Somewhere between $250 billion and $4 trillion, a ghost lives.
The yen carry trade (borrow cheaply in Japan, invest where yields are higher) has been the subject of a strange argument since the August 2024 unwind sent the Nikkei down 12% in a single session. One camp says the trade is dead. The other says it never left. Both cite data. Both sound confident. They cannot both be right, and at USD/JPY 160.46 with the vice finance minister invoking “resolute measures” for the first time in his tenure, the answer has stopped being academic.
What the futures show
Start with what is visible. The CFTC’s Commitments of Traders report tracks speculative positioning in yen futures on the Chicago Mercantile Exchange. The March data tells a story of its own. In early March, speculative positions flipped from net long to -16,600 contracts short. A week later: -41,400. By March 20: -67,800. The latest reading, for March 27, showed a slight trim to -62,800. Inside a fortnight, net shorts had more than quadrupled. The shorts are not just back. They are running.
The “carry trade is dead” argument leans on this data. At their 2024 peak, net shorts reached roughly 180,000 contracts. A drop to -19,000 by early February 2026 looked like capitulation. The trade appeared to have bled out quietly over 18 months of BOJ rate hikes and narrowing differentials.
But futures are, as the BIS put it in its post-mortem of the August 2024 episode, “only the tip of the iceberg.”
What the futures hide
The BIS estimated total carry trade positioning going into August 2024 at roughly ¥40 trillion ($250 billion). Even that figure, they noted, was “biased down due to data gaps.” The number covered only what could be inferred from reported positions. It excluded the bulk of over-the-counter derivatives, FX swaps and structured products that constitute the real plumbing of yen-funded leverage.
The FX swap market between yen and other currencies runs to $14 trillion in notional value. Not all of that is carry. Much of it is legitimate hedging by exporters, importers and insurance companies managing currency exposure on foreign assets. But the BIS itself has warned that the boundary between hedging and speculation blurs in practice. A Japanese life insurer buying unhedged US Treasuries is, functionally, running a carry trade. An asset manager rolling yen-funded forwards every three months is doing the same thing with different paperwork.
The “carry trade is alive” camp (BCA Research called it “a ticking time bomb” in February) points to this structural layer. Their argument is that the August 2024 unwind flushed out the fast money in futures, the hedge funds and CTAs that show up in CFTC data. What it did not flush out was the slow money: institutional portfolios, corporate treasury hedges and derivative structures with longer tenors that unwind over quarters, not days.
Estimates of this broader exposure range from $1 trillion to $4 trillion. Nobody knows the real number. The BIS has said as much.
Why it matters now
On March 27, Finance Minister Katayama told reporters that Japan would respond to yen weakness with “bold steps.” On March 30, Vice Finance Minister for International Affairs Mimura went further, warning that “if this situation continues, resolute measures will soon be necessary” and that the government’s aim was “all-fronts” (全方位). It was the first time Mimura had used the phrase 断固たる措置 since taking office in July 2024. That phrase, in MOF’s carefully graded lexicon, is the final verbal step before actual dollar-selling intervention.
More striking was the scope. Mimura said speculative activity was rising not just in FX but in crude oil futures, and that both markets were in the government’s sights. Bloomberg reported separately that the Ministry of Finance had contacted major domestic banks to gather views on the feasibility of intervening in oil futures markets, with yen stabilisation as the objective. If confirmed, this would represent a new front in Japan’s currency defence: attacking the oil-driven forced flow at its source rather than merely absorbing its effect in the spot FX market.
Two weeks earlier, Tokyo and Seoul issued a joint statement expressing alarm at the pace of won and yen depreciation, a rare coordinated signal. In late February, Nikkei Asia reported that Washington had indicated willingness to support coordinated FX intervention if Japan requested it. A rate check conducted by the New York Fed in January lent credibility to that report. If the US is on side, the arithmetic shifts: Aozora Bank’s chief market strategist Moroga Akira told Reuters that if MOF commits to intervention, they would aim to move USD/JPY by roughly five yen in one go. Unilateral action might manage one yen. The gap between those two numbers is, again, the amplitude question.
The BOJ held rates at 0.75% at its March meeting, but the Summary of Opinions revealed a split board, with some members arguing for a hike despite Middle East uncertainty.
USD/JPY hit 160.46 in the Tokyo morning session before Mimura’s remarks pulled it back below 160. The US-Japan 10-year spread is 205 basis points, edging into what traders call the “carry unwind danger zone” below 200bps, where the cost of holding yen-funded dollar positions begins to erode faster than the yield compensates.
The forced flows are running in opposite directions. Oil at $116 a barrel (Hormuz disruption) creates roughly ¥900 billion a month in additional forced dollar buying by energy importers, weakening the yen. Against that, March 31 fiscal year-end repatriation pulls yen back to Japan, and MOF intervention would do the same with far more force.
If the carry trade is already gone, MOF intervention into year-end repatriation would strengthen the yen by a few hundred pips, the market would absorb it and grind back to pre-intervention levels over the following months, as it did after the 2022 interventions.
If $1-4 trillion in derivative exposure is still out there, the arithmetic changes. Intervention that pushes USD/JPY into the mid-150s could start triggering margin calls on positions built and rolled forward at higher levels. Those margin calls would force more yen buying, which would strengthen the yen further, which would trigger more margin calls. The feedback loop is self-reinforcing: sell dollars to buy yen to meet margin, which makes the yen stronger, which creates more margin calls. That mechanism is what turned August 5, 2024 from a bad day into a 12% crash.
The difference between the two scenarios is not direction. It is amplitude. Both end with a stronger yen. One ends with a 3% move. The other ends with something closer to 10%.
The other side
A vocal minority of macro traders argue that no yen carry trade remains in the US market, that it was fully unwound by mid-2025. Their evidence centres on CFTC positioning, bank lending data and the behaviour of USD/JPY during recent risk-off episodes, which they say shows no carry-trade signature.
They may be right about the US hedge fund community. The futures data supports the case, and the speed of the March 2026 rebuild in shorts (from net long to -67,800 inside a fortnight) looks more like fresh tactical bets than legacy positions being maintained.
But hedge funds are not the whole market. Japanese life insurers manage over ¥390 trillion in combined assets, with a substantial share held in foreign securities. Bloomberg data from March 2025 showed the nine largest covered only 46% of their foreign holdings with yen hedges, leaving the rest exposed to currency moves. Mrs Watanabe, the collective term for Japanese retail FX traders, held ¥4.2 trillion in margin positions as of late 2025. European and Asian institutional investors have their own yen-funded books that do not appear in Chicago futures data.
The August 2024 unwind was triggered by futures positioning. The next one, if it comes, would be triggered by something slower and harder to see: the point at which rising Japanese yields make it cheaper to invest at home than to bear the currency risk abroad. That process has been under way since the BOJ started hiking. It accelerates with every basis point added to JGB yields and every yen of intervention that reminds foreign holders that the currency can move violently against them.
What we do not know
Honest accounting demands a list of uncertainties.
We do not know the true size of outstanding yen-funded positions. The BIS has the best data and admits it is incomplete. Estimates range across an order of magnitude. Anyone who quotes a precise figure is guessing.
The timing of MOF action is equally opaque. Mimura’s language has reached the final verbal step, but the oil futures dimension adds a variable that has no precedent. Intervening in crude oil markets to defend the yen would be operationally and politically distinct from selling dollars in spot FX; whether it is even practical at scale remains an open question.
On monetary policy, the BOJ board itself appears undecided between the domestic inflation argument (hike to contain import costs) and the external shock argument (hold to avoid tightening into an oil crisis). The March Summary of Opinions suggests no consensus. But Governor Ueda, speaking before the lower house budget committee on March 30, offered a pointed warning: if markets come to believe that short-term rates are not being adjusted appropriately and that inflation may overshoot, long-term rates face upside risk. The implication is that inaction on rates could itself destabilise the bond market. With the 10-year JGB yield at 2.39%, a 27-year high, this is not a theoretical concern.
Then there is March 31 (tomorrow, as this is written). Fiscal year-end repatriation is a calendar certainty. Its magnitude is not. If Japanese corporates and funds bring home more yen than expected, the convergence with Mimura’s rhetoric could produce a sharp move. If repatriation disappoints, the yen stays above 160, and MOF has to decide whether to spend reserves fighting oil-driven fundamentals or try the untested route of crude oil futures intervention.
The carry trade may be a ghost. But the risk is asymmetric. If the ghost is real and MOF intervenes, a 10% yen move would reprice every foreign-funded position in Japan (equities, JGBs, credit) in a matter of days. If the ghost is not real, the same intervention produces a 3% move that fades within a fortnight.
The amplitude question may take months to resolve. The rate question will not. Governor Ueda told parliament on March 30 that failing to raise short-term rates risks destabilising the long end. The BOJ’s Summary of Opinions shows board members arguing for hikes even with oil above $110. Whether the next hike comes in April or July, the direction is no longer in dispute. Japanese interest rates are going up.
That is the one variable in this story that does not depend on whether the ghost is real. And it is the variable that matters most for the sector we have been writing about since the beginning. Net interest margins, after three decades at zero, are repricing. The carry trade determines the speed of the yen. The rate cycle determines who earns more from every basis point of it.
Data as of 2026-03-30 ~10:30 JST.