Japan Spent a Record $98.6 Billion Defending the Yen. It Sold US Treasuries to Do It — and Two Weeks Later, the Yen Was Back Where It Started.

The Ministry of Finance's own monthly release confirms 30 July-26 August 2026 as the period covered by its most recent foreign-exchange intervention report — the period in which Japan's spending hit a record. Analysts Atsushi Takeda (Itochu Research Institute) and Prashant Newnaha (TD Securities), cited in reporting on the Treasury-holdings drop, both concluded Japan 'most likely' sold US Treasuries — and specifically short-dated ones, which are more liquid and carry less risk of disturbing long-term yields, a distinction Newnaha called 'probable' given their liquidity advantages. The $87.8bn fall in foreign-securities holdings tracks closely with the scale of the intervention itself, which is the evidence for the sourcing claim rather than an official confirmation — Tokyo does not publish which securities it sells. OMFIF's analysis frames the US Treasury's direct participation as unusual: the US has intervened in currency markets only twice this century before this, in 2000 (buying euros with G7 partners) and 2011 (selling yen after the Fukushima disaster), both times as 50/50 Fed-Treasury operations alongside collective G7 action — a genuinely multilateral posture this year's unilateral US-Japan pairing does not share. By 1 September, with the Bank of Japan's policy rate still at 1.0% against a Federal Reserve target range of 3.5-3.75%, USD/JPY was trading in the 159.98-160.00 range — evidence, per one analysis of the intervention's durability, that 'repeated intervention with only brief currency relief would suggest that policy is fighting — rather than reversing — the underlying rate differential.'

OMFIF's analysis is explicit that the yen's weakness reflects 'doubts about Japan's macroeconomic policy' rather than a one-off shock: the Bank of Japan has maintained accommodative policy despite inflation, and the resulting rate differential against the Federal Reserve is what fuels carry-trade activity, where investors borrow cheap yen to fund higher-yielding positions elsewhere. US Treasury Secretary Bessent's participation is read in the same analysis as motivated partly by contagion risk — 'a weakening yen could drag down other Asian currencies' — and partly by a more self-interested concern: that continued yen selling would eventually require Japan to sell US government bonds to fund it, exactly the outcome that materialized in August. The Savior Wealth analysis frames the September data point starkly: with the BOJ still at 1.0% and the Fed at 3.50-3.75%, intervention is 'fighting — rather than reversing' the rate gap, meaning the underlying cause was never addressed by the August spending, only its symptom. Likely rather than confirmed, because neither government has stated the intervention's precise decision threshold or trigger point in the reporting available — the rate-differential explanation is the analysts' read of the pattern, not an official rationale.

The Federal Reserve's expansion of its Foreign and International Monetary Authorities (FIMA) Repo Facility, noted in the Savior Wealth analysis as part of this episode, gives Japan (and other central banks) a way to raise dollars against Treasury collateral without selling the underlying bonds outright — precisely the tool that would prevent a repeat of August's $87.8bn holdings drop. If it is used before the next intervention, both central banks benefit: Japan avoids further balance-sheet erosion and the US avoids the yield pressure that comes with a large sudden seller entering the Treasury market. On the market side, the same analysis documents heavy speculative positioning — net short yen futures at 63,298 contracts as of 25 August — meaning leveraged funds with short-yen exposure had unrealized losses that a stronger yen (even briefly) let some close out or hedge. Uncertain, because no source here documents actual trading behavior confirming funds exploited the window, only the positioning data that makes it plausible.

The $87.8bn reduction in Japan's foreign-securities holdings is the most direct, measured cost of this episode — money spent for an effect that, per the September data, had substantially faded within about two weeks. That is a genuinely poor trade in isolation: a record monthly spend that did not durably move the exchange rate the underlying rate differential still explains. The Savior Wealth analysis's carry-trade section adds a second, more speculative loss scenario: it lists the August 2024 unwind as a precedent — a episode in which the TOPIX fell 12% in a single day, the S&P 500 fell 3%, the VIX traded above 60, and Bitcoin and Ethereum fell as much as 20% — and notes that net short yen positioning (63,298 contracts as of 25 August 2026) remains large enough that a disorderly unwind, though smaller than 2024's, 'remains possible.' Rated likely rather than confirmed because the Treasury-holdings loss is measured and reported directly, while the carry-trade unwind risk is a documented possibility in the cited analysis, not an event that has actually occurred in this cycle.