Japan’s GPIF lost ¥5.25 trillion as yen strength and rising yields fuel debate over pension allocation, market risk and political pressure.
A stronger yen was among the outcomes Tokyo sought this summer, and Japan's Ministry of Finance (MOF; 財務省) spent ¥15.4 trillion buying yen on July 30, July 31 and August 3. The intervention campaign included a coordinated operation with the United States on July 31. The same currency move helped push the country's public pension reserve fund into a loss. The Government Pension Investment Fund (GPIF; 年金積立金管理運用独立行政法人), one of the world's largest pension funds, reported on November 6 that it lost ¥5.25 trillion, a return of −1.65 percent, in July–September, the second quarter of the fiscal year that runs from April 2026 to March 2027. At ¥158.19 per US$1, the Bank of Japan's (BoJ; 日本銀行) Tokyo 17:00 rate for November 5, 2026, the loss equals about US$33.2 billion. GPIF President Kazuto Uchida attributed the result to bond yields rising at home and abroad and to a yen that strengthened against the dollar and the euro, partly as a result of yen-buying intervention by Japan. The fund's FY2026 Second-Quarter Investment Results report provides the supporting data, including returns across the four asset classes and movements in government bond yields and exchange rates.
The loss arrived after four months in which senior members of the government had called for measures to encourage GPIF to put more of its ¥314.4 trillion (about US$2.0 trillion) into Japanese assets, and markets had been watching for any sign of a tilt. None appeared. At the end of September the fund held 26.49 percent in domestic bonds, 24.05 percent in foreign bonds, 24.80 percent in domestic equities and 24.67 percent in foreign equities, each close to its 25 percent target. This analysis argues that the quarter settles the short-term question and sharpens the long-term one. The foreign-currency exposure that produced most of the loss is the same exposure that advocates of buying at home want reduced, though their reasons range from pension risk to support for the yen and demand for government debt. Higher Japanese yields have given the pension case for a larger domestic bond share real substance. That case can be settled legitimately only inside GPIF's own verification process, on the return and risk terms that process sets. At GPIF's scale, any shift would be read in currency and bond markets as policy, whatever its stated reason.
Before the politics, the arithmetic of the loss shows how concentrated it was. Foreign bonds lost ¥5.23 trillion, or 6.52 percent, which by itself accounts for 99.6 percent of the fund's net loss. Domestic bonds lost ¥1.45 trillion and foreign equities ¥1.46 trillion, while domestic equities earned ¥2.88 trillion, a 3.71 percent return. Interest and dividends still brought in ¥1.30 trillion during the quarter, a reminder that the loss was a fall in market values while the portfolio's income kept arriving.
Chart 1: Foreign bonds produced almost the entire quarterly loss. Source: GPIF, Investment results for 2Q of fiscal 2026, November 6, 2026; preliminary figures, gross of fees. Data cut-off: November 6, 2026.
Separating the foreign bond result into its currency and market components suggests that the stronger yen accounted for a substantial share of the loss. GPIF's market table puts the yen at ¥162.53 per dollar at the end of June and ¥157.14 at the end of September, and at ¥185.82 and ¥178.48 per euro, respectively. These movements provide an illustration of the currency translation effect on foreign assets, although they do not capture the index's full currency composition. Measured in yen, a foreign asset's return combines its local-currency return with the exchange-rate movement:
: return measured in yen; : local-currency return; : currency return (yen appreciation against the dollar or euro)
GPIF's foreign bond benchmark, the FTSE World Government Bond Index excluding Japan and China, returned −6.65 percent in yen. Solving for the local-currency part gives , or between −2.8 and −3.5 percent depending on whether the euro or the dollar move is used. These are IndraStra estimates, since the index holds bonds denominated in more currencies than these two, and the calculations do not account for their respective weights. The results nevertheless illustrate how both yen appreciation and declines in underlying foreign bond prices contributed materially to the quarterly loss. The market component reflected a broader repricing of interest-rate, inflation and fiscal risks across major economies. The policy environment also tightened during September: the European Central Bank raised its key rates on September 10, followed by the Federal Reserve on September 16 and the Bank of Japan on September 18, when it increased its policy rate to around 1.25 percent. These decisions formed part of the wider backdrop to rising bond yields, but should not be treated as their sole cause. Long-term yields also reflected changing inflation expectations, government borrowing requirements, fiscal uncertainty and investor demand for duration.
A single losing quarter needs the context of the fund's longer record, which remains strong. Even after July–September, GPIF's return for the first half of the fiscal year was 6.42 percent, or ¥18.84 trillion, because April–June had produced 8.20 percent. Since it began investing in markets in fiscal 2001, the fund has earned cumulative returns of ¥215.77 trillion (about US$1.36 trillion), an annualized 4.83 percent, and it has absorbed seven losing years along the way. Nothing in the quarter points to a failing strategy; the live debate concerns which risks the fund should carry, and who should decide.
Who decides is set by a framework the quarterly numbers show GPIF following closely. Its fifth medium-term plan, in force since April 1, 2025, fixes 25 percent for each of the four asset classes, with permitted deviations of six percentage points for domestic bonds, domestic equities and foreign equities, five points for foreign bonds and nine points for bonds or equities as a whole. The September figures used a sliver of that room. The rise in the domestic bond share, from 25.59 percent in June, is partly seasonal; the share tends to increase in March and September because of cash management in the pension system. It also includes about ¥5.4 trillion of reserves held in the government's pension special account, which GPIF counts as domestic bonds. Removing that amount from both sides gives
Numerator: domestic bonds minus the ¥5.4 trillion special-account reserves counted as domestic bonds; Denominator: total reserves minus the same special-account amount.
a figure in line with the roughly 25 percent reported after the release. Koji Okuda of the Daiichi Life Research Institute put it plainly: "It appears GPIF is barely making use of the permitted deviation ranges from its model portfolio."
That restraint stands out against the political pressure that built through July. Finance Minister Satsuki Katayama told reporters on July 10 that she wanted to pursue measures that would encourage pension funds, starting with GPIF, to invest more in Japanese financial assets, allowing citizens to share in the benefits of economic growth as Japan moved beyond its era of near-zero interest rates. Four days later she argued that the existing 25 percent policy portfolio should not be treated as immutable and suggested that if the government's growth strategy made yen assets more attractive, a review and revision could follow. She also acknowledged that GPIF operated under its own rules and that the health ministry would need to be consulted. On July 17, Prime Minister Sanae Takaichi told the upper house budget committee that it was important to pursue measures encouraging further investment in Japanese financial assets by households and by pension funds including GPIF.
Pushback came in the Diet and from the ministry that oversees the fund. In the upper house finance committee on July 16, Isamu Ueda of Komeito told Katayama it would be inappropriate for the finance ministry or politicians, lacking investment expertise, to steer where the fund invests, and warned that returns could suffer. The minister who actually oversees GPIF was cooler still. Health Minister Kenichiro Ueno said on July 14 that GPIF must invest solely in the interest of the insured, from a long-term perspective, that he did not consider the investment environment to have diverged greatly from the portfolio's assumptions, and that GPIF reviewed the portfolio every year and would consider changes if needed. He pointed instead to the fund's growing investments in domestic private equity and other alternative assets, which are capped at 5 percent of the portfolio and stood at 1.72 percent in September.
GPIF's management committee records indicate that the portfolio verification process received renewed attention during the period of growing political pressure. On March 6, GPIF's management committee heard from its policy portfolio verification project team that no review of the portfolio was needed, and on May 22 the committee voted unanimously that neither further verification nor a review was required for fiscal 2025. Then, on August 21, the committee met with a report on the project team's discussions listed first among its reporting items. The Daiwa Institute of Research noted that an August meeting was the first in seven years, the last having been held in 2019, when the five-yearly portfolio review was under discussion. It also cautioned that the committee may simply have wanted a record showing that it had examined the question.
On September 8, Ueno said GPIF was continuing to verify the portfolio in a timely and appropriate way, while repeating that conditions had not diverged greatly. The agendas for the September 28 and October 23 meetings carry no further project team item, and as of November 9 GPIF had not posted a summary of the August discussion on its meeting page.
The pressure would be easier to dismiss if the market environment had stood still, and the yield case for Japanese bonds has changed sharply since the current mix was designed. When GPIF set 25 percent for each asset class in April 2020, the 10-year Japanese government bond (JGB) yield was 0.03 percent at the end of March. When it kept that mix for the plan starting April 2025, the yield was 1.50 percent at the end of March 2025. On September 2, 2026, it closed at 3.006 percent, the first close above 3 percent since September 1996 in the finance ministry's daily series, and it peaked at 3.11 percent on October 7 before falling back to 2.70 percent on November 6. GPIF's own history shows how much yields matter to the mix. In 2014, after the health minister asked GPIF to bring its review forward, the fund cut its domestic bond target from 60 percent to 35 percent, judging that a bond-heavy portfolio could no longer meet the required real return. In 2020 it cut the target again, to 25 percent, as domestic yields fell further.
The strongest argument for a higher domestic bond share is therefore one made on pension grounds, and it deserves to be taken seriously. GPIF's mandate is to earn a real return, meaning the return above nominal wage growth, of 1.9 percent at minimum risk. The Daiichi Life Research Institute has calculated that the current mix assumes domestic bonds will earn a real return of about −0.8 percent. If that improves to about −0.6 percent, a mix with 30 percent in domestic bonds would meet the 1.9 percent target; at about −0.4 percent, the 2014-style 35 percent mix would roughly meet it. Pensions are paid in yen, and moving money from foreign bonds into JGBs would cut the fund's currency risk while leaving equities untouched. The July–September quarter is exactly the kind of period in which such a mix would have lost less. These points have merit, and a fund that ignored a doubling of domestic yields since it last reaffirmed its mix would be failing its own process.
The case has limits that the same evidence exposes. GPIF holds foreign bonds partly because, at its size, doing so avoids distorting the domestic bond market, a consideration that grows as the reserves grow. A higher JGB share also narrows the international diversification of the bond portfolio, and a portfolio review requires updating expected returns, risks and correlations for all four asset classes together, not raising one assumption because one yield has moved. The yield itself has proved unstable: the 10-year rate gave back about 0.4 percentage points between its October peak and November 6. A portfolio redesigned on the strength of a few months of yields would be exposed to the next reversal, and higher yields carry a risk of their own, since rising rates are what produced the domestic bond loss in the quarter just reported.
Those objections weigh more as the sums grow, and size is what turns a portfolio question into a market event. Total pension reserves stood at ¥319.78 trillion on September 30, so each percentage point of reallocation moves about ¥3.2 trillion. The table compares the main routes with the intervention that preceded the quarter's yen rally.
| Route to more domestic assets | Shift (percentage points) | Amount (¥ trillion) | Amount (US$ billion) |
|---|---|---|---|
| One percentage point of reserves | 1.0 | 3.2 | 20 |
| Domestic bonds to the top of the band (26.49% to 31%) | 4.5 | 14.4 | 91 |
| Policy target for domestic bonds raised to 30% | 5.0 | 16.0 | 101 |
| All domestic assets to the theoretical maximum (51.29% to 61%) | 9.7 | 31.1 | 196 |
| Policy target for domestic bonds raised to 35% | 10.0 | 32.0 | 202 |
| Memo: MOF yen-buying intervention, July 30–August 3, 2026 | n/a | 15.4 | 97 |
Table 1: A modest change in GPIF's mix rivals a quarter's intervention. Sources: IndraStra calculations from GPIF reserves of ¥319,780.2 billion at September 30, 2026, GPIF deviation bands and MOF intervention data; dollar amounts at ¥158.19 per US$1 (BoJ, November 5, 2026). The 61 percent maximum follows from the minimum foreign weights (20 percent for bonds, 19 percent for equities). Data cut-off: November 6, 2026.
The amounts in the table follow from one line of arithmetic,
: amount moved (in yen or dollars); , : new and current portfolio weights; : total pension reserves (¥319.78 trillion at end-September 2026)
and they show why the motive behind any change matters. Raising the domestic bond target to 30 percent, entirely at the expense of foreign bonds, would require selling roughly as much foreign currency as the finance ministry sold in its three days of intervention. In September 2025, the United States Treasury and Japan's finance ministry agreed that government investment vehicles such as pension funds would continue to invest abroad for risk-adjusted return and diversification purposes, not targeting exchange rates for competitive purposes. Asked in July whether her remarks were consistent with that statement, Katayama said nothing about it had changed. A reallocation justified by pension return and risk would fit that commitment. One that looked timed to the yen or to JGB auctions would test it, and would also invite the question Ueda raised about who is really managing the money.
The state's other actions do not all point in the direction the July remarks suggested. The BoJ decided on October 30 to more than double the pace of its sales of exchange-traded funds (ETFs) to about ¥860 billion a year at book value, or roughly ¥2.46 trillion at market value, from January 2027, even as ministers urged the pension fund to own more Japanese assets. The two decisions have different purposes, but together they show that "more domestic investment" is not a single coordinated policy. They also leave GPIF to justify on its own terms whatever it decides.
If the decision belongs to GPIF's own process, each party has a different task before that process reports. For GPIF, the most useful step would be to publish the reasoning of the verification project team when the August summary appears, including the return assumptions it tested, so that any decision, whether to keep the mix or to change it, is seen to rest on the 1.9 percent mandate. For the government, Ueno's formulation, that the fund invests solely for the insured and reviews its mix annually, gives ministers a way to argue for domestic growth through policy rather than through the portfolio, including the alternative-asset channel that remains well below its cap. For investors in yen and JGBs, deviation-band use is the signal to watch. A domestic bond share that stays well above 25 percent outside March and September would say more than any speech.
Those signals will arrive on a known calendar, and the next dates should show whether the August meeting marked a turn or a formality. GPIF's third-quarter results are due in early February 2027 if last year's timetable holds, and its verification of the portfolio for fiscal 2026 would, on last year's pattern, reach the management committee in the spring. The current plan runs to the end of March 2030. Before then, the mix can change only if GPIF's own committee finds that conditions have moved far enough from the plan's assumptions and the health minister approves a revised plan; Ueno has said repeatedly that conditions have not diverged that far. The ¥5.25 trillion loss showed what the currency exposure costs in a bad quarter. Whether Japan's pension giant should carry less of it is a fair question, and markets will trust the answer only if it comes from the fund's own analysis of what its pensioners need.
Apple News, Google News, Feedly, Flipboard, and WhatsApp Channel
COVER IMAGE: The cover image is AI-generated and is used for illustrative purposes only. It does not depict actual people, places, or events.
