The Quietest Meeting of the Week
Two central banks met this week, half a world and one war apart. On Wednesday, the Federal Reserve held its target range at 3.50 - 3.75% and devoted part of its statement to the obvious: "the implications of developments in the Middle East for the U.S. economy are uncertain." On Thursday, the Bank of Japan held its policy rate at 0.75% - by an 8 - 1 vote, with the dissenting board member formally proposing a hike to 1.00%.
The Fed made every front page. The BoJ barely made the tape. Markets spent Friday watching oil instead, and the S&P 500 fell again as crude climbed.
Here is why Thursday's non-event deserves more attention than it got: the dollar closed the week at 159.23 yen - less than 2% from 161.95, the yen's weakest level since 1986, and the exact neighborhood where the most violent market unwind of this decade began nineteen months ago.
We used Barebone AI to rebuild the entire timeline of the yen carry trade - every policy decision, yen level, and unwind episode from Japan's bubble years through Thursday's vote - and to measure what the last rupture actually did to markets.
The headline numbers: $6.4 trillion erased from global equities in three weeks. A 12% one-day crash in Tokyo, the worst since 1987. And then - the part everyone forgets - a full recovery in the S&P 500 within four months. That's the precedent. Here's the mechanism.
Three Decades of Free Money
In the early 1990s, Japan's twin bubbles in stocks and real estate burst, and the economy slid into a balance-sheet recession that lasted a generation. The Bank of Japan responded by cutting rates to the floor - and then leaving them there. Zero in the late 1990s. Negative (−0.1%) from 2016 until March 2024. Today's 0.75% is, remarkably, the highest Japanese policy rate since 1995.
That is the entire range of Japanese monetary policy for thirty years: zero, give or take three-quarters of a point.
Three decades of free money created what the trading world treats as a cheat code. Borrow where money costs nothing, invest where it pays something, pocket the difference. That strategy - funding positions in a low-rate currency to buy higher-yielding assets elsewhere - is a carry trade, and the yen became the world's favorite currency to do it in.
Borrowed yen flowed into Treasuries, into European bonds, into Mexican pesos, and into US tech stocks. Quietly, Japanese liquidity became part of the oxygen supply of global markets.
How the Trade Works - and Why Nobody Knows Its Size
The mechanics are simple. Borrow yen at a fraction of a percent. Swap them into dollars. Buy something that yields more - the 10-year Treasury closed Friday at 4.39% - or something that you hope compounds faster, like AI-era tech equities. At the peak of the gap, with the Fed at 5.50% and the BoJ below zero, the riskless spread between the two policy rates was 5.6 percentage points. Add the leverage that hedge funds routinely apply through FX forwards and futures, and a 5-point spread becomes a double-digit return.
The catch: the whole structure is a short position against the yen. If the yen strengthens faster than the spread accrues, the trade loses money - and because it's levered, it doesn't lose money politely. Margin calls force funds to sell whatever the borrowed yen bought, which weakens those assets, which triggers more selling, which strengthens the yen further as positions are bought back. The exit is always smaller than the room.
How much money is in the room? That's the uncomfortable part - nobody knows. The estimates that circulated after the 2024 blowup spanned two orders of magnitude:
| Measure |
Estimate |
Who counted it |
| Speculative yen FX forwards (hedge funds) |
~$160B |
BIS proxy, 2024 |
| Yen bank lending to non-banks outside Japan |
~$880B (¥133T) |
BIS, Q1 2024 |
| Dollar - yen carry positioning at peak |
$500B+ |
UBS estimate, 2024 |
| Japan's public-sector "government carry trade" |
~$24T |
Deutsche Bank framing |
A range that wide isn't sloppy research - it's the finding. The carry trade isn't one position on one exchange; it's a funding pattern threaded through banks, hedge funds, insurers, and Japan's own government balance sheet. Nobody can see all of it. Including the people in it.
Three Days in August
The script's worth of history you need is August 2024. The sequence:
| Date |
What happened |
| Jul 31, 2024 |
BoJ surprise hike, ~0.1% → 0.25% |
| Aug 2, 2024 |
US payrolls: +114k jobs vs ~175k expected |
| Jul 29 - Aug 5, 2024 |
Yen rallies more than 6% against the dollar |
| Aug 5, 2024 |
Nikkei falls 12%+, worst day since 1987; S&P 500 −3%; VIX spikes to its highest since COVID |
Note what the trigger was not. The BoJ's hike was 15 basis points - a rounding error by Fed standards. What moved was the expected path: a Japanese central bank suddenly hiking into a US economy that suddenly looked weak enough to force Fed cuts. The rate gap was narrowing from both ends at once, and every levered yen borrower headed for the exit in the same week.
By Bloomberg's tally, $6.4 trillion was wiped from global stock markets in the three weeks into August 5. JPMorgan estimated that by mid-August, 65 - 75% of global carry positioning had been unwound. For a strategy most retail investors had never heard of, it was a remarkable demonstration: a 15-basis-point move in Tokyo helped erase more value than most countries' entire stock markets are worth.
The Round Trip
Here is the part the 2024 post-mortems didn't predict: the trade came back.
The dollar - yen rate tells the whole story in one line. The run to 161.95 in July 2024. The violent snap to 139.58 by mid-September. And then a nineteen-month grind right back to where it started - 159.23 at Friday's close, with a print just shy of 160 on Wednesday.
Why would traders rebuild a position that just blew up? Because the math still works. Even after BoJ hikes and Fed cuts, the policy gap is roughly 3 percentage points - half its peak, but still real money once levered. And because Japan handed the trade a new tailwind: a government that called a snap election on a platform of tax cuts and bigger deficits, and backed a ¥21.3 trillion stimulus package.
Prime Minister Sanae Takaichi pledged to "break free of the spell of excessive fiscal austerity."
Markets took her at her word. Fiscal expansion plus a cautious central bank is a recipe for a weaker yen - and superlong Japanese government bond yields have spent much of the past year around record highs as investors reprice Japan's debt load. The 10-year JGB now yields more than 2% for the first time in a generation. The yen weakened anyway.
The Gap Is Half Closed
The carry trade's fuel is the rate differential, so track the two engines:
|
Peak (Jul 2023 - Mar 2024) |
March 20, 2026 |
| Fed funds (upper bound) |
5.50% |
3.75% |
| BoJ policy rate |
−0.10% |
0.75% |
| Policy gap |
5.60 pts |
3.00 pts |
The BoJ's December hike to 0.75% - its third since exiting negative rates - was the move that pushed JGB yields through 2%. Since then it has held twice, in January and again on Thursday. Governor Kazuo Ueda's stated reason for waiting: the bank updates its quarterly forecasts in April, and it wants new readings of underlying inflation that strip out government subsidies, which have flattered the headline numbers. January's headline CPI cooled to 1.5%, the slowest in two years - but officials themselves treat that figure as distorted.
The pressure to keep hiking hasn't gone anywhere. Japan's unions entered this spring's shunto wage round demanding almost 6%, after two consecutive years of settlements above 5% - exactly the wage-price dynamic the BoJ spent thirty years praying for. One board member already voted for 1.00% this week. The consensus among forecasters puts the next hike in the second quarter, as early as the April 27 - 28 meeting.
Meanwhile the Fed has cut three times since September and is now holding, with a war premium working its way into oil. The gap is closing from both ends - slowly, predictably, the way it's supposed to. August 2024 is what happened when it closed suddenly. That is the distinction that matters from here.
The Honest Caveat: The Crash Was a Buying Opportunity
If the article stopped here, it would read like a warning to sell everything before Japan hikes. The full data says something more interesting.
In our 30-year study of VIX spikes, we classified the August 2024 unwind as a knee-jerk panic - a violent positioning flush with no underlying economic break - rather than a systemic crisis. The aftermath made the case emphatically:
Buying the S&P 500 at the close of the worst day - August 5, 2024 - returned +11.5% in three months, +16.9% in six, and +21.5% over the following year. The investors hurt worst weren't the ones who held through it; they were the levered ones forced to sell into it, and the panicked ones who joined them.
Two more honest complications. First, the "great unwind" thesis has been wrong for nineteen months and counting - strategists who declared the carry trade finished in August 2024 watched it get rebuilt at a 3-point spread. A trade that pays you daily to stay in it doesn't die of old age. Second, the same opacity that makes the trade scary cuts both ways: if nobody can measure it, nobody can prove it's dangerously large right now either.
The asymmetry to respect is what an unwind collides with. August 2024 hit a fundamentally healthy US economy, so it stayed a knee-jerk panic. The same FX shock arriving alongside a genuine US slowdown - say, an oil-driven one - would belong to the other category in our study: systemic episodes, where buying the dip won only 40% of the time at 12 months. Same trigger, different world, opposite outcome.
What to Watch From Here
No directives - just the dashboard we'd watch:
The April 27 - 28 BoJ meeting. New quarterly forecasts, new subsidy-stripped inflation estimates, and the first realistic window for a hike to 1.00% - a one-handle on Japanese rates for the first time in three decades. Thursday's 8 - 1 vote says the debate is already live inside the board.
The speed of the yen, not the level. The 2024 lesson: the level (161, 159, 155) matters less than velocity. A 6% yen rally inside a week is the signature of forced unwinding. Slow appreciation is digestion; fast appreciation is margin calls.
Shunto settlements through the spring. Wage deals above 5% for a third straight year lock in the BoJ's case that inflation is finally homegrown - and keep the hiking cycle alive regardless of who holds the prime minister's office.
The second engine. August 2024 required two things at once: a hawkish Tokyo surprise and sudden US growth fear. The first ingredient is scheduled; the second is what oil above its March highs would supply. Watching either alone misses the mechanism.
Where Japanese money goes when home pays. With 10-year JGBs above 2% and superlong yields at records, Japanese institutions - for decades the world's most reliable buyers of foreign bonds - finally earn a real return at home. Repatriation is the slow-motion version of the carry unwind: no crash, just a steady bid disappearing from Treasuries and global risk assets.
The yen carry trade spent thirty years becoming the plumbing of global markets, broke spectacularly once, and was rebuilt within a year and a half at the same exchange rate. The question isn't whether the Bank of Japan finishes the job it started - one of its own board members just voted to. The question is how much of the world's risk is still quietly funded in yen when it does.
Data: Barebone | Sources: Bank of Japan Statement on Monetary Policy (March 19, 2026), Federal Reserve FOMC statement (March 18, 2026), BIS Bulletin No. 90 and BIS Quarterly Review (September 2024), Rengo/shunto coverage via Japan Times (March 2026) | Data as of March 20, 2026 close
