Half the ¥13.8trn Yen Rally Is Gone (August 2026): Goldman Puts Tokyo's Ready Cash at $200bn — and a 65% Chance the BoJ Hikes in September

Two weeks after the first joint US-Japan yen-buying operation since 1998, about half of what it bought has been handed back. The yen reached 155.20 per dollar on Monday 3 August, weakened past 159 on 11 August, approached 160 on Wednesday 12 August, and was trading either side of 159 on Friday 14 August — roughly 3.95 of the 8.79 yen won from July's four-decade low of 163.99, surrendered without a single change in the thing that caused the weakness. The genuinely new information is not the decay, which was the base case. It is that Tokyo's firepower has now been measured and is not the constraint: Goldman Sachs Research puts Japan's ready cash at about $200bn of a roughly $1trn reserve, with the Fed's FIMA facility able to make the rest liquid. What has not moved is the thing that sets the level — a Federal Reserve target range of 3.50%-3.75% against a Bank of Japan policy rate of 1.00% — and the market now prices a 65% chance the BoJ starts closing that gap on 18 September.

That combination is what makes this a mechanism story rather than a scoreboard. It does not prove intervention failed — a disorderly move was interrupted, and that was the stated objective. It proves something narrower and more useful: a currency's level is a claim on a yield differential, and buying the currency does not buy the differential. When the money is demonstrably there and the level still leaks, the constraint has been located, and it is not in Tokyo's vault. A chart shows you a violent move and an intervention headline; it cannot tell you whether the level holds. Fundamentals connect the cause — the rate differential, one of the five factors Pip Theory scores — to the currency.

Two documents published since the operation carry more forward content than any intervention total. The first is Goldman's estimate of what Japan can still spend, which answers the question the market had been guessing at. The second is the Bank of Japan's own July Summary of Opinions, released on 10 August, in which the Board says the hike pace could run "faster than market expectations" and names the weak yen as one of the reasons.

What actually happened: the rally decayed on schedule

The two weeks after the intervention are the cleanest natural experiment the yen has offered in years, because only one variable was allowed to change.

Read down the right-hand column and the mechanism states itself. Across those two weeks there was no policy change on either side, no confirmed second operation, and no reversal of the conditions that took the yen to 163.99 in the first place. What there was, on each of those days, was the same interest-rate differential paying the same people the same amount to do the same thing: late on 12 August the 10-year Treasury yielded 4.690% against 2.839% on the 10-year Japanese government bond, a gap of more than 1.8 percentage points on the long end to go with roughly 2.5 on the policy rate. The yen gave back about 3.95 of its 8.79-yen rally in the ordinary course of business rather than in response to any new event.

The pace is the only thing the coordination bought. The unilateral 30 July operation surrendered close to half its move inside a day; the confirmed joint one took a fortnight to reach the same place. That is a real difference — it is the difference between a headline and a fortnight of breathing room — but it is a difference in half-life, not in destination.

What pushed the dollar higher over the first of those two weeks was mostly external. Traders were positioning for a possible deal to reopen the Strait of Hormuz and for Friday's US employment report, and the dollar index rose 0.32% to 99.98 on Thursday after hitting a six-week low on Monday. "When you strengthen the yen, you by definition weaken the dollar and that's part of what happened in the first part of the week since Friday," John Velis, FX and macro strategist at BNY, told CNBC. "But the idea that maybe there's some good news on a ceasefire or a deal in the Persian Gulf has taken some of the dollar premium off with lower oil prices and so forth."

The constraint has been located, and it is not the war chest

The obvious question after a retracement is why the authorities do not simply do it again. Until this week the honest answer was that nobody outside the Ministry of Finance knew how much dry powder was left in a usable form. That gap has now been filled with a number.

Goldman Sachs Research estimates that of Japan's roughly $1trn in US dollar reserves, about $200bn sits in cash or cash equivalents — which, on the same bank's estimate, is close to the size of the operation just conducted. "They already have at their disposal enough to do another couple rounds of what we just saw," Karen Fishman, a Goldman Sachs Research strategist, said on the bank's Exchanges podcast, adding: "Realistically, they wouldn't come close to using all of that, but I think that just sort of hits home the point that they have plenty of capacity to keep intervening if they wish" (CNBC).

That table is the reason the story has changed shape. A reader watching the yen slide back toward 160 could reasonably conclude that Tokyo had run out of ammunition and the market had called the bluff. The estimate says the opposite: the ammunition is there, it is liquid, and the FIMA route makes the rest of it reachable without the self-defeating side-effect of dumping Treasuries. What is decaying is not the capacity to intervene. It is the effect of intervening.

Once you accept that, the whole question moves to the other side of the trade — and Goldman's own framing puts it there. Whether Tokyo pulls the trigger again "may hinge on the carry differential between Japanese and U.S. borrowing rates, which remains the overwhelming driver of the exchange rate", according to Praneet Shah, the bank's head of FX options trading. He put the scale of what a rate move has to overcome plainly: the BoJ would need to hike faster than expected to shift carry dynamics that have driven a 45% yen depreciation over five years.

The record: ¥13.8trn in two days, then an official confirmation

The yen spent July grinding to fresh four-decade lows, reaching 163.99 per dollar in the final week — its weakest since 1986. Then it moved almost vertically twice.

On Thursday 30 July, in New York hours, it ran from roughly 162.8 to as strong as 157.8 in about an hour. Tokyo said nothing, but the size can be inferred from Bank of Japan current-account projections, and Bloomberg's analysis of those accounts put it at approximately ¥8.45trn — about $53bn, some estimates closer to $59bn, which would be the largest single-day yen-buying operation ever conducted. Within 24 hours the yen had handed back close to half of it, closing Friday around 159.27.

Then came Friday's operation, and this one was different. A comparison of BoJ accounts released the following Monday against money brokers' forecasts pointed to a further ¥5.33trn, or roughly $34bn; Reuters put the figure as high as $36.6bn. On Sunday the US President said publicly that Washington was helping to support the yen, and on Monday Japan's Ministry of Finance confirmed the joint action, stating that it had countered "excessive volatility and disorderly movements in the Japanese yen in recent months". Finance Minister Satsuki Katayama told reporters Japan "will not hesitate conducting further coordinated intervention". Treasury Secretary Scott Bessent said Washington "strongly support[s] Japan's decisive market and monetary steps to correct the substantial undervaluation of the yen", and that it would not hesitate to join further operations.

The market response looked, for about a day, like the tell. The yen reached 155.20 on Monday — its firmest since early May, about 3.8% stronger across the two sessions — and on Tuesday was still trading near 156.8, which was not the previous week's fast fade. By Thursday that distinction had shrunk to a difference in speed. The confirmed joint operation bought roughly four days of retention where the solo one bought roughly one; both ended up leaking the move back at a rate set by the carry, and the honest conclusion is that coordination extended the half-life rather than changing the decay.

The genuinely new thing: the US paid in euros, not dollars

The New York Fed executes intervention in two capacities — for the Federal Reserve's own account under FOMC direction, and as fiscal agent for the Treasury's Exchange Stabilization Fund (NY Fed). Historically, US operations have been financed roughly equally from those two pots. What is unusual here is not who executed it but what was sold: Reuters and the FT reported that the US bought yen by selling euros from its reserves.

Mechanically, that still works. Yen strength comes from the yen leg — someone is bidding for yen in size, and the cross-rate arithmetic transmits it to USD/JPY whatever sits on the other side. But two consequences follow, and both are easy to miss.

First, the US added no direct dollar selling. A dollar-funded operation puts fresh dollar supply into the market on top of the yen bid; a euro-funded one does not. The dollar's fall against the yen came from the yen bid and the signal, not from Washington selling its own currency.

Second — and this is the constraint that governs everything — it caps the size at a pot you can look up. US official reserve assets totalled about $250.9bn at 24 July 2026, but most of that is gold, SDRs and the IMF reserve position. The deployable foreign currency, across securities and deposits, looked like this (US Treasury):

Read that against Japan's Thursday operation alone — up to roughly $59bn in a single session. The entire American euro holding is smaller than one day of Japanese firepower, and the whole foreign-currency reserve is smaller than two.

The conclusion is not that the US contribution is trivial. It is that the contribution is informational, not volumetric. Washington cannot out-trade the carry flow and does not appear to be trying to; what it supplies is the credible prospect of an unpredictable second participant. Escalating beyond the euro pot would mean changing instruments — selling dollars outright, monetising SDR holdings, or warehousing currency with the Fed. Those routes exist; none has been used, and the Treasury did not disclose what it spent. For the historical version of a coordinated operation that did reset a trend, see the Plaza Accord, and for the mechanics in general, how and why central banks step in.

The FIMA repo: the sentence that mattered more than the totals

Buried under the intervention arithmetic on 3 August was a procedural announcement with more forward content than any of the figures. Japan's Ministry of Finance said it plans to use the Federal Reserve's FIMA repo facility for future interventions.

The facility — the Fed's standing repo line for Foreign and International Monetary Authorities, made permanent in July 2021 — lets a foreign central bank borrow dollars overnight against US Treasury securities it already owns, rather than selling them. To understand why that is the most important sentence of the week, follow the funding leg of any yen-buying operation.

Japan is the largest foreign holder of US government debt, so the selling route has always carried a self-defeating quality: raising dollars by liquidating Treasuries pushes American yields up, and a higher US yield is precisely what makes the yen weak. An intervention funded that way partially finances its own reversal. It is also the reason Washington had an interest in participating that has nothing to do with Japan. Louise Loo, head of Asia economics at Oxford Economics, told CNBC that avoiding a scenario in which Japan dumped Treasuries was "possibly one of the key reasons" behind US involvement, adding that the emphasis on FIMA "was a clue that they'd like to avoid forced-selling as much as possible."

Masahiko Loo, senior macro strategist at State Street, put the weight of it plainly: the signal "may be bigger than the intervention itself." His reasoning is that naming the facility "tells markets Japan can raise dollar liquidity without selling Treasuries", addressing the worry that intervention "could pressure U.S. funding markets through short-end UST sales."

What this changes for a reader trying to price the next operation is the ceiling, not the direction. The euro pot caps what Washington can contribute at roughly $25.9bn. FIMA changes what Tokyo can contribute — not by making it larger in yen terms, but by removing the side-effect that made large operations counterproductive. Japan could always spend; the question was what the spending did to the rate gap on the way out. This is the first credible answer to that question, and it is why the next intervention, if it comes, is a cleaner test of the mechanism than the last one was.

Not everyone reads the American role as a strengthening of it. Robin Brooks, senior fellow at the Brookings Institution, argued that the euro funding is "confusing markets and will prove counterproductive", because it "invariably will have markets wondering why the U.S. didn't just fund Yen buying out of Dollars." The critique and the FIMA point are consistent with each other: both say the operation's effect is being carried by signalling rather than by volume, and signalling is only as durable as its next confirmation.

What the coordination changes — and what it doesn't

Coordination changes the risk of holding a short-yen position, not the return on holding one. While a dollar deposit pays roughly 2.5 points more than a yen deposit, the carry incentive keeps regenerating the flow. Coordination raises the toll on the road; it does not close the road — and the fortnight since priced exactly that distinction. The unilateral operation surrendered nearly half its gain inside a day, because only the price had changed. The confirmed joint one took two weeks to give back the same proportion, because the reaction function changed too. A higher toll slows traffic; it does not redirect it.

That is the practical content of the phrase "buying time", and it is worth being literal about the units. Tokyo bought roughly two weeks with up to $85bn, and the calendar it was buying time until is 18 September. The cost is paid in reserves and in the credibility of a threat that has to keep being believed; the counterparty pays nothing to wait. What has changed since is that the destination of the wait now looks more like a rate rise than another cheque — which is the only version in which the time bought turns out to have been worth the price.

The real driver: the interest-rate gap

Currencies, over any horizon that matters, follow the flow of capital — and capital chases yield. This is the single biggest reason the yen got to 163.99 in the first place.

The Bank of Japan has been normalising policy, and on 16 June 2026 it raised its policy rate to 1%, the highest since 1995. On 31 July it held there on an 8–1 vote while warning that underlying inflation could exceed its target — the detail covered in our note on the July BoJ decision. That is a milestone for a country that spent years at or below zero, but it has to be measured against the other side of the trade. On 29 July the Federal Reserve held its target range at 3.50%–3.75% for a fifth consecutive meeting, leaving a gap of roughly 2.5 to 2.75 percentage points between dollar and yen rates.

That spread is the engine of the carry trade: borrow in low-yielding yen, convert to dollars, earn the higher US rate. While markets are calm and the gap is wide, the trade is self-reinforcing — every yen borrowed and sold is downward pressure on the currency. Intervention can lean against the flow; it cannot repeal the arithmetic. Our explainer on how the carry trade works covers the mechanism in full.

Official rate data: Bank of Japan, the FOMC's 29 July statement, and FRED.

The September sequence now decides the follow-through

For two years the standing assumption behind every yen forecast was that the gap would close from the US side — the Fed would cut, and the carry trade would lose its fuel. That assumption is now in question, and the intervention does nothing to settle it.

At the July meeting the FOMC held, but the vote was 9–3, and all three dissenters — Beth Hammack, Neel Kashkari and Lorie Logan — preferred a quarter-point increase. The statement noted that inflation "remains elevated relative to the Committee's 2 percent goal, in part reflecting supply shocks that have driven price increases in certain sectors, including energy." The bond market took the point: the 30-year Treasury yield rose above 5.2%, its highest since 2007, and the 10-year climbed toward 4.74% — driven by the worry that policy is not tight enough for the inflation in front of it. Swaps put roughly a 60% probability on a September increase under Chair Kevin Warsh.

That is the part the second week of August revised, and it revised it in the yen's favour. July CPI rose 0.1% on the month, in line with forecasts, and the annual rate eased to 3.4% from 3.5% — the detail is in our note on the July CPI report. July producer prices then fell 0.03% against a consensus of +0.2%, though both core measures firmed underneath, which our PPI note unpacks. Neither print settled the argument, but the market treated the pair as removing the urgency from a September increase: Treasury yields pulled back after the CPI release and the S&P 500 closed at a record 7,798.99 on 13 August as traders read the data as reinforcing bets the Fed would hold. For the yen that matters through one channel only. A Fed that does not hike stops widening the gap; it does not narrow it. Only the other central bank can do that.

Two dates now matter more than any intervention total, and their order matters as much as their content. The FOMC meets on 15–16 September; the Bank of Japan meets on 17–18 September and announces on the second day. The Fed goes first by roughly 48 hours, which means the BoJ will set policy with the American answer already on the screen while the Fed sets policy without the Japanese one. Asymmetric information about a rate gap is not a detail — it is the difference between one central bank reacting and both guessing.

On the American side the hawkish case kept building through the week. Fed Governor Lisa Cook said on 5 August that she was open to the idea that the central bank may need to raise its rate target to deal with inflation that is "too high", while San Francisco Fed President Mary Daly said she was "completely supportive" of the July hold and wanted more data before September.

On the Japanese side the case is building from a direction that is easy to miss, and it is the most underrated fact of the week. The BoJ's minutes of the June meeting, published on 5 August, record the Board discussing the weak yen not as a market problem but as an inflation input. Members noted that price rises were partly "a reflection of the pass-through of cost increases driven by the recent depreciation of the yen", and one member, arguing for the June rate rise, observed that because "import prices had also been driven up by exchange rate developments" and this burdened "a considerable number of firms, including small and micro firms", it had "become more appropriate than before to adjust the degree of monetary accommodation" (Bank of Japan).

Follow that logic to its end and the intervention is competing with the thing most likely to work. A weaker yen raises Japanese import costs, which raises inflation, which strengthens the argument for a BoJ hike, which narrows the rate gap, which is the only mechanism that reliably strengthens the yen. Official yen buying interrupts that chain at the first link. It is not a contradiction the authorities have to resolve — the Ministry of Finance owns the currency and the Bank of Japan owns the rate, and they are answering different questions — but it does mean the fastest route to a stronger yen runs through a weaker one, and intervention slows the journey it is trying to complete.

That dissent is now on the record with a name and a number. At the July meeting the Board held at 1.0% by 8–1, with Takata Hajime voting against and proposing 1.25%, on the grounds that "the situation had shifted to a new phase in which the Bank needs to adopt a nimble approach in response to upside risks to prices caused by demand shocks stemming from overseas developments" (Statement on Monetary Policy, 31 July 2026).

The reasoning behind the other eight votes arrived on 10 August, and it is the most consequential document of the fortnight for anyone trying to price the yen. The July Summary of Opinions is not a set of minutes but a collection of the views members submitted, and read as a whole it describes a Board that has changed the question it is asking:

• On the pace: "Given that underlying CPI inflation has been approaching 2 percent and greater consideration should be given to upside risks to prices than before, it could be considered that the pace of policy interest rate hikes will be faster than market expectations, depending on developments in economic activity and prices as well as financial conditions."
• On the balance of risk: "risks to prices are significantly skewed to the upside", with the output gap indicating "supply shortages and excess demand".
• On the yen's own role: "For prices, the situation in the Middle East, the expansion in AI-related demand, and the depreciation of the yen all exert upward pressure."
• On the objective itself: "The focus of monetary policy has shifted from 'lifting underlying CPI inflation to 2 percent' to 'avoiding further upward deviation in underlying CPI inflation.' It cannot be said that 'the risk of waiting is marginal,' and it is therefore necessary for the Bank to accelerate the pace of adjustment to the degree of monetary accommodation."
• On the framework: the Board is now in "a new phase in which the Bank needs to adopt a nimble approach ... and to discuss the size of a rate hike, rather than adhering to a certain pace of rate hikes", and one member observed that the policy rate sits "below the lower bound of the broadly estimated range" for neutral.

Take those five lines together and the significance for the currency is direct rather than atmospheric. A central bank that has stopped asking how to lift inflation and started asking how to stop it overshooting is a central bank whose reaction function now points at the rate gap. The market has moved accordingly: pricing implies a 65% chance of a 25-basis-point hike on 18 September and about 40 basis points of tightening by year-end. That pricing is also the risk. "If they don't deliver" a September hike, Fishman said, "that would put renewed downward pressure on the yen" — which is the same sentence read from the other end.

So the yen sits between two live September meetings, and the fortnight since the intervention has narrowed which of them is doing the work. The August data pushed the Fed toward holding and the BoJ's own Board toward hiking, which collapses the range of outcomes onto the Japanese decision:

• BoJ hikes, Fed holds — now the priced base case at about 65%. The gap narrows for the first time in this cycle from the side that can actually narrow it, and the joint operation gets the fundamental follow-through it has been missing. This is the only branch in which 155 becomes a floor rather than the speed bump it has already proved to be once. Note what it costs Tokyo: nothing in reserves. A hike is the intervention that pays for itself.
• Both hold — the branch the market is 35% on, and the one that hurts. The carry incentive persists unchanged and the yen's level becomes a function of how often the market expects to be hit. Because pricing has already moved to expect a hike, a hold is not neutral: it is a disappointment that has to be unwound, which is precisely Fishman's point about renewed downward pressure. This is also the branch in which repetition depletes the $200bn cash pile and the ~$25.9bn euro pot starts to matter.
• Fed hikes, BoJ waits — now the tail, not the base case. The gap widens toward 3 points. No plausible intervention budget offsets that, and the authorities would be spending reserves to slow a flow their own policy divergence is accelerating. July's inflation prints made this less likely without removing it, since both core PPI measures firmed even as the headline fell.

There is a fourth path nobody chooses. The yen is a funding currency in calm markets — the thing you borrow — and flips to a safe haven in a crisis, as carry trades unwind violently (our anatomy of the 2024 unwind). A volatility spike remains the fastest possible reversal, and no authority needs to spend a yen to cause it.

The second channel: an oil shock lands on an energy importer

The rate gap explains the trend. A second factor explains why the yen had so little natural support to begin with: commodities, through Japan's oil-import bill.

Crude surged as the Middle East conflict escalated, with Brent pushing above $87 a barrel and traffic through the Strait of Hormuz thinning sharply (Al Jazeera). For most majors that is a risk-sentiment story; for the yen it is direct. Japan imports almost all of its energy, so dearer oil widens the import bill and creates fresh real-money yen selling as importers buy the dollars to pay for crude. It is a tax on the yen — and it is the same energy shock the FOMC cited for elevated inflation, which keeps US yields high. One shock, both channels.

Crude has since retreated as de-escalation hopes built, relieving pressure from this side: Brent closed Thursday 6 August at $82.49 after a 3.8% rebound, but was still heading for a weekly loss of roughly 7% on the prospect of a deal reopening the Strait of Hormuz — the mechanism is set out in our note on the strait and the commodity currencies. See also why gold fell despite the war and oil and the commodity currencies.

The BoJ's own Board is not treating that relief as permanent, and the reason is specific rather than generic. Its June minutes record one member arguing that even if the Middle East conflict ends, inflationary pressure would persist because of "the higher costs, such as for shipping and storage, associated with the procurement of alternative sources of supply", and a second noting that crude procurement costs were likely to stay elevated "because it was likely to take time to fully ensure safe passage through the Strait of Hormuz even after the end of the conflict". The July Summary of Opinions repeats the point with the softer oil price already in hand: crude and naphtha benchmarks "have declined from their peak levels reached in April", and supply and demand are balanced partly because delayed tankers have exited the Persian Gulf — but "once these temporary factors dissipate, supply and demand conditions could tighten again". For a currency whose central bank is weighing a September hike, an energy-cost floor is not a footnote — it is the input that keeps the hike case alive after the headline oil price falls. The live JPY page shows how the rate, risk and commodity factors score the yen today, and the USD page the other side of the pair.

What to watch next

The direction won't turn on the size of the next cheque. It will turn when a fundamental input shifts. The signposts, in rough order of importance:

• The September sequence, in order. Fed 15–16 September, BoJ 17–18. With 65% priced on a Japanese hike and the market no longer pressing for an American one, this is now the only branch that can narrow the gap rather than hold it still. The intervention changes none of it.
• Whether 160 draws an operation. The level, not the calendar, is the test. The market has already walked spot back toward the figure that preceded the last intervention; an operation there says the authorities are defending a band, and its absence says July was about disorder rather than level — and Goldman's estimate removes "they cannot afford it" as an explanation for the absence.
• Japanese inflation data before the meeting. The Board has told the market its trigger is upside price risk, so a firm CPI print does more for the yen than any intervention headline. The Summary of Opinions also flags that consumer-goods price rises are expected "to accelerate again toward early autumn".
• What the US sells next. If a further operation is again euro-funded, the ceiling is the ~$25.9bn pot. A shift to selling dollars or monetising SDR holdings would signal a far larger commitment — that would be the genuine regime change.
• Whether Japan actually draws on FIMA. Naming the facility is cheap; using it is observable, and a first drawing would confirm that Tokyo can scale up without pushing US yields — and its own headwind — higher.
• The official MOF figures. Japan's monthly release published 31 July covered 29 June to 29 July, stopping just short of both operations, so the confirmed totals land in the next one (Ministry of Finance). Those numbers validate or revise the ¥13.8trn estimate.
• US data between now and September. The direction of surprise matters more than the level. Goldman's Shah pointed to July 2024, when one of the most effective rounds of BoJ-MOF intervention landed on a US CPI miss compounded by a payrolls miss days later: soft American data does the yen's work for it by weakening the hike case, and it also makes a fresh operation more likely to stick. July payrolls fell 23,000 with 103,000 cut from the prior two months — the detail is in our payrolls note — and the August report is the next test of whether that was a turn or a wobble.
• The oil price and the Strait of Hormuz. Elevated crude taxes Japan's trade balance and keeps US yields high. Sustained de-escalation relieves both channels at once.
• Global risk sentiment. A volatility spike could trigger a carry unwind and a sharp yen rally regardless of what either central bank or treasury does. Fastest possible reversal, least predictable. See also the GPIF and the yen for the structural flow story underneath.

To learn how Pip Theory builds its fundamental currency-strength scores from five factors, see the methodology overview.

Educational macro context only — not investment advice.