bn) on 10 Aug 2026 — but under a separate Policy Board mandate for international cooperation and emergency FX liquidity to Japanese banks, capped at 5-year foreign government paper. That money is not the war chest.

The FEFSA is a levered carry trade, not a rainy-day fund. At 31 March 2025 it held ¥191.4trn of assets against ¥97.6trn of Financing Bills outstanding — roughly 51% debt-funded. It is structurally short yen and long dollars, financed in yen T-bills. Cumulative FX valuation gain through FY2024: ≈¥50.3trn. Yen-buying intervention is therefore deleveraging: sell the dollar asset, retire the yen bill.

Only about an eighth of the reserves is cash. End-July 2026: securities $927bn (72.0% of total), deposits $162bn (12.6%), gold $110bn, SDRs $61bn, IMF position $11bn. And 99.7% of the deposits sit with foreign central banks and the BIS — $432m with commercial banks, i.e. nothing. So intervention at scale is always a securities operation eventually.

The Fed swap line cannot fund this, and its being uncapped is why that matters. Swap dollars land on the BOJ's balance sheet, and the BOJ has no authority to intervene on its own account. The FOMC applies a liquidity purpose test to swaps and a separate disorderly-markets test to intervention. The BOJ's 2026 drawings have all been $1–6m — $2m on 7 July, $2m on 21 July — recurring readiness tests, not funding. The whole policy argument is about the capped facility (FIMA repo, $60bn) while the uncapped one sits idle.

"Japan drains the foreign repo pool" is mostly wrong now. It fitted 2022. It does not fit 2026: in May 2026 MOF's reserve securities fell $75.6bn while its deposits at central banks and the BIS were flat (+$39mn). Japan funded ~$75bn of intervention out of bills and coupons without touching its cash at all.

On 31 July 2026 the MOF bought yen in coordination with the US Treasury — the first US participation in yen-supporting intervention since 1998, executed under the September 2025 US–Japan Finance Ministers' Joint Statement. Minister Katayama's 3 August statement added the line that matters here: "Japan also plans to utilize the Federal Reserve's Foreign and International Monetary Authorities (FIMA) Repo Facility in the future."

As of the 5 August H.4.1 the facility was still drawn at $0. Secretary Bessent has publicly called for it to be "upsized". That is a request to change the plumbing described in §06.

Gold volume has been fixed at 27.20mn fine troy oz throughout — the entire rise from $49bn to $110bn is price, not buying. Source: MOF, released 7 Aug 2026.

Only the dollar-denominated part is intervention ammunition. Gold, SDRs and the IMF reserve position are on the template but are not sellable-in-a-morning dollar assets. That leaves $1,090bn of FX reserves, of which the genuinely near-cash tranche is much smaller.

Derived from the FY2024 FEFSA disclosure (31 Mar 2025, ¥ converted at ~156) — the only official maturity cut MOF publishes, once a year each November. The ≤1yr bucket jumped +7.2pp in FY2024, to 30.5% of the securities book: the portfolio was deliberately shortened. Independent corroboration: Goldman Sachs put cash and equivalents at ~$200bn (Aug 2026), on a narrower definition.

The 26% non-government bucket is not broken out. MOF's 2005 management policy names agencies, supranationals and ABS as eligible; it does not say how much is held. Currency composition is withheld by explicit standing policy, "because it could affect the FX market".

A Cabinet answer to a Diet question, 28 June 2011: the Fund holds foreign-currency securities 「米国債を中心に」 — centred on US Treasuries — and holdings by country of issue and trading plans are not disclosed "because of the risk of unforeseen effects on financial and FX markets."

The most recent answer, 23 December 2025, holds the same line, while adding that management is conducted 「我が国の判断により」 — at Japan's own judgement — an explicit rebuttal of the idea that the portfolio is run to please Washington.

Drawing down the foreign repo pool is not a Fed operation at all — it is Japan spending its own cash, which happens to be parked overnight with the Fed rather than in a bank. No approval, no disclosure, no stigma, no cap. It is the reason the pool is now 99.5% of the Fed's entire reverse-repo book: the domestic ON RRP has drained to $1.65bn while the foreign pool sits at $318bn.

Fed staff wrote the answer down. New York Fed Staff Report 983 (2021) states that FIMA repo proceeds are "generally expected to result in transfer of dollars to local institutions with funding needs and/or fund FX spot interventions" — with footnote 17 constraining the direction: "FX intervention funded in this way would only be in the direction of currency sales to strengthen the domestic currency vis-à-vis the U.S. dollar." That is exactly yen-buying.

There is no contractual bar — the 2014 USD–Yen swap agreement contains only a recital about market functioning. The block is institutional: swap dollars land at the BOJ, which has no legal authority to intervene on its own account; intervention is MOF money in the FEFSA. The 1962–98 network was the intervention network, but the causality runs the other way from the usual telling: the FOMC eliminated the standing facilities on 17 November 1998 "because of their long disuse" and ahead of the euro, participants agreeing the lines were "no longer needed for their traditional use, exchange-market intervention." Demand had evaporated; the channel was not closed to stop it. Rebuilt in 2007 on a lender-of-last-resort rationale.

If Washington joins in — as it did on 31 July 2026 — it fires from the Exchange Stabilization Fund, decided by the Treasury Secretary, executed by the NY Fed as fiscal agent. The ESF's FX book was $18.8bn at 30 June 2026 — $12.95bn of euro-denominated assets plus $5.79bn of yen-denominated assets, both figures in dollars. Warehousing with the Fed is capped at $5bn. The Fed's SOMA holds a deliberately matched book, $19.1bn at 31 March 2026, but deploying it is an FOMC decision, not the Treasury's to command. So Treasury alone has ~$19bn; with the FOMC alongside, ~$38bn.

Either way an order of magnitude below what Japan deploys alone — which is why joint intervention is a signal about intent, not about size.

The G7's Bari formulation — reaffirmed by the G7 in Paris on 19 May 2026 — allows intervention to counter "excess volatility and disorderly movements in exchange rates". The operative bilateral document is the US–Japan Finance Ministers' Joint Statement of 11 September 2025, which restates that test and commits both sides to monthly public disclosure. The July 2026 Treasury FX Report had already recorded Japan as intervening in 0 of 12 months and "exceptionally transparent" — the pre-condition for a joint operation being read as cooperative rather than manipulative.

Not because the series agree precisely — two of the three are TIC-sourced, and the TIC-implied figure is ~$15bn short of the MOF-implied one. Because of what the last two lines say:

The decisive lines are the last two: MOF's own securities fell by almost exactly the intervention size while its deposits did not move. Bills then locate which securities — Japan's short-term holdings fell $59.8bn against a $61.0bn fall in the global foreign-official aggregate. Caveat: those are different universes (Japan = all holders; the aggregate = all countries, official only), so "98% of it" is a coincidence of magnitudes, not an identity. What it does establish is that no other reserve manager was selling bills in size that month. Reading: ~80% funded out of bills, the balance out of coupons, cash untouched. Inference, but tight.

TIC Table 3, US$bn. Bills are 8–12% of Japan's total UST holdings but absorb almost all the intervention. That is the point: the coupon book is long-duration and sitting on a 4.6% unrealised loss, so you sell the bills.

H.4.1 Wednesday levels, US$bn. The −$36.2bn fall over the three weeks to 5 Aug 2026 is the 6th largest in 602 observations since 2015. Caveat: the −$17.9bn week to 29 July pre-dates the 31 July operation, MOF's end-July deposits were flat, and the pool is never disaggregated by country. It cannot be attributed to Japan from public data. Same problem in reverse in Apr–May 2026: the pool fell $22.9bn over the month while Japan's own deposits were flat, so most of that decline belongs to somebody else.

MOF template, US$bn. Securities do all the moving; the deposit line is remarkably inert across four intervention episodes — $135bn in 2021, $162bn now. Japan does not run its cash balance up before intervening.