Nippon Life booked ¥500 billion in domestic bond losses in the first half of fiscal 2025. The same firm posted a record ¥1.01 trillion in basic profit for the year. Both figures come off the same balance sheet, and the market has yet to decide which to believe.
Earlier this year this blog argued that Japan’s listed life insurers look structurally cheap on embedded value. Daiichi Life Group (8750), renamed from Dai-ichi Life Holdings on 1 April 2026, and T&D Holdings (8795) trade at 0.6 to 0.7 times P/EV, while European peers change hands at 0.8 to 1.0. One paragraph of that piece brushed past the growing unrealised losses on domestic bond holdings. Accounting noise, it said. Economic substance elsewhere.
That line now deserves scrutiny. Japanese long-dated yields are back above 3% for the first time since the late 1990s. For thirty years life insurers have lived with negative spread – guaranteed rates on old policies exceeding what new money could earn. The era is ending.
The headline numbers
Nippon Life’s domestic bond losses stood at ¥5.45 trillion as of end-December 2025, up ¥763 billion from September. Including Dai-ichi Life, Meiji Yasuda and Sumitomo Life, the four-firm total reached ¥13.25 trillion, an increase of roughly ¥2 trillion in a single quarter. The four-firm total sat at ¥8.5 trillion at end-March 2025. Meiji Yasuda’s losses expanded 8.6-fold year-on-year over that earlier reporting period.
Japan’s Financial Services Agency has brought forward its regular inquiry into how insurers are managing their bond books.
The 30-year JGB currently trades near a 3.6% yield. On 20 January 2026 it briefly touched a record 3.88% after rising 0.27 percentage points in a single session. A super-long bought at 0.5% with 20 years to maturity now fetches around 55 to 60 per cent of par in the secondary market. The longer the remaining duration, the deeper the paper loss.
Read the headlines alone and this looks like a crisis.
Fukoku’s pivot
Fukoku Mutual Life is the first of the large insurers to publish its FY2026 investment plan, as industry practice has it. The direction signals where the sector is heading.
A year ago the tone was bullish. Morimi Junya, then head of financial planning, said yields matched investment levels and the firm would keep adding to super-long bonds. By February 2026 he had softened the stance, saying further rate rises during the year meant no urgency to add JGB exposure. Then on 16 April, Onodera Yusuke – newly promoted to executive officer for financial planning – confirmed Fukoku would add just ¥110 billion of JGBs in FY2026 against the ¥480 billion it had pencilled in for FY2025. That is a 77 per cent cut. Mr Onodera expects super-long yields to stay roughly flat next year.
This matters beyond Fukoku. The dominant buyers of super-long JGBs have been stepping back. The Ministry of Finance’s 2026 JGB issuance plan cut issuance across the super-long tenor, a quiet admission that the demand curve has shifted.
The question now is how insurers manage what they already hold.
What J-ICS actually changes
In July 2025 Japan’s FSA published the regulations for its economic-value-based solvency regime, known as J-ICS. It aligns with the international Insurance Capital Standard. Measurement began in FY2025 and the first reporting date is 31 March 2026.
Under the old regime, bonds classified as held-to-maturity were insulated from market-value shocks. A bond bought at ¥100 and worth ¥55 in the market stayed on the books at ¥100. Management could sleep.
J-ICS marks both assets and liabilities to market. Unrealised losses feed straight into the solvency ratio.
The point most easily lost: liabilities revalue too. If the discount rate rises from 0.5% to 3.6%, the present value of claims payable thirty years from now shrinks considerably. A ¥45 drop on the asset side can be absorbed, or exceeded, by the fall in liability value. Qualifying capital can rise.
But there is a second effect. J-ICS loads a mass-surrender risk charge into required capital when rates rise: as deposit rates climb, more policyholders walk from low-guaranteed-rate savings products into bank accounts. The denominator of the solvency ratio grows. Daiichi explicitly notes in its Q1 FY2025 results that higher rates push required capital up through this channel.
What J-ICS really changes is volatility, not the underlying level. Daiichi’s group ESR ranges between 170 and 200 per cent under the new basis, but a 20bp move in the 30-year can swing the reading noticeably. That volatility is what now governs insurers’ investment behaviour.
The accounting has changed. The economics have not.
Three options
For a life insurer holding a portfolio of low-coupon bonds at deep paper losses, three paths exist.
Hold. Paper losses shrink as maturity approaches; the bonds redeem at par. The cost is reduced liquidity and a solvency ratio that bounces around with J-ICS.
Sell and reinvest. A 20-year bond at ¥55 against par crystallises a ¥45 loss. Put the ¥55 into a new issue at 3.5% and annual coupon income rises from ¥0.50 to ¥1.93. On interest differential alone, recovering the ¥45 loss takes over three decades. Not economic in isolation.
Nippon Life has chosen this path selectively. In its October 2025 investor briefing, Tsuzuki Akira, executive officer for financial planning, disclosed that the firm had realised ¥500 billion in domestic bond losses in H1 FY2025 through ¥1.5 trillion of portfolio turnover, and that full-year rotation would reach ¥3 trillion. Mr Tsuzuki was explicit about the motivation: avoiding forced impairment. If paper losses exceed 30 per cent of book value with no prospect of recovery, accounting rules require an impairment charge. Much of what looks like strategy is defensive.
Rotate gradually. Most of the sector is doing this. Hold the worst bonds; let medium-duration paper mature and reinvest at prevailing yields; slow or shorten new super-long buying. Taiyo Life, part of T&D Holdings, has said publicly that it will replace low-coupon paper with higher-yielding issues.
The equity cushion
Nippon Life booked ¥500 billion of bond losses in H1 FY2025. The same firm posted a basic profit of ¥1.01 trillion for FY2024 – the first Japanese life insurer ever to cross that mark. The two sit on the same balance sheet. The reconciliation runs through another line item entirely: unrealised equity gains.
Nippon Life’s domestic equity unrealised gain alone stood at ¥10.56 trillion at end-December 2025. Policy shareholdings acquired at book value decades ago generate large realised gains when sold at market. Selling equities to offset bond losses lifts portfolio yield without denting basic profit. Low-coupon paper leaves the book; higher-coupon paper replaces it.
There is no accounting trick here. Since 2023 the Tokyo Stock Exchange has pressed companies below 1x book value to improve capital efficiency and cut cross-shareholdings. Insurers were not the primary target. But the reform has handed them the fuel they need to absorb bond losses without wrecking the income statement. Governance pressure meant to serve equity-holders turns out to have been an incidental gift to insurance balance-sheet managers.
Daiichi and T&D
Kikuta Tetsuya, Daiichi’s chief executive, runs the third strategy while still adding to super-longs at the margin. The operating insurer’s legacy book carries contracts from earlier decades, but the average effective liability cost sits well below current super-long yields. With 30-year JGBs at 3.6%, the spread on new investment is comfortably positive. In a May 2025 interview Mr Kikuta called the recent yield spike out of step with fundamentals and said volatility should ease by year-end. The implication: current yields are a buying level.
Daiichi’s domestic bond losses stood at roughly ¥2 trillion at end-March 2025, per Mr Kikuta’s May 2025 remarks; the figure has grown since, in line with the sector-wide increase. Daiichi’s IR materials show that a 50bp fall in yen rates would drop ESR by 19 percentage points, while a 50bp rise would add 4 points. The sensitivity is asymmetric. What keeps the solvency reading inside the 170-200% range despite this is the equity cushion described above: when rates fall, bonds reprice up; when rates rise, the equity offset softens the hit on ESR through realised gains.
A ¥100 billion buyback, repeated annually for five consecutive years, and a payout ratio raised in stages from 30 to 45 per cent over two years say the board is betting on EV compounding rather than waiting out the bond losses.
T&D follows a similar playbook. Its FY2024 buyback matched Daiichi’s ¥100 billion and in May 2025 it re-set its payout target to roughly 60 per cent of group adjusted profit on a five-year rolling average. Cumulative buybacks over FY2021-2024 came to ¥250 billion. A new long-term vision published on 3 April 2026 reserves ¥500 billion for strategic investments; the market read this as a tilt away from buybacks and the stock fell 7.9% on the announcement. The operating companies – Taiyo Life in the consumer market, Daido Life in the corporate – are both replacing low-coupon paper with higher-yielding issues as older holdings mature.
Both firms share the same instinct: sit on loss-making paper rather than fire-sell it, ride out the J-ICS volatility and absorb any drawdown through equity realisations. As mid-sized mutuals exit the super-long market, listed insurers with spare capital face thinner competition for attractive issues.
The J-curve arithmetic
Life-insurer earnings follow a J-curve. Near-term losses give way, as low-yield paper rolls off, to an acceleration in interest spread. The pattern will be familiar to anyone who has watched a private-equity fund mature.
Every year a roughly fixed share of the portfolio matures.
A 30-year bond bought in 2000 at 0.5% redeems at par in 2030. Loss: zero. Put the ¥100 back to work at 3.5% and annual coupon income jumps sevenfold, from ¥0.50 to ¥3.50.
A 20-year bond bought in 2010 at 1.0% matures in the same year. Same arithmetic.
This repeats each year as the low-yield stack built between 2000 and 2020 rolls off. Interest spread – the gap between portfolio yield and guaranteed rate, times reserves – widens mechanically.
The early evidence is already there. Nippon Life’s FY2024 interest spread rose 94 per cent year-on-year to ¥551 billion, a record high. Kampo Life booked ¥142.5 billion in FY2024, with a portfolio yield of 1.91% against an average guaranteed rate of 1.61%, a 30 basis-point gap. Widen that gap to 200 basis points as roll-off completes and Kampo’s spread income compounds several-fold over ten years.
None of this requires the Bank of Japan to tighten further. Current yields merely need to hold. If yields fall, bond prices recover and solvency ratios improve; spread growth slows but still delivers several times the absolute level of the 0.5% era.
Where the risks sit
Three risks cut against the sanguine case.
Speed of the rate move. Level is not the problem; velocity is. Daiichi’s asymmetric sensitivity gives the first tell: a 50bp fall costs 19 points of ESR, a 50bp rise adds only 4. On price effect alone, sharper rates should help. But J-ICS’s mass-surrender capital charge kicks in on the way up, expanding the denominator. Days like 20 January 2026, a 27bp move in a single session, activate both mechanisms at once. A slow grind higher gets absorbed; an acceleration forces selling.
Policyholder runs. If the BOJ policy rate reaches 1.5 to 2.0 per cent and bank term deposits start paying 1.5 per cent, the calculus for savers changes. Holders of 1.0% guaranteed-rate savings products will notice. Mass surrenders turn paper losses into real ones as insurers sell bonds at market to meet redemptions. J-ICS requires this to be modelled in full. The policy rate sits at 0.75% today, which leaves distance before the trigger. Japan offers no modern precedent: the 1997-2001 surrender wave reflected insurer failure, not a shift in relative rates, and the 1980s bubble predated current contract structures.
Fiscal confidence. Japan’s gross JGB debt is forecast to reach ¥1,145 trillion at end-FY2026, with debt-to-GDP around 187 per cent, the highest among large economies. The FY2026 budget assumes an interest rate of 3.0% and earmarks ¥31.3 trillion for debt service. If refinancing averages above 3.0 to 3.5 per cent, the debt-service line starts crowding out everything else. That 40-year touch of 4.215% in January was a market warning. If Prime Minister Takaichi’s government leans harder into fiscal stimulus, a repeat is plausible. Persistent fiscal premium keeps yields elevated – good for reinvestment – but also keeps the tail risk of forced selling alive. The best outcome for insurers is high yields that stay calm.
All three are tail risks on present evidence. Domestic inflation sits between 2 and 3 per cent. The policy rate at 0.75% is nowhere near the surrender trigger. The government has raised its budget-assumed interest rate from 2.0% to 3.0%, which one could read as prudence. A 10-year JGB yield of 5% would require inflation well above 3%, real growth above 3% or a term premium that detaches from fundamentals. Japan’s demographics make the first two unlikely.
Why the market keeps it cheap
The argument so far says the market is wrong. A steady 0.6x P/EV deserves a harder look, though. The section above named three cyclical risks tied to rates and capital. Five structural concerns cut across them.
The biggest drag is structural demand. Japan’s population is shrinking; new policy sales remain under pressure. If EV growth falters, the re-rating thesis collapses on its own terms.
Overseas expansion is unproven. Daiichi has acquired Protective Life in America and TAL in Australia, among other overseas subsidiaries. Currency moves and foreign rate environments could inflict losses independent of the Japanese spread improvement story.
J-ICS disclosure risk is real. The first reporting date in March 2026 will give the market its first clean look at economic-value balance sheets. Any ESR reading below the 170-200% range would punish the stock. Transparency cuts both ways.
The back book is not fully clean. Bubble-era policies at 5%-plus are mostly run off, but a layer of early-2000s contracts at 3-4% remains. On those, the spread improvement story is still playing catch-up.
Governance is on the list too. In February 2026 Daiichi Life Group disclosed that 64 employees across three of its insurance subsidiaries had taken competitor product and customer information while seconded to banks and distributors between 2021 and late 2025. Similar patterns at Nippon Life, Meiji Yasuda and Sumitomo Life brought the total instance count above 3,500 across the four firms over the same period. All four ended the secondment practice on 1 April 2026; Daiichi’s chief executive and group chair each returned 30 per cent of one month’s salary. The direct ratings and ESR impact is modest, but a governance-premium discount looks durable until the post-mortem is absorbed.
Each of these is a real risk. None is an existential one. And they do not correlate tightly: demographics does not cause a J-ICS shock, overseas losses do not cause back-book rot, a secondment scandal does not rewrite liability profiles. Pricing five independent tails as one compound event is a category error, not a judgement. The memory of the 1997-2001 life-insurer failures runs deep, and the number ¥13 trillion revives it.
The market is pricing the scar, not the future
Life-insurer stocks sit at 0.6x P/EV because of memory. Between 1997 and 2001 several insurers failed. The ¥13 trillion in unrealised bond losses echoes that moment. It should not. What is happening now is the exit from negative spread, not a return to it. Nippon Life’s basic profit crossing ¥1 trillion for the first time sits alongside growing bond losses on the same balance sheet. The market has yet to reconcile the two.
J-ICS has made the transition visible. What it also provides is transparency comparable to Solvency II in Europe – exactly the regime under which European life insurers trade at 0.8 to 1.0x P/EV. J-ICS is Japan’s Solvency II. It lays the institutional groundwork for the Japanese discount to close.
The market is pricing the transition pain, not the destination.
So long as yields stay between 3 and 4 per cent, each roll-off cycle widens spread income and grows EV. A P/EV re-rating from 0.6x to 0.85x alone implies a 42 per cent upside. Compound in EV growth – T&D’s four-year average ROEV of 7.9%, Daiichi’s adjusted ROE above 10% – at 8 per cent annually over four years and you add another 36 per cent. Compounded, roughly 93 per cent price return over four years, or 18 per cent annualised; add a 4 per cent dividend yield and the total lands in the low 20s annualised. The figure assumes re-rating and EV growth happen independently and strips out forced-selling and overseas-loss tail risks; take it as an order of magnitude.
Daiichi Life Group (8750) and T&D Holdings (8795) are the listed vehicles for this J-curve. The horizon is the four years over which the bulk of low-yield paper rolls off.
This piece is the author’s analysis of public information and not investment advice. Nippon Life, Meiji Yasuda, Sumitomo Life and Fukoku Mutual are unlisted mutuals. Figures are drawn from company disclosures, Bloomberg and Nikkei reporting as of April 2026.
— Gyokuro