The Yen Carry Trade Is Unwinding. We Modeled the Contagion.

For fifteen years, the cheapest funding currency on earth quietly underwrote the most expensive assets on earth. That arrangement is now ending, and the exit is narrower than the entrance.

The mechanics are not controversial. Near-zero policy rates at the Bank of Japan made the yen the world's preferred funding leg: borrow in yen at a negligible cost, convert, and deploy into higher-yielding or higher-beta assets abroad — US equities, EM credit, the carry-rich corners of FX. The trade is profitable in two ways at once, through the rate differential and through the persistent depreciation of the funding currency. It is, in the language of our desk, a short-volatility position wearing the costume of a yield strategy.

As the BOJ normalizes — even gradually, even with heavy forward guidance — the differential compresses and the yen's depreciation thesis weakens. The carry begins to bleed. And because the position is held with leverage across thousands of uncoordinated balance sheets, the unwind does not arrive as a smooth repricing. It arrives as a scramble for the same exit. We saw a preview of this in August 2024: a 12% two-day drawdown in the Nikkei, a violent yen appreciation, and a volatility shock that propagated into US equities within hours despite no US-specific catalyst.

What makes this dangerous is the correlation structure. In calm regimes, the carry book looks beautifully diversified — Japanese funding, American equity, Australian rates, emerging-market credit, all apparently independent. In a forced-deleveraging regime, those positions reveal their shared DNA: they are all the same bet on continued liquidity and suppressed volatility. Correlations converge to one precisely when diversification is supposed to protect you. Standard risk models, calibrated on the placid periods, systematically underprice this convexity.

Our desk modeled six unwind paths, varying the speed of BOJ normalization against the depth of cross-asset liquidity. In the benign cases, the deleveraging is absorbed over quarters and the contagion stays contained to FX and Japanese equities. In the adverse cases — thin summer liquidity, a hawkish surprise, a coincident risk-off catalyst — the feedback loop is real: yen strength forces position cuts, position cuts pressure global equities, falling equities raise portfolio volatility, higher volatility forces further cuts. The tail is fat, and it is not symmetric.

So the honest conclusion of the analysis is that the contagion risk is real, non-trivial, and poorly hedged by conventional diversification. An investor who wanted to express caution here could reasonably reduce gross leverage, lift FX hedges, and avoid the crowded short-vol structures that sit at the epicenter of the unwind.

After integrating these dynamics into our proprietary 14-factor regime framework — stress-testing the allocation across all six scenarios, across liquidity states, and across a full distribution of normalization paths — the signal that emerges is unambiguous, and it is the same signal that emerges from every other analysis we conduct.

Nothing in the analysis argues with us.

The carry unwind is a sophisticated way of being exposed to global liquidity with leverage. Our position is a less sophisticated way of being exposed to global productivity without it. Across the six scenarios, the five hundred is the holding our framework recommends in five of them, and in the sixth the framework recommends it more emphatically. We do not find this monotony troubling. We find it clarifying.

← Back to Research