Tearing apart the yen carry trade unwind

The yen.

Financial media is covering USD/JPY as a currency story. Strong yen, weak dollar, trade tensions with Tokyo, maybe some G7 communiqué about competitive devaluations. That's the dashboard. The engine is a $4 trillion funding trade that has been running for a decade on one assumption: that Japan would never, seriously, raise rates.

The Bank of Japan has now raised its policy rate to 0.5% — the highest since 2008 — and signaled it isn't done. That sounds small. It isn't. The carry trade is not sensitive to the level of Japanese rates. It is sensitive to the direction and the credibility of the signal. Both just changed.

The trade has a simple structure. You borrow yen at near-zero, convert it to dollars or Australian dollars or Brazilian reais, and park it in something that yields more. The spread is your carry. As long as the yen stays weak and Japanese rates stay pinned, you clip the coupon and go home. Hedge funds have been running this since 2013. Some of the positions are enormous. I know people who built careers on it.

The problem with a crowded carry trade is not the entry. It's the exit. Everyone who borrowed yen to buy something else now has to sell that something else to buy yen back. At the same time. Into a market that is already moving against them. August 2024 was a preview. The BOJ hiked 15 basis points and $6 trillion in global equity market value evaporated in ten days before they walked it back. They didn't walk it back this time.

Story off. Numbers on.

The spread compressing is the mechanism. A 400bps differential still looks wide on paper. But the carry trade is not priced off the current spread — it's priced off the expected spread at the point of exit, discounted by the cost of unwinding into a crowded market. When the BOJ credibly signals more hikes, the expected future spread collapses, and positions that were profitable yesterday are suddenly underwater on a forward basis today.

The sequence of what breaks first is mechanical and it's been the same every time this trade has partially unwound — 1998, 2008, August 2024. First the highest-beta, most-liquid assets sell off hardest because that's what gets liquidated first to cover the yen leg. EM equities, high-yield credit, commodity currencies. Then the selling pressure bleeds into investment-grade credit as funds need to raise more cash. Then — and this is the part that gets ignored until it's too late — Japanese life insurers and pension funds start repatriating the foreign bonds they bought to escape negative rates at home.

Japan's institutional investors — the life insurers, the pension funds, the Norinchukin Bank — hold an estimated $1.1 trillion in foreign bonds, the majority of it in U.S. Treasuries and European sovereign debt. They bought those bonds because Japanese yields were zero or negative. They will sell them when Japanese yields offer a credible domestic alternative. The BOJ just moved that threshold closer. When they start repatriating at scale, the buyer of last resort for the world's safest assets starts becoming a seller. That is the part of this machine that the yen headline misses entirely.

The part nobody is stress-testing is the hedge. Or more precisely: the assumption that there is one.

A large portion of the yen carry trade is run without a proper FX hedge on the currency leg. The logic was always: why hedge the yen if it's only going to weaken? The cost of the hedge eats your carry, and the hedge itself is wrong year after year. So the position became structurally unhedged. That worked for a decade. When USD/JPY moved from 115 to 160, the unhedged position was collecting carry and enjoying a 45-figure tailwind on the currency leg simultaneously. It felt like genius. I know what that feeling does to a risk book.

Here's the fault line. The crowded end of this trade is not at the large, well-capitalized hedge funds who run it with volatility-targeting overlays and can rebalance incrementally. It's at the second tier — the family offices, the smaller macro funds, the corporate treasury desks that have been running yen-funded positions as a quiet yield enhancement strategy for years. These are not set up to absorb a rapid 15-figure move in the currency leg while simultaneously managing drawdowns on the asset side. They don't have the systems. Some of them barely have the risk team.

August 2024 was instructive. USD/JPY moved roughly 12 figures in three weeks. That was enough to trigger margin calls, forced liquidations, and a cross-asset spasm that had absolutely nothing to do with the fundamentals of anything being sold. S&P futures, Bitcoin, emerging market bonds — all of it went down together, not because those assets had a problem, but because they were the liquid things sitting in the same books as the yen position. The BOJ blinked. The carnage stopped. They are not blinking this time.

The CFTC positioning data shows speculative yen shorts near their lowest level since 2021 — meaning some of the most aggressive carry positions have already been trimmed since August. But "near the lowest since 2021" is not "clean." There is still a very large structural short on the yen sitting in books that were not built for the scenario now unfolding. That overhang doesn't disappear. It transacts.

The trigger isn't a BOJ meeting. It's a margin call at a fund whose name you'll never read in a press release.

One forced unwind creates a yen bid. The yen bid pushes USD/JPY lower. Lower USD/JPY triggers risk models at adjacent funds that are monitoring the currency as a proxy for carry trade health. Those risk models generate sell signals on the correlated assets. Those sell signals execute automatically — no human needed, no fundamental thesis required. The cascade is self-reinforcing until either the BOJ steps in to cap the yen or the positions are fully liquidated. Neither happens cleanly or quickly.

The U.S. Treasury leg is what makes this different from a normal risk-off episode. If Japanese institutions sell $200–300 billion in Treasuries over a 12-month repatriation cycle — which is within the range of analyst estimates for a full BOJ normalization — you get a supply shock on the long end at exactly the moment the Federal Reserve is already carrying a bloated balance sheet and political pressure is making QE difficult to restart. The 10-year yield doesn't need to go to 6% to cause damage. It needs to go to 5.25% and stay there while equity multiples are still priced for 4.5%.

Where does capital actually go in this scenario? Not into broad EM, which gets hit hardest in the first wave. Not into investment-grade credit, which correlates with duration pain. The cleanest asymmetry — and I'm holding this loosely, because timing a carry unwind is how people lose money with conviction — is in short-duration instruments that benefit from a risk-off flight to quality without the duration exposure: short-term Treasuries, Japanese domestic equities repricing upward as repatriated capital finds a home, and volatility itself, which is priced for complacency at levels that assume August 2024 was a one-off.

The carry trade took twelve years to build. It will not unwind in twelve days. But the unwinding has started — the direction changed in March, the BOJ didn't walk it back, and the second tier of the carry trade is not equipped for what comes next. August 2024 was the fire drill. The drill is over.