In 2025 Japan ran a current-account surplus of ¥31.9trn, a record and the second in a row. Yet the yen has not stopped falling. Even after the Bank of Japan raised rates in June 2026, the currency trades around ¥161 to the dollar, close to its weakest since 1986. A record creditor’s currency sits at a four-decade low. Unless that contradiction can be resolved, the question every market participant is asking has no answer: where does the yen go next?

The market mostly watches a single clock: the gap between Japanese and American interest rates. Widen it and the yen falls; narrow it and the yen rises. On tariffs, the assumption is that Section 122 will slide into Section 301 with the burden roughly unchanged. So the rate stays range-bound. That is the standard reading.

My argument is that there are three clocks, and all three now point the same way. The short clock is flow: positioning, futures, intervention. The medium clock tracks the rate differential. The long clock measures the currency’s underlying strength, its national power. Three hands rarely align. And the safety net that carry sellers have leaned on, the belief that the level always reverts, is being pulled away by the drift of that long hand. One trigger could run the tape backwards: a collapse of the legal basis for American tariffs, forcing the crowded carry trade to unwind. The market prices that as tail noise. A moving set of hands and an untriggered switch: those are the two halves of the case.

The short clock: shorts are stacked, the defence is losing its bite

Positioning leans yen-weak. Speculators are net short, and on the CFTC’s reading leveraged funds were short about 115,000 contracts as of 23 June, having added again the week before without covering. At ¥12.5m a contract that is roughly ¥1.4trn of notional. The carry trade is crowded on the short side, as one-sided as it has been in years. That crowding cuts both ways. With no trigger it is fuel for further weakness; pull the trigger and it becomes the spring for a violent snap-back. More on that when we reach the vessel.

The defence is the other half. Between 28 April and 27 May the authorities bought yen on a record monthly scale, about ¥11.7trn. The reserves that fund it fell 5.6%, or $77bn, in May, the steepest drop since 2000; they still hold more than $1.3trn. This is not an empty magazine. The problem is potency. The yen kept falling after the intervention and the effect did not hold. The market still fears another round, which caps the topside, but the authorities know the effect is short-lived.

On the supposed front-running of the 24 July deadline: it shows on the American import side but not on Japan’s. Export volumes were roughly flat in the 2025 fiscal year, up 0.6%. So I will not make a rebound in front-loaded trade the main character. What the short clock shows is the weight of the shorts and a defensive line that no longer bites.

The medium clock: the gap widens, and the same oil shock does it

The Bank of Japan raised its policy rate by 0.25 points to 1.00% in June, a level last seen in 1995, 31 years ago. Governor Kazuo Ueda was absent, in hospital, and the vote was 7-1. The next meeting, on 31 July, is expected to hold, and the consensus, Nomura included, puts the following hike in December.

America, by contrast, has held at 3.50-3.75% for four meetings. June was the first under Kevin Warsh, the new chairman who took over in May; the language pointing to easing was dropped, and the dot plot rose to 3.8% by year-end. The market has begun to price a hike this year, perhaps as soon as October.

Here is the asymmetry most people miss. Both central banks face the same supply shock: the oil spike that followed the Iran crisis from late February, with WTI averaging about $94.5 a barrel, roughly 1.45 times a year earlier. They read it in opposite directions. The Federal Reserve leans towards a hike, treating it as inflationary; Ueda, in a May speech, warned against chasing supply-driven price rises with monetary policy. The lesson of the first oil shock was not to be late; the governor argues the initial conditions differ and signals he is in no hurry.

So the same oil price widens the gap rather than closing it. The Fed leans in, the BOJ leans away. The gap is about 2.6 points now (American 3.50-3.75% against Japanese 1.00%) and set to widen into the autumn. The June hike is the proof: the BOJ moved and the yen still sank to ¥161. What binds the currency is not the level of rates but the gap that will not close. What actually funds the carry is not the policy gap but the 10-year spread, and there, American 4.4% against Japanese 2.6%, it is still about 1.8 points wide. No realistic BOJ path (a terminal rate of 1.50%, 1.75% at a push) closes the distance to American yields in the high threes.

The long clock: the level it should revert to has itself moved

It is the hand that should worry carry sellers most.

The real effective exchange rate (REER), the broadest measure of a currency’s strength, stood at 66.33 in March 2026 (2020=100), below the level at which the series began in 1970, a record low. Back then the yen was fixed at ¥360 to the dollar and the index sat around 75. Today’s yen has fallen through even that ¥360-era strength. At its 1995 peak the index was about three times current levels. Daisaku Ueno of MUFG Morgan Stanley reads the decline as, in all likelihood, a fall in national power driven by an ageing, shrinking population.

The textbook carry trader sells the low-yield currency to earn the gap, trusting that long-run purchasing-power parity will hold the floor. Currencies do not run for ever; fundamentals pull them back. For the yen, the anchor that should pull it back has itself drifted down. REER measures a relative price level, not a spot fair value, so the next step is interpretation rather than measurement: the ¥110-120 the market long treated as the floor has, to my reading, moved towards ¥140-150. The carry short is therefore not a contrarian bet on mean reversion but a trend-following bet on structure. The safety net the market quietly assumes is gone.

Why did the anchor move? National power says nothing on its own, so measure it by which balance-of-payments flows actually touch the spot yen.

The record surplus rests on income, not trade. Japan’s primary-income balance was ¥41.6trn in 2025, of which direct-investment income, the main driver, was around ¥25trn. Most of it is reinvested abroad and never comes home as yen buying. Japan has moved from a trade-surplus country to a mature creditor that keeps its earnings offshore. The money stays out.

Behind that, two flows pay out in foreign currency as they go. Energy and digital. The energy deficit was ¥24.2trn in 2024, a reflexive drain that swells as the yen weakens; self-sufficiency sits in the low teens, and the only structural lever to shrink it is restarting nuclear plants. The digital deficit hit a record ¥6.7trn in the same year, led by the ¥2.5trn of ’telecommunications, computer and information services’ that includes cloud fees. Wider use of generative AI pushes it higher still. Foreign services cannot be swapped out when the yen falls, so an inelastic sale of yen leaves the country month after month. In 2024 that digital deficit overtook the inbound-tourism surplus of ¥5.9trn.

Semiconductors need their sign split. The goods side is in surplus: exports of chipmaking equipment hit a record in 2024, concentrated on China, Taiwan and South Korea, and electronic components grew 8.2% in 2025, part of what tipped the fiscal-2025 trade balance back into surplus. Japan holds the equipment and the materials, the picks and shovels. But the deficit shows up one layer higher. Advanced logic and AI GPUs are imported, and above them sits the cloud-software-and-services layer that is the digital deficit. The surplus side also rides the capex cycle and leans on China, so it is first to be cut if export controls tighten. A cyclical, geopolitically exposed surplus against a repeating, inelastic deficit: the reliable long-run yen driver is the latter.

The paradox now resolves. The record surplus is, in currency terms, hollow. Earnings stay offshore or sit exposed to geopolitics; the bills go out in real time and do not flex. The net flow that touches the spot yen runs opposite to the headline. That is why a record surplus and a four-decade-low yen share the same year. The hinge between the long and medium clocks is demography: a shrinking workforce caps potential growth and pushes down the neutral rate the BOJ can hold, an estimated 1.1-2.5% with a floor near 1.00%. So the medium-term gap is structurally hard to close. The long hand ties down the root of the medium one.

With three hands pointing to a weaker yen, the trigger that could run the tape backwards is neither rates nor flow. It is the legal vessel of American tariffs.

Follow the sequence and the vessel has been rebuilt twice in short order. The July 2025 US-Japan deal set the reciprocal tariff at 15%, cut the car tariff from 27.5% to 15% and had Japan put up a $550bn (about ¥80trn) investment facility. Then, on 20 February 2026, the US Supreme Court ruled the reciprocal tariffs unlawful under the International Emergency Economic Powers Act (IEEPA); collection stopped on the 24th. A 10% surcharge under Section 122 of the Trade Act took its place, leaving Japan with roughly 13% once the standard rate is added. That Section 122 measure also lapses, at its statutory 150-day limit, on 24 July.

The replacement is ready. The Office of the US Trade Representative (USTR) has, on what it argues is a failure to police forced labour, proposed Section 301 tariffs on 60 economies including Japan, dated 2 June. The 15 economies that already ban forced-labour imports get 10%; the other 45 get 12.5%, and Japan sits in the latter. The 301 duties stack on the most-favoured-nation rate, while goods under Section 232 of the Trade Expansion Act (cars, steel, pharmaceuticals, critical minerals) are carved out. Item by item, 232 or 301 is assigned, and the whole is engineered back towards the 15% that existed before the ruling. Comments close on 6 July, the hearing falls on 7 July and a final decision is expected in July, near enough to the Section 122 expiry. A separate 301 case on industrial overcapacity covers 16 economies and more than 75% of American imports. The Section 232 product tariffs (cars at 15%, steel, aluminium and copper, semiconductors and pharmaceuticals under review) run on independently.

Here is the heart of it. Japan paid for a 15% deal with an ¥80trn investment facility. The tariff part of that deal has lost its legal footing twice and been rebuilt twice. The design of Section 301 is to restore, on a different legal basis, the negotiated 15% the ruling struck down. The price paid stays put; only the vessel that was bought is swapped out. The effect, and seemingly the aim, is to keep tariff revenue roughly flat. The variable is not the rate but the legal basis. This is reflexivity: the pricing is held constant while the mechanism behind it is continuously replaced.

The trigger the market has not priced is the vessel breaking once more. The Court of International Trade ruled Section 122 unlawful in May, though collection continues under appeal. Refunds on entries already cleared, it has been suggested, are possible. Should the judicial doubt reach Section 301 and refunds accelerate, it would bite as a loss of confidence in American policy.

One objection deserves a pre-emptive answer. A normal risk-off bids the dollar as a haven. But this shock starts inside America’s own institutions. The haven that should be bought is itself the impaired asset. So in this case, and this case alone, risk-off can sit alongside a weaker dollar. There the crowded yen carry unwinds by force, and the yen snaps the other way more sharply than by any other route.

Note that this is one of the few places to hold any conviction about a stronger yen. The three clocks point to weakness. Even if a judicial shock sends the yen sharply higher, two things keep that strength from lasting. One is the long anchor: with the level it should return to having moved down, a spike decays. The other is the dollar’s footing. In August 2024 a build-up of yen shorts unwound at once and the dollar fell from ¥161 to around ¥142 in a matter of weeks; the Fed was then heading into cuts, and a softer dollar pushed the yen up. Now the Fed leans hawkish and the dollar has support. The V can come; the 2024 amplitude is unlikely. The trigger is sharp, but nothing follows it.

Pull it together and the base case is continued yen weakness. The hands align, the intervention loses its bite, the gap widens into the autumn. I am not calling a bottom for the yen. The direction is what can be bet, and the direction is down.

Get the weighting right. The weight sits on the three clocks that are actually flowing. The vessel breaking (a judicial spillover to Section 301, refunds accelerating) is a latent tail with no sign of it yet: no suit, no refund surge. So it is not a bet but an unpriced thing to watch. Unpriced does not mean probable. The BOJ surprise the market braces for is already in the price, was delivered in June and did not work. The edge that is left lies in the legal structure. Bet the direction; watch the vessel.

A word on sectors, without naming names. The chipmaking-equipment makers that anchor the export surplus are a real strength, but their reliance on China carries the structural risk of tighter export controls. A weaker yen lifts export margins, yet alongside dearer oil it feeds an import-led inflation that erodes households and domestic demand. That is a description of mechanism, not a recommendation. A domestic-light position that reads the weak yen as nothing but a tailwind is exposed to another force: falling real purchasing power. The value trap here is worth watching as mechanics, not as a scolding.

The yen is a relief valve for a rate gap that will not close, pushed along an anchor that has drifted down. On top of that structure sits a tariff whose legal basis is swapped out, again and again. With three clocks pointing the same way, the only thing that could turn the tape is the vessel breaking, which nobody has priced. In July, one of those vessels quietly lapses.