A plain-English guide to the carry trade.

A carry trade borrows in a currency with a low interest rate and invests in a currency with a high interest rate, pocketing the rate differential. The trader takes FX risk in exchange for a steady carry. When the FX rate moves favourably, the trader wins twice; when it moves adversely, the trader can lose more than the carry. USD/JPY is the canonical example because the US-Japan rate differential has been wider than any other G10 pair through the post-2022 cycle, and the Japanese yen has been the lowest-cost funding currency in G10 since the 1990s. With USD/JPY trading above 161 today and MoF unable to break the structural pull through verbal escalation alone, the carry trade is the dominant flow. This is the explainer on how it is constructed, the math behind it, and the three ways it historically ends.

The arithmetic of carry

Consider the canonical example: a trader borrows in JPY at Japan's 2-year rate (call it J, currently around 0.6 percent) and invests in USD at the US 2-year rate (call it U, currently around 4.4 percent). Ignoring spreads, the trader earns:

carry = U − J ≈ 380 basis points per year

That is the gross yield differential. From it, the trader must subtract the cost of the FX hedge if hedged, or take on the FX risk if unhedged. The unhedged version is the "real" carry trade; the trader receives 380 basis points of annual carry but is exposed to USD/JPY movement.

A useful way to think about the risk side: how much could USD/JPY fall before the trader is flat for the year? If the pair sits at 161, a 3.80 percent fall to roughly 155 would wipe out a year of carry. Anything beyond that is real loss.

The differential drives the level

Why the trade works on average

Currency carry has been a positively-rewarded strategy in the post-1980 G10 sample by roughly 4 to 6 percent annualised on an equal-weighted long-high-yield, short-low-yield basket. The return comes from the rate differential; the volatility comes from periodic FX moves against the carry. Two academic explanations dominate:

• The forward-rate bias. The forward FX rate, derived from interest-rate parity, says the high-rate currency should depreciate. In practice it does not, on average; it tends to appreciate slightly or hold flat. The gap between the parity-implied path and the realised path is the carry return. Why this gap persists is a deep academic question; the leading explanation is compensation-for-tail-risk.
• The risk-premium story. Carry traders take on left-tail FX risk that other investors don't want. The carry return is the compensation for that exposure, similar to selling out-of-the-money options. The pattern of returns (slow steady gains punctuated by sharp drawdowns) is consistent with this read.

The three ways a carry trade ends

1. Rate convergence. The high-rate central bank cuts, or the low-rate one hikes, or both. The differential narrows; the carry compresses; flows reverse. This is the slowest and most-controlled way for a carry trade to end. The 2024 BoJ rate normalisation (from 0 to 0.25 percent) was the start of a rate-convergence regime for USD/JPY; it has not finished.
2. Volatility spike. An exogenous risk event (financial crisis, geopolitical shock, central-bank surprise) pushes FX volatility higher. The risk-adjusted carry compresses; leveraged carry positions get stopped out. The 1998 LTCM episode, the 2008 yen rally, and the 2024 August yen squeeze are the three canonical examples. All three produced 5 to 10 percent USD/JPY rallies in the yen direction within days.
3. Intervention. The low-rate country's central bank or finance ministry directly sells dollars and buys the funding currency. We covered Japan's intervention framework in detail. The 2022 and 2024 MoF operations both produced 4 to 6 percent USD/JPY corrections; both reversed within 6 to 9 months as the underlying carry reasserted itself.

Why intervention alone cannot end the trade

The reason MoF has been reluctant to act this week despite step-five verbal escalation: the structural carry is too wide. At a 580-basis-point rate differential, an intervention that knocks USD/JPY down 4 percent is a single-cycle move; the carry trade rebuilds within months because the rate-differential cash flow is still attractive. The 2024 May intervention erased within 12 months. The 2022 intervention erased within 9.

The math is unforgiving. At 580 basis points of annual carry, the rate differential pays for a 4 percent intervention drawdown in eight months. Anything less than a structural narrowing of the rate gap leaves the trade intact. MoF can buy time. They cannot, by intervention alone, end the trade.

The components of an actual carry book

A working carry book is more sophisticated than the simple USD-vs-JPY example. A discretionary macro desk's carry-trade exposure typically combines:

• Spot positions: Long USD/JPY directly, sized to risk-budget.
• Forward overlays: Buying USD forward at the parity-implied rate, capturing the parity gap if the spot does not move as parity says.
• Options structures: Selling out-of-the-money USD/JPY puts or buying calls to express the carry view with limited downside. Risk reversals (selling puts, buying calls) capture the skew premium that funds the carry's tail risk.
• Basket diversification: Spreading the carry across multiple G10 high-low pairs (USD/JPY, AUD/JPY, NZD/JPY, MXN/JPY in the EM context) to reduce single-pair tail risk while preserving carry yield.
• Stop-loss discipline: Pre-defined exit levels (often 5 to 7 percent against position) to cap tail risk. The August 2024 yen squeeze stopped out most systematic carry strategies in a single session because stops triggered en masse.

How to read the carry trade from the price

1. USD/JPY level vs 2y rate differential. The chart above is the cleanest single read. If the pair sits at the level implied by the differential, the carry trade is in equilibrium; if it is below, the market is paying for tail-risk insurance; if above, the carry is being over-extracted.
2. One-month risk-reversal skew. The relative pricing of USD/JPY puts vs calls. When skew moves toward puts (yen-call premium), the market is pricing intervention risk; when toward calls, the market is pricing carry continuation.
3. Realised vs implied vol. When implied vol on USD/JPY rises above realised, dealers are hedging tail risk; that often precedes the trade ending. Today's one-month implied vol on USD/JPY at roughly 8.5 percent annualised is above the trailing realised of 6.5 percent; the option market is pricing some intervention risk, but not a full-tail event.

What the carry trade does not tell you

1. The terminal level. Carry trades have no inherent target; they end when the conditions that created them end. USD/JPY at 161 today implies the carry has further to extend if the conditions persist, not that it has reached a target.
2. The timing of the unwind. The historical pattern is "extends gradually, ends suddenly". The August 2024 episode unwound 5 percent in three sessions on a BoJ surprise. No model forecasts the timing reliably.
3. Whether intervention will succeed. MoF operations buy time but do not end the carry. Reading "MoF will act" as "carry trade will end" is the single most expensive misreading on a yen desk.
4. The cross-asset implications. Yen carry flows show up in cross-asset correlations in unexpected places: long yen-funded EM bond positions, long yen-funded US equity positions, long yen-funded gold positions. A yen squeeze produces forced selling in those assets, not just USD/JPY.

Related reading

• Today's analysis applies this directly: End of week: MoF talked, didn't act.
• The intervention framework: A plain-English guide to FX intervention
• The cross-asset framework: A plain-English guide to currency correlations
• The dollar-side story: A plain-English guide to the dollar smile