For thirty years Japan lent the world money at close to zero. Investors borrowed yen to buy higher-yielding assets, and Japan’s own savers followed them abroad. None of it has been shown to set the price of American capital. What it changes is how much capital is available — and how fast it can leave. The question this page asks is not whether the yen is fluctuating (short-term noise), but whether a move is persisting (regime hardening).
As of 7 Oct 26 · The carry is unwinding; whether it persists remains open.
The yen gained 37% against the dollar over nineteen months, 123 to 90, pausing once — alongside the credit crisis.
The Bank of Japan’s bond buying drove Japanese savings overseas; GPIF, the world’s largest pension fund, cut domestic bonds from 60% to 35%.
The Nikkei’s worst day since 1987. Markets recovered within weeks; the yen’s rally held three months, then reversed.
The carry rebuilt and the yen hit its weakest month-end since at least 2000 (162.61). Tokyo stepped in, then Washington.
For decades, Japan offered something unusual: money that was extremely cheap to borrow. Investors could borrow in yen, move that money into countries where interest rates were higher, and earn the difference. As long as Japanese rates stayed low and the yen stayed weak, the trade worked.
The risk appears when that relationship reverses. If Japanese rates rise or the yen strengthens, the trade becomes less profitable. Investors may sell the assets they bought and repay their yen borrowings. If enough investors do this at the same time, what began as a currency move can become a broader withdrawal of capital from global markets.
The fast channel is leveraged capital. Hedge funds and other investors can unwind positions quickly. The positions can be large: the BIS put the speculative yen carry trade entering August 2024 at roughly ¥40 trillion ($250 billion). Selling can hit public markets and reduce the capacity of institutional investors to make new commitments.
The slow channel is Japanese savings. Pension funds, insurers and other institutions have spent years investing outside Japan because returns at home were so low. The Bank of Japan pushed them further: in 2014, GPIF cut its allocation to domestic bonds from 60% to 35% and doubled its allocation to foreign equities. As Japanese rates rise, some of that incentive weakens. Capital that once had a strong reason to leave Japan may increasingly have a reason to stay home — or return.
The distinction is persistence. August 2024 created a sharp market shock without materially changing capital availability: our Capital Regime remained Capacity-Constrained from June through August, while Capital Availability moved only from −2.31 to −1.88. A sustained reversal is different. It can reduce the flow of capital into dollar assets and eventually change the conditions under which capital is allocated.
| Channel | How capital moves | Transmission |
|---|---|---|
| Fast · Leveraged carry | Investors borrow cheaply in yen and buy higher-returning assets elsewhere. When the economics reverse, positions can unwind quickly. | Public-market selling → lower portfolio values → less capacity for new commitments |
| Slow · Japanese savings | Institutions invest abroad when returns at home are unattractive. As Japanese yields rise, more capital can remain at home or return. | Less Japanese capital flowing abroad → lower foreign capital supply to U.S. markets |
Same origin, different clock. The leveraged trade can reverse in days; institutional allocation can take quarters or years. The first creates market shocks. The second can change the underlying availability of capital. Only a persistent move becomes a regime signal.
What the evidence does not show. Japan’s rising interest rates do not appear to directly set U.S. long-term rates. Our tests find Japan moving with the global rate cycle rather than leading it. That keeps the transmission mechanism where the evidence supports it: through capital flows, not directly through the price of U.S. capital.
The August 2024 unwind was fast and partial. The carry rebuilt. This time, however, the rate gap has narrowed from both ends and intervention has expanded from Japan to the United States. The trade remains profitable, but the conditions that sustained it are weakening — while governments are increasingly involved in managing the adjustment.
The yen carry trade sits outside the domestic liquidity measures that anchor the Capital Regime. But it can still change the capital available to the U.S. system: quickly through leveraged-market unwinds, and slowly through Japanese institutional flows.
The charts track observable carry incentives, yen positioning, and the persistence of currency movements. They do not measure the entire global yen carry trade or establish that Japanese rates determine the price of U.S. capital.
The history charts begin in January 2000, consistent with the Regime-Era Register, and use month-end observations. The panels below retain each source’s reporting frequency.
Era bands provide historical context; they do not establish causation. Shaded regions identify unwind episodes. Leveraged positioning measures the size of outstanding positions, not the direction of future currency movements. The six-month persistence threshold is a display convention, not a validated predictive signal; its predictive value remains untested on sealed, forward-looking data.
| Panel | Source / frequency | Construction and limits |
|---|---|---|
| USD/JPY — History and Defense Line | Federal Reserve H.10 noon buying rates in New York, via FRED · DEXJPUS · daily | Higher = weaker yen. History uses month-end values; the current month is partial until month-end. 160 is a market reference, not an official intervention threshold. Red marks Japanese intervention from the Ministry of Finance record. |
| Short-rate gap and Carry Spread | Federal Reserve Bank of New York via FRED · EFFR; Bank of Japan · uncollateralized overnight call rate · daily | EFFR minus the BoJ call rate, in percentage points: monthly averages in History, daily in Indicators. Each market’s holidays are carried forward onto the U.S. trading calendar. Market rate against market rate. |
| Yen Volatility | Federal Reserve Board (H.10) via FRED · DEXJPUS · daily | Realized volatility: the 21-day standard deviation of daily log returns, annualized (× √252), in percent. |
| Yen Persistence | Federal Reserve Board (H.10) via FRED · DEXJPUS · month-end | See Persistence below. |
| Leveraged Yen Position | CFTC Traders in Financial Futures · weekly | Leveraged-fund net position, futures + options combined, thousand contracts. See Leveraged position below. |
| August 2024 Unwind | Federal Reserve Board (H.10) · DEXJPUS; Nasdaq, Inc. · NASDAQ100; Cboe · VIXCLS (Copyright, 2016, Chicago Board Options Exchange, Inc. Reprinted with permission.) — via FRED · daily | Yen strength (USD/JPY inverted) and the Nasdaq-100, each indexed to 100 thirty trading days before 5 August 2024; the VIX close on 5 August 2024 appears in the label. Event study only; co-movement does not establish causality. |
| Japan vs U.S. 10Y Yields | Japan’s Ministry of Finance · JGB 10Y (issuer’s record); Federal Reserve Board (H.15) via FRED · DGS10 · daily | See Hypothesis tests below. |
| Japan · U.S. Treasury Flows | U.S. Treasury TIC · SLT Tables 3 and 5 · monthly | Japan’s net long-term Treasury flows and bills, $bn; valuation changes excluded. + = foreign position increased; the Grand Total ties to Treasury’s release. Arrives about seven weeks after the month. |
| Japanese Capital Abroad | Japan’s Ministry of Finance · weekly securities flows | Four-week rolling sum of net purchases of foreign equities and foreign long-term debt, ¥tn. Long-term only. No tested threshold. |
| Regime-Era Register strip | Capital Exponent · Regime-Era Register | Month-end. Matches the site’s master series in every overlapping month (320 months). |
The position is the CFTC’s leveraged-fund net position in yen futures and options combined, the fuller measure of the leveraged book. Futures-only reads −115k on 30 June, and 2015 was deeper.
One contract is ¥12.5 million, so −30k ≈ ¥0.38tn ≈ $2.4bn: a visible slice, on one exchange, of a speculative book the BIS put near ¥40 trillion ($250 billion). The dollar size uses USD/JPY on each report date.
The series is a deliberate exception to Layer 0’s exclusion of positioning: it measures the size of the leveraged book, not a direction signal.
Persistence counts the consecutive months in which the three-month change in the yen (month-end USD/JPY, in yen terms) has held its sign. The proposed threshold is six months. It is display-only until tested forward on sealed data. October is provisional until month-end.
Three pre-registered tests are cited on this page.
Japan’s rates against the U.S. term premium: retained, and labelled. Its follow-up, with German and UK 10-year yields as a global-rate control: a common global factor; the result is timing-sensitive, and the averages specification is used.
Japanese Treasury flows (TIC): not detected, and underpowered. The power branch is not robust, so the wording is chosen to hold under both branches.
USD/JPY 160 is the line markets watch for Japanese intervention. The 2024 and 2026 operations came with USD/JPY around 157–164; the first, in 2022, at 144–150. Japan’s Ministry of Finance decides; the Bank of Japan executes.
The page plots the Japan side only, from the Ministry of Finance record. For 30 July–26 August 2026 the period total is ¥15,399.3 billion; daily detail for that period arrives with the Ministry’s Q3 release in early November.
U.S. participation was reported on 3 August 2026 (Al Jazeera), and the FIMA push on 5 August 2026 (Bloomberg). The U.S. side is not plotted until a public primary record exists.
TIC data arrive about seven weeks after the month. Figures before 2023 are Treasury’s revised vintage and can differ from originally announced numbers.
The current month is provisional until month-end, in both the USD/JPY history and the persistence count.
In the frozen v1.0 point-in-time series, Rm moved from 5.91% to 5.78% and Capital Availability from −2.31 to −1.88 between June and August 2024. The regime was Capacity-Constrained throughout.
Data notices
This product uses the FRED® API but is not endorsed or certified by the Federal Reserve Bank of St. Louis.
The Effective Federal Funds Rate (EFFR) is subject to the Terms of Use posted at newyorkfed.org. The New York Fed is not responsible for publication of EFFR by Capital Exponent, does not sanction or endorse any particular republication, and has no liability for your use.
Disclaimer
Capital Exponent LLC provides research and business advisory services that are informational and educational in nature. Nothing published or communicated by Capital Exponent constitutes personalized investment advice, an offer or solicitation, or a recommendation to buy, sell, or hold any security or financial instrument. Capital Exponent LLC is not a registered investment adviser or broker-dealer. Investment and business decisions remain the sole responsibility of the reader or client. Past performance is not indicative of future results.