The Carry

---

Of all the leverage in the global financial system, the most important is the kind nobody can quite measure, that appears on no single balance sheet, that has no manager and no ticker and no annual report — and that funds, invisibly, a meaningful slice of every other trade in the world. It is called the yen carry trade, and to understand why a committee of central bankers in Tokyo can crash the Nasdaq in an afternoon, you have to understand it. It is the deep plumbing beneath the floorboards of the whole house, and the water in it has started, for the first time in thirty years, to flow the wrong way.

The mechanism is almost embarrassingly simple. For decades, Japan held interest rates at zero, or below zero, while the rest of the world paid something for money. So a hedge fund — or a bank, or a Japanese pension, or a retail trader in Tokyo, or a global macro desk in London — could borrow yen for essentially nothing, sell those yen, and use the proceeds to buy something that yielded more: U.S. Treasuries paying four percent, Mexican bonds paying nine, Nvidia stock, the S&P 500, gold, bitcoin, an Australian property. The trade earns the difference — the "carry" — between the near-zero cost of the borrowed yen and the return on whatever you bought with it. As long as the yen does not rise and the funding stays cheap, it is the closest thing finance offers to free money, and it has been printing for thirty years. The whole world has been, in a sense, long everything and short the yen, financed in Tokyo.

How big is "big"?

Nobody knows exactly, which is itself the danger. The honest answer to "how large is the yen carry trade" is a range so wide it tells you the system does not understand its own exposure. On the narrowest definition — speculative, yen-funded leveraged positions held by hedge funds and the like — estimates run to $350–500 billion, and after the August 2024 tremor Morgan Stanley still put the outstanding figure near $500 billion. That is the visible tip. On the broadest definition — every cross-border position in the world ultimately financed by cheap yen, including the vast foreign holdings of Japanese institutions — Deutsche Bank has estimated the global carry trade at something on the order of $20 trillion.

Twenty trillion dollars. Set that number down and look at it. It is comparable in scale to the entire annual output of the United States. Even if the true figure is a fraction of it, even if the "real" exposed amount is one or two trillion, the point stands: this is not a niche strategy. It is a structural feature of the global financial system, a column of leverage so large and so diffuse that no regulator monitors it, no clearinghouse nets it, and no one — genuinely no one — can tell you how much of the world's asset prices are propped on borrowed yen. When the cost of the most basic input to all of it — the price of borrowing yen — begins to rise, the system cannot even locate, let alone defuse, its own fault lines.

And there is a second, quieter column beside it. Japanese institutions — life insurers, banks, pension funds, the Government Pension Investment Fund, asset managers — hold something like $5 trillion in overseas assets, accumulated over decades of sending savings abroad in search of the yield they could never find at home. That is not speculative carry; it is patient capital. But it answers to the same arithmetic. The day Japanese bonds yield enough to bring that money home, the largest creditor nation on earth has a reason to repatriate — to sell foreign assets and buy domestic ones — and the flow that has supported global markets for a generation can run in reverse.

The foundation moves

For the entire life of the carry trade, the one thing you could count on was that the Bank of Japan would keep its rate near zero. That assumption is now dead. The BOJ has entered the first genuine tightening cycle in a generation, and the numbers mark a regime change.

In December 2025 the Bank of Japan raised its policy rate to 0.75% — the highest level since 1995, a thirty-year high — and held there through the early months of 2026. That sounds trivial against American or European rates, and in absolute terms it is tiny. But the carry trade does not run on the level of Japanese rates; it runs on their direction and on the differential against the rest of the world. For thirty years the direction was flat-at-zero. Now it is up. And the long end has moved far more violently than the policy rate: the 10-year Japanese government bond yield has climbed to roughly 2.77% — its highest in nearly three decades, up from around 1.1% at the start of 2025 — and the 20-year JGB touched 3.555% in May 2026, a multi-decade high. Japanese government bonds, which for a generation paid nothing and did nothing, now actually pay something. That single fact corrodes the entire edifice.

The reason is twofold, and both halves matter. The first is that the BOJ is not only raising its policy rate but withdrawing its thumb from the long end: it has cut its monthly government-bond purchases from about ¥5.7 trillion in August 2024 to roughly ¥2.9 trillion by early 2026, removing the suppressive buying that pinned yields down for years. The market, left to price JGBs more on its own, is demanding far more yield — and inflation above the BOJ's 2% target for a fourth consecutive year gives it every reason to. The second is political and combustible: Japan's leadership has leaned toward easier money and fiscal support even as the central bank tightens and the bond market revolts, a tension that the most credible analysts warn could push the 20-year yield toward 4.00% and, if it climbs without a matching rise in Japanese productivity, trigger a disorderly unwind. The foundation is not just moving. It is being pulled from two directions at once.

The dress rehearsal

We are not speculating about what an unwind looks like. We have the footage. On a single day in early August 2024, the world got a compressed, terrifying preview.

The trigger was almost absurdly small by the standards of global finance: the Bank of Japan raised its rate by a sliver, to 0.25%, and signaled more to come, while at the same moment weak U.S. jobs data made American assets suddenly less attractive. The yen, which carry traders were collectively short, began to rise. And because the trade is leveraged and crowded, the rise fed on itself: as the yen strengthened, every carry position started losing money on the currency leg, forcing traders to buy back yen to cut losses, which pushed the yen higher, which forced more buying, in the self-reinforcing spiral that is the signature of a crowded leveraged trade going into reverse. To raise the yen to repay the loans, traders sold what the borrowed yen had bought — and they sold all of it at once, indiscriminately, because in a margin call you sell what you can, not what you want to.

The damage was global and instantaneous. Japan's Nikkei crashed 12.4% in a single session — its worst day since the 1987 crash. Bitcoin fell from around $64,000 to $49,000 in 48 hours, with hundreds of billions wiped from crypto. Volatility gauges spiked, hedge funds were forced to liquidate, and markets thousands of miles from Tokyo — American tech, emerging-market bonds, anything that had been bought with cheap yen — sold off together, for no reason intrinsic to any of them except that they shared a common funding source that had suddenly become expensive. It lasted only days; markets bounced; the world moved on and mostly forgot. But it was a controlled demonstration of the core fact: a small move in Japanese monetary policy can transmit, through the carry trade, into a violent simultaneous repricing of unrelated assets across the planet. August 2024 was a fire drill. The building did not burn down. It only filled with smoke long enough to show everyone where the exits weren't.

Why the slow grind is worse than the shock

The comforting reading of August 2024 is that the system proved resilient: it wobbled, it recovered, the carry trade partially unwound and life went on. The discomforting reading — the correct one — is that August 2024 unwound only a fraction of the trade, on a temporary scare, and still managed a once-in-a-generation single-day crash. The fundamentals that caused it have not reversed; they have intensified. In 2024 the BOJ rate was 0.25% and the 10-year JGB yielded around 1%. Today the rate is 0.75% and the 10-year yields close to 2.77%. The differential that powers the trade is narrower now than it was when the trade blew up, and the direction of travel is unmistakably toward narrower still.

The greater danger this time may not be another single cathartic crash but something harder to fight: a slow grind. A sudden shock, paradoxically, is survivable, because it forces a fast, cleansing unwind and a snap-back. A grind is worse. If JGB yields climb steadily, month after month, the carry differential erodes steadily, and the trade becomes steadily less profitable to hold and steadily more tempting to exit — not in a single panic that everyone hedges against, but in a continuous, grinding repricing across multiple asset classes that no one can time and no hedge cleanly covers. The repatriation of that $5 trillion in Japanese overseas savings, in particular, would not arrive as a thunderclap. It would arrive as a persistent bid for yen and a persistent offer in foreign assets, a tide going out so gradually that each individual day looks ordinary and only the cumulative line, months later, reveals that the support beneath global markets has been quietly withdrawn.

The differential closes from both ends

There is a compounding danger that the focus on Tokyo obscures, and it is essential to the timing. The carry trade's profitability is not a function of Japanese rates alone; it is a function of the gap between Japanese rates and everyone else's. And right now that gap is being squeezed shut from both directions at once. On one side, Japan is raising — policy rate up to 0.75%, JGB yields up to thirty-year highs, the BOJ draining its bond purchases. On the other side, the rest of the developed world, having spent two years at the top of its own hiking cycle, is widely expected to be cutting — the Federal Reserve and the European Central Bank leaning toward lower rates as their inflation fades. A carry trade earns the difference between what you pay to borrow and what you earn on what you buy. When Japan's borrowing cost rises and America's lending yield falls, the difference compresses twice as fast as either move alone would suggest. The trade is being attacked from both ends of its own arithmetic.

This is what makes the present moment genuinely different from the long history of false alarms. The carry trade has been pronounced dead many times over thirty years, and it kept paying, because every previous scare was a one-sided wobble: a brief yen spike that faded, a single BOJ feint that came to nothing, against an unchanged backdrop of a vast and durable rate gap. What is happening now is structural rather than episodic. The differential that is the trade's entire reason to exist is narrowing on a sustained basis, driven by opposite-direction central-bank cycles that are unlikely to reverse soon. Every month the gap closes a little more, the carry pays a little less, and the rationale for holding the position — to say nothing of adding to it — weakens a little further. A trade does not need a crash to die. It can simply stop being worth doing, and a $20-trillion edifice of "stop being worth doing," exited gradually by millions of holders who each individually decide the juice is no longer worth the squeeze, is precisely the slow grind that no one can hedge.

And the political overlay makes the path more treacherous, not less. Japan's government has tilted toward monetary easing and fiscal stimulus even as its own central bank tightens and its own bond market sells off — a three-way tension between an easing-minded administration, a tightening BOJ, and a bond market demanding ever-higher yields. Policy incoherence of exactly this kind is what converts an orderly normalization into a disorderly one. If markets come to believe Japan will be forced to choose between defending the yen and funding its government, or between fighting inflation and pleasing its politicians, the yen can move in ways no carry model prices — and it is the yen, not the JGB, that pulls the trigger.

The hidden link in everything you own

Here is why this belongs in your thinking even if you have never traded a yen in your life, even if you own nothing but an American index fund. The carry trade is a hidden correlation engine. In normal times, your U.S. tech stocks and Mexican bonds and gold and bitcoin are different things with different drivers, and you imagine you are diversified across them. But to the extent they have all been bought, at the margin, with leverage sourced from cheap yen, they share a single secret dependency — and when that dependency is stressed, they stop being different things and become the same thing: collateral to be sold to repay a yen loan. The diversification you thought you had evaporates at exactly the moment you need it, because the assets were never as independent as they looked. They were all, invisibly, the same trade.

This is the deepest reason the yen carry trade matters more than its obscurity suggests. It is a transmission mechanism that can turn a Japanese monetary decision into a global, cross-asset, correlated sell-off — converting the "diversified" global portfolio into a single leveraged bet on the price of borrowing yen staying low. For thirty years that bet paid, reliably, boringly, which is exactly how it grew to $20 trillion of loosely-defined exposure: nothing recruits capital like a trade that has worked for a generation. And nothing is more dangerous than a crowded trade that everyone has come to treat as a structural constant rather than the leveraged position it actually is.

What the watchers are watching

The signals to watch are not obscure, and the professionals are already watching them, which is its own kind of warning — crowded exits get more dangerous, not less, when everyone is eyeing the same door. Watch the yen itself: a sustained, sharp strengthening of the yen is the lit fuse, because it is the currency move that turns carry profits into losses and forces the buy-back spiral. Watch the JGB yield, especially the 20-year as it presses toward 4%: every basis point higher narrows the carry and strengthens the case for Japanese money to come home. Watch the BOJ's meetings — the April 2026 decision to hold came on a split vote, with dissenters pushing to raise to 1%, a sign that the tightening pressure inside the central bank is real and building. And watch the gap between what Japan's politicians want (easier money, fiscal support) and what its bond market is forcing (higher yields), because that tension is precisely the kind of policy incoherence that produces the disorderly outcome rather than the managed one.

None of this is a prediction that the carry trade unwinds tomorrow, or violently, or at all on any particular schedule. Japan may engineer the most delicate normalization in monetary history; the yen may behave; the $20 trillion may bleed down gently over years with no crash at all. That benign path exists and should be stated honestly. But the risk is asymmetric and underpriced, because the trade is enormous, leveraged, crowded, undocumented, and built on an assumption — perpetually cheap yen — that is, for the first time in thirty years, demonstrably false. The cost of the world's cheapest money is rising. The single largest source of hidden leverage in global finance is being slowly, deliberately drained. And we already have the August 2024 footage of what it looks like when even a corner of it lets go.

The carry trade is the quietest large thing in the world, and quiet large things are exactly the ones that, when they finally move, move everything at once. For thirty years it was the floor under global asset prices, so reliable that the market stopped thinking of it as leverage at all. It is leverage. It always was. And the foundation is moving — not in a forecast, not in a thesis, but in the cold print of a thirty-year-high bond yield and a central bank that has, at last, started to take the cheap money away. The world borrowed yen to buy everything. Someday it has to buy the yen back. The only question that has ever mattered is whether it does so slowly, or all at once.