Japan's end of ultra-cheap money is reshaping the yen carry trade, global leverage, bond flows and the cost of capital.
For three decades the yen was more than a currency. From October 1995 to November 2025, the monthly average of the overnight call rate steered by the Bank of Japan (BOJ; 日本銀行) never rose above 0.55 percent, and for most of those 362 months it sat below 0.3 percent. With money that cheap, Japan became, in effect, a lender to everyone else. Investors borrowed yen at close to nothing, converted it into dollars, pesos or reais, and bought assets that paid more. The Bank for International Settlements (BIS) describes the yen as the predominant carry trade funding currency. Japan's monetary policy exported cheap money.
That arrangement is ending, slowly and on the record. On September 18, 2026, the BOJ's Policy Board voted 7-2 to guide the uncollateralized overnight call rate to around 1.25 percent, effective September 24, a level the call rate last averaged in June 1995. The 10-year Japanese government bond (JGB) yield reached 3.097 percent on October 2, its highest since August 1996, and the 30-year yield stood at 4.166 percent on October 5, the highest in a Ministry of Finance (MOF) series that begins in 1999. The risk is not necessarily a dramatic collapse of yen-funded positions. It is that gradual normalization changes the return calculations behind trillions of dollars of global capital, one marginal decision at a time.
The tempting conclusion, that the carry trade is finished, does not survive the data. The Federal Reserve (Fed) raised its target range to 3.75 to 4 percent on September 16, so the gap between the two policy rates is still about 2.6 percentage points. In real terms Japan's policy remains loose: with the national consumer price index (CPI) up 1.9 percent in the year to August, a simple ex-post real policy-rate measure is about minus 0.65 percent, while the Fed's midpoint sits roughly half a point above United States (US) inflation of 3.4 percent. The BOJ itself says that real interest rates "have remained at low levels" and that accommodative financial conditions will be maintained. This is the end of free money, not the end of cheap capital. What Japan is withdrawing is narrower and more consequential: an implicit subsidy to global leverage. Yen-funded strategies do not become impossible. They become less profitable, more sensitive to the currency, more demanding of capital and more volatile.
Seen this way, 1.25 percent matters more as a regime than as a number. Six moves, beginning with the end of negative rates in March 2024, have taken the policy rate from below zero to 1.25 percent, and the September statement says the BOJ "will continue to raise the policy interest rate". The Summary of Opinions released on October 1 is blunter in places: one member argued that the BOJ "will need to accelerate the pace of rate hikes" if prices deviate upward, and another noted that the interval between hikes "will be shorter than before." The next meeting, with a new Outlook Report, is scheduled for October 29-30 as of October 6. The relevant question for a borrower is therefore no longer whether 1.25 percent is high. It is whether Japan is still a reliable source of near-zero funding. On the BOJ's own guidance, it is not.
How much money depends on the answer is harder to say than headline figures suggest. The BIS locational banking statistics show that reporting banks held US$2.26 trillion of cross-border claims denominated in yen at the end of March 2026, about ¥361 trillion at the BOJ's end-March rate of ¥159.63 per dollar. That is the source of the widely quoted ¥360 trillion. It is an outer envelope, not a measure of speculation, because it includes interbank and intragroup lending and ordinary bank loans abroad. The narrower BIS global liquidity indicator, yen credit to non-bank borrowers outside Japan, stood at ¥65.9 trillion (about US$418 billion at ¥157.58, the BOJ's October 2 rate) and was up just 0.3 percent from a year earlier, after three quarters of decline. When BIS economists tried to size the carry trade itself before the 2024 turmoil, they arrived at ¥40 trillion (US$250 billion) as a "rough middle ballpark" and warned that the estimate was biased down by data gaps, because much of the activity runs through currency forwards, swaps and options that never touch a bank balance sheet. The yen carry trade is not one trade on one ledger. It is a financial architecture built on the assumption that yen funding will stay cheap and stable, and its true size cannot be read off any single table. One comparison gives the scale: Japan's own holdings of foreign long-term debt securities were ¥360.5 trillion at the end of 2025, roughly the same figure, held on entirely different balance sheets.
The mechanics explain why the regime matters more than the level. Let be the number of yen per dollar, so a rise in is a weaker yen. A trader who borrows one yen at , converts it, invests at the foreign rate and converts back earns
A weakening yen () adds to the return; a strengthening yen subtracts from it. Writing expected yen appreciation as , with , the attractiveness of the trade before it is placed is
where is hedging and transaction cost and is the premium demanded for volatility and crash risk. Carry loses appeal when the rate gap narrows, expected appreciation rises or volatility raises . Covered interest parity explains why the currency term cannot be engineered away. A forward contract prices the rate gap into the exchange rate, so a fully hedged position earns approximately
little beyond the cross-currency basis . Unhedged carry is therefore a bet on the exchange rate, and a hedged foreign bond is only as attractive as its hedge is cheap.
Current numbers make the bet concrete. On October 1, the one-year Treasury yield was 4.44 percent and the one-year JGB yield 1.668 percent, a gap of 2.77 points, with the dollar at ¥158.39. The yen appreciation that erases a full year of that carry is
a move to about ¥154.2 per dollar. The yen was stronger than that, at ¥153.12, as recently as September 9. Leverage magnifies the asymmetry. With five units of dollar assets for every unit of equity and four units of yen debt, a sudden yen appreciation of changes equity by
so a 5 percent move removes about 24 percent of equity, roughly a year and a half of the 15.5 percent annual return that the same leverage would earn if the exchange rate stood still. The rate gap still pays, but far less per unit of risk. Dividing the three-month rate differential by the realized volatility of the dollar-yen rate over the previous three months gives a carry-to-risk ratio of about 0.74 in early July 2024 and 0.29 on October 1, 2026, a decline of about 60 percent. The BIS found that this ratio peaked in the first quarter of 2024, before Japanese tightening began.
What happens when that ratio collapses suddenly was demonstrated in the summer of 2024. The BOJ raised its rate to 0.25 percent on July 31, 2024. Two days later a weak US payrolls report landed on crowded positions. Between July 11 and August 5 the yen strengthened by 12.6 percent against the dollar, equal to about two and a half years of the carry then available, in under four weeks. On August 5 Japan's Tokyo Stock Price Index (TOPIX) fell 12 percent and the Cboe Volatility Index (VIX) spiked above 60 in off-hours trading; the BIS analysis finds that deleveraging and rising margins amplified the reaction to a single data release, and notes that the Mexican peso, the Brazilian real and the South African rand were the high-yield currencies hit hardest. On August 7 Deputy Governor Uchida Shinichi said that the BOJ "will not raise its policy interest rate when financial and capital markets are unstable". By the end of that week the S&P 500 had recovered its losses since Monday, and the BIS judged that exchange-rate moves had not been outsized by the standard of earlier carry crashes.
The 2026 record reads less like a repeat than a slow-motion version of the same adjustment. Speculative positioning in yen futures, which the BIS calls "only the tip of the iceberg," shows the pattern. Non-commercial traders on the Chicago Mercantile Exchange were net short 163,412 contracts on July 28, 2026, about ¥2.04 trillion and not far below the ¥2.30 trillion peak of July 2024; by September 29 they were net long 55,440 contracts, according to Commodity Futures Trading Commission (CFTC) data. That swing of about ¥2.7 trillion (US$17 billion) coincided with a yen appreciation of 6.2 percent between July 1 and September 9, spread over ten weeks rather than compressed into a weekend, and the yen actually weakened after the September hike. The evidence is consistent with an unwind that is happening through the calendar rather than through a crash, although futures data cannot show what is happening in the far larger over-the-counter market.
The more consequential adjustment may be taking place inside Japan. Banks and other financial corporations hold about 95 percent of Japan's foreign long-term debt securities, and for them the relevant comparison is not a rate gap but a hedged yield against a domestic one. Using the three-month rate differential of 2.67 points on October 1 as the hedging cost, and ignoring the cross-currency basis,
so a currency-hedged 10-year Treasury yields about half a point less than a 10-year JGB, and at 30 years the shortfall is about 1.2 points. This is not new: the hedged pickup has been negative in every month since October 2022, when Fed tightening drove hedging costs above the yield gap, and it was worse in 2023. What normalization has changed is the alternative. The unhedged gap between US and Japanese 10-year yields has halved, from a monthly average of 3.98 points in October 2023 to 2.00 points in September 2026, and a domestic yield above 3 percent now gives Japanese institutions something to buy at home without taking any currency risk. An unhedged investor in 10-year Treasuries now loses the pickup if the yen appreciates by just over 2 percent in a year.
Flow data show who is responding. MOF portfolio statistics record net sales of foreign long-term bonds by Japanese residents of ¥2.76 trillion in January-August 2026, against net purchases of ¥11.0 trillion in the same months of 2025. The selling is concentrated: banks' own accounts sold a net ¥9.25 trillion (about US$59 billion) and life insurers ¥1.47 trillion, while trust accounts, commonly used as a proxy for pension money, bought ¥10.22 trillion. The pension side is rule-bound: the Government Pension Investment Fund (GPIF) targets 25 percent in foreign bonds and held 24.60 percent at the end of June, so deviations below its target can create a mechanical rebalancing demand for foreign bonds. Weekly data continue the theme, with net sales of ¥1.90 trillion in the week of September 13-19 and ¥684.5 billion the following week. Hedge funds attract the headlines, but domestic balance sheets are larger and slower, and once they turn, they tend to stay turned.
The bond market is where the domestic and global stories meet. Higher JGB yields make Japanese assets more attractive to Japanese savers, which reduces the incentive for capital to leave, which removes a marginal buyer from foreign bond markets, which can push global yields and risk premiums up at the margin. Each link is plausible; none is mechanical. Japan held US$1,103.9 billion of US Treasury securities in July 2026, still the largest foreign holding at 11.9 percent of the total, but US$135 billion less than in February, a decline that partly reflects lower bond prices. The US 10-year yield, 5.31 percent on October 5, has its own domestic drivers in a hiking Fed and 3.4 percent inflation, so the data cannot attribute its rise to Japan. What can be said is that one of the most dependable buyers of foreign duration now has a credible reason to buy less.
The same yields raise the price of Japan's own debt. The International Monetary Fund (IMF) puts gross public debt at 206.8 percent of gross domestic product (GDP) in 2025 and 202.9 percent in 2026. The fiscal 2026 budget, as drafted, allocates ¥31.28 trillion to debt service, 25.6 percent of spending, including ¥13.0 trillion of interest, calculated on an assumed long-term rate of 3.0 percent that the 10-year yield now slightly exceeds. Prime Minister Takaichi Sanae told the Diet on October 5 that the government would appropriately control full-year JGB issuance while pursuing what she calls responsible, proactive fiscal policy. Fiscal credibility is now part of the price of yen funding, because it shapes the term premium demanded at home.
For leveraged investors outside Japan, the implication is a change in the cost and fragility of positions rather than their disappearance. High-yield currencies such as the peso and the real, emerging-market local bonds and leveraged equity strategies were the first casualties of the August 2024 squeeze, and the BIS also recorded losses of up to 20 percent in Bitcoin and Ethereum as retail traders met margin calls. None of these markets depends on yen funding alone, but a funding currency that can appreciate several percent in a few weeks forces leveraged holders to post more margin, hold smaller positions or demand higher returns. That is how a modest change in Japanese rates travels into asset prices far from Tokyo without a single dramatic event.
The chain that follows is gradual by construction. The BOJ raises rates in small steps; yen funding becomes dearer; the expected cost of currency moves rises as the yen becomes less one-directional; leveraged positions offer less return per unit of risk, so some are trimmed; demand for yen-funded foreign assets eases while demand for yen rises; foreign asset prices face marginal pressure; global yields and risk premiums adjust. The strongest objection is that none of this has happened in a way that markets have punished. The differential still exceeds 2.5 points, the yen fell after the September hike, Japanese real rates are negative, and the Uchida principle suggests that the BOJ will pause whenever markets wobble, which caps the probability of a disorderly unwind. That objection has merit, and it is why a crash is the wrong base case. It does not defeat the argument, because the argument concerns the denominator rather than the numerator: carry per unit of risk has fallen by more than half, and the domestic alternative for Japanese institutions has improved for the first time in a generation.
Several outcomes are likely. The BOJ is likely to continue moving in small steps, with the October 29-30 meeting live as of October 6; JGB yields are likely to stay well above their pre-2024 range; and Japanese banks and insurers will probably keep favoring domestic bonds at the margin while the GPIF continues to rebalance mechanically. Another August-2024-style shock remains possible if a US growth scare coincides with yen strength, although speculative futures positioning has already swung to net long, leaving less fuel than in 2024. It is uncertain whether Japanese repatriation will be large enough to lift global term premiums measurably, whether fiscal pressure will push the long end of the JGB curve higher regardless of BOJ policy, and how much carry exposure sits in derivatives that no public dataset captures. Investors and policymakers may need to treat the yen less as a funding leg and more as a risk factor.
Japan's departure from the zero-rate world marks the gradual removal of one of the most persistent sources of cheap global funding. Carry economics are deteriorating, not because the rate gap has vanished but because it is narrower, less dependable and more volatile. Japanese investors are reassessing foreign assets, starting with the banks and insurers whose choices move slowly and last. Leveraged portfolios have become more sensitive to the currency that once seemed the safest leg of the trade. The result need not be a single shock. It is more likely to be a prolonged repricing of what capital costs and where it flows, set by a country that for thirty years charged almost nothing for it.
Apple News, Google News, Feedly, Flipboard, and WhatsApp Channel
