Savings & pensions: what can we learn from Japan and Norway?
Both countries built very large state investment portfolios — but did they actually save the money? Norway did. Japan largely borrowed and invested it. The useful lesson is to read assets, debt and pension promises together, rather than judging a country on its gross debt or the size of its fund alone.
1Did they save the money?
Norway's oil fund holds about $2.1 trillion, roughly 385% of its GDP, built from saved petroleum revenue — the state owns far more than it owes. Japan's GPIF holds about $1.9 trillion, roughly 45% of GDP, and sits inside a much larger ~192%-of-GDP stack of state financial assets — the flip side of even larger borrowing, much of it bonds the Bank of Japan bought with printed money. Same headline, opposite balance sheets. The sections below put both in context: where the world's retirement systems bank their capital, how six systems compare on debt and savings, and what Japan's three debt numbers — gross, net, and interest paid — actually say.
2The world pension map
How the world's retirement systems bank long-term capital
Hover a highlighted country to see how its retirement system works and who runs the money. Mostly fundedMixedMostly pay-as-you-go
Sources: OECD Pension Markets in Focus / Pensions at a Glance 2025, national pension supervisors and fund annual reports, ICI, APRA, NBIM, GPIF, CPPIB, NPS, PPF, France Assureurs, DNB, AP funds, Alecta and Swiss federal pension guidance.
3Six systems compared
States save too, or don't: Norway banked an oil windfall, Japan borrowed a portfolio into existence, America wrote itself an IOU, France let its reserve fund run off.
Same ageing problem, six answers. The row that cuts through the noise is net government debt: green means the state owns more than it owes (a net saver), red means it is a net debtor. Only Norway and Denmark are in the black.
Fiscal & pension scorecard
Debt, savings, and who actually banked the money
| 🇺🇸 US | 🇬🇧 UK | 🇫🇷 France | 🇯🇵 Japan | 🇩🇰 Denmark | 🇳🇴 Norway | |
|---|---|---|---|---|---|---|
| Approach | PAYG + private DC | PAYG + workplace | Pure PAYG | Borrow & invest | Fully funded | Oil wealth fund |
| Gov debt — gross % of GDP | 124% | 102% | 116% | 207% | 28% | 45% |
| Gov debt — net − = net saver | 97%~83% consolidated | 94% | 108% | 137%~77% consolidated | −6%net creditor | −384%oil-fund wealth |
| State investment fund size, % of GDP | Soc. Security fund~8% of GDP | None | FRR~0.7% · in run-off | GPIF~45% · $1.9T | NoneATP is worker-owned | GPFG oil fund~385% · $2.1T |
| Whose money is it? how the pile was built | An IOU to itselfpayroll surplus → Treasuries | — | Old earmarked leviesno fresh inflows | BorrowedJGBs (BoJ QE) + pension contributions | Workers' own savingsthey own it, not the state | Saved oil moneypetroleum export revenue |
| Household (private) savings financial assets, % of GDP | ~450% | ~230% | ~220% | ~360% | ~380% | ~150% |
| Funded pension pots % of GDP | ~153%401(k) / IRA | ~76%workplace DB/DC | ~12% | ~30%+ GPIF reserve | ~206%largest in OECD | ~12% |
| PAYG promises, unfunded accrued to date, % of GDP | ~181%Social Security | ~224%state pension | ~397% | ~200% | ~24%tax-financed | ~320% |
Norway vs Japan. Norway saved an oil windfall, so its fund is genuine net wealth — the state owns 384% of GDP more than it owes. Japan borrowed to build its pile: GPIF (~45% of GDP) is a real pension reserve of workers' contributions and will help pay pensions, but the much bigger ~192%-of-GDP asset stack is the flip side of even larger debt — mostly bonds the Bank of Japan bought with printed money — a state-scale carry trade, not saved money. Net it all out and Japan is still a debtor (~77% of GDP), just a smaller one than the US.
Sources — gross & net debt: IMF WEO & Fiscal Monitor (2025); Denmark & Norway net positions from Statistics Denmark and Statistics Norway national accounts (both net creditors; the IMF publishes no net-debt series for Norway). Consolidated net debt (Japan ~77%, US ~83%) from Chien, Du & Lustig, “Japan’s Debt Puzzle” (JEP 2025). State funds: NBIM (GPFG ~$2.1T), GPIF (~$1.9T), SSA (OASI/DI Trust Fund), FRR. Household financial assets: Fed Z.1, BoJ flow of funds, ONS, INSEE, Statistics DK/NO. Funded pension assets: OECD Pension Markets in Focus 2025. Unfunded PAYG promises (accrued-to-date): Eurostat nasa_10_pens1 2021 (France, Denmark), SSA closed-group (US), ONS (UK state pension), MHLW/Oguro (Japan), Statistics Norway (folketrygden). The unfunded-promise and net-position rows use different national bases and are not strictly like-for-like — treat as orders of magnitude.
4One debt, three lenses
Gross, net, and what it costs
General government debt since 1980, three ways. Gross debt is the headline everyone quotes — and on it Japan looks broke. Net debt subtracts what the state owns — and Japan drops by a third (consolidate the BoJ and the public pension funds, as Chien, Du & Lustig do, and it falls to ~77% of GDP, below the US). Interest paid is what actually hits the budget — and there the "world's most indebted country" has been paying less than half what the US pays. The lesson of the Japan debate: never read the gross number alone.
01Gross debt, % of GDP
02Net debt, % of GDP
03Interest paid, % of GDP
Source: Dealroom.co analysis based on the IMF DataMapper — general government gross debt (GGXWDG_NGDP, WEO), net debt (Fiscal Monitor) and interest paid on public debt (ie, Global Debt Database; latest year lags by one, and no EU27 aggregate is published for net debt or interest). Pulled live; the latest year is a WEO estimate. The country pills rebuild all three panels. The dashed green line on the net panel is a different measure — Japan's fully consolidated net debt (central + local government + public pension funds + Bank of Japan + public financial institutions, financial assets marked to market) from Chien, Cole & Lustig, “What about Japan?” (St. Louis Fed WP 2023-028, rev. 2025), Table 1 year-end snapshots (1997 24.7%, 2012 118.4%, 2023 94.3%) with the published Q2-2024 update of ~77–78% (Chien, Du & Lustig, JEP 2025; St. Louis Fed, Apr 2025). Segments are straight lines between the authors' snapshot years; the gap against the IMF net line is the assets the IMF measure leaves out.
“The Japanese government is a giant sovereign wealth fund — funded not with oil money like Norway, but with money borrowed from the Japanese people.”
Why Japan heldThe debt is owed to itself
~88% of JGBs are held domestically; the Bank of Japan alone holds ~43–46%, and the interest the state pays it flows back as seigniorage. Households sit on a record ¥2,386T (~$14.7T) of financial assets, and the country runs record current-account surpluses (¥31.9T in 2025) — a captive, yen-denominated funding pool no euro-area treasury enjoys.
The reframeNet debt is the honest number
Consolidate the government with the BoJ and public pension funds (GPIF) and Japan holds a financial portfolio worth ~192% of GDP — foreign bonds, equities, FX reserves. True net debt: ~77% of GDP, arguably the lowest of the majors. Japan ran a state-scale carry trade — borrow at ~0% in yen, invest abroad at higher returns — and “got very, very lucky” as foreign assets soared and the yen fell.
What's changingThe inflation trigger has been pulled
The BoJ exited negative rates in March 2024 and has hiked to 1.00% (June 2026) — the highest since 1995. The 30-year JGB yield hit 4.0%; debt service takes a record ¥31.3T (~26%) of the FY2026 budget; the IMF projects interest costs doubling by 2031. The model that made 250% carryable — free money and a captive buyer — is being dismantled in real time.
5Can the promises be kept?
Pension promises are just debt that doesn't appear in the debt statistics. Each bloc is having the same argument about whether that implicit debt is a crisis — with a different accent.
United StatesThe 2032 cliff — crisis or invention?
The arithmetic isn't disputed: the 2026 Trustees Report puts OASI depletion in Q4 2032 with 78% of benefits still payable (83% on the combined funds, 2034), and the 75-year gap at 4.42% of taxable payroll (~1.5% of GDP). The fight is what to call it. Krugman: a pay-as-you-go program “won't go bankrupt” — the 25-year gap is 1.06% of GDP, less than one year's proposed military increase; the “crisis” is an invention to justify cuts, as in 2004. CRFB / MacGuineas: “Washington is sleepwalking into a retirement crisis” — the automatic ~22% cut is $18,400/year for a typical retiring couple, and fixing it today already takes a 34% payroll-tax rise or a 25% benefit cut. Kotlikoff: on an infinite horizon the hole is $71.9T — “we're not broke in 20 years, we're broke now.”
JapanAdjusted by formula, tested by inflation
Japan answered the sustainability question by formula: the 2004 reform capped contributions and lets “macroeconomic slide” automatically trim benefit indexation until the system balances over ~100 years, verified by an actuarial check every five years. The 2024 valuation puts the replacement rate on a path from 61.2% today to ~50% by 2057 — solvent on paper, with GPIF’s ¥260T as buffer, but at the price of steadily thinner pensions. The implicit debt (~¥1,110T accrued) never appears in the JGB statistics — and the new test is inflation: the slide was designed for a deflationary world, and real benefit cuts now happen in plain sight.
EuropeThe biggest promises, the hardest politics
Europe promised the most and finds the politics hardest. Accrued unfunded entitlements run from 192% of GDP (Sweden) to 496% (Spain); some states built auto-stabilisers (Sweden’s notional accounts, Germany’s sustainability factor, Finland’s life-expectancy link), but the flagship reform went the other way: France’s 2023 rise of the retirement age from 62 to 64 — carried against mass protests — was suspended in the December 2025 budget deal as the price of political survival. The EU’s answer is the funded pillar: the SIU pensions package (auto-enrolment, dashboards) is an attempt to grow assets against the promises, which is exactly what the funded-vs-promised scatter on the main analysis measures.
The numbers only look incompatible because the horizons differ: Krugman's “1% of GDP” and Kotlikoff's “$72 trillion” are the same shortfall expressed over 25 years vs forever. Europe's 200–500%-of-GDP entitlements in the funded-vs-promised scatter on the main analysis are the same accrued-to-date arithmetic applied honestly to every EU state — and Japan's lesson cuts both ways: implicit debt is carryable if the state holds assets, savings stay home and rates stay low. Fund the promises and the ratio stops mattering; that is the whole argument of this page.
Sources: 2026 OASDI Trustees Report (June 9, 2026); Paul Krugman, “Social Security is Facing a Political Crisis” (Jun 11, 2026), “The Clean Little Secret of Social Security” (Mar 2025) and “Inventing a Crisis” (NYT, 2004); CRFB, Analysis of the 2026 Trustees Report; Laurence Kotlikoff (Substack, Jun 2026); Chien, Du & Lustig, “Japan's Debt Puzzle” (JEP, Fall 2025); Money & Macro, “Why Japan isn't broke yet” (May 2026); IMF WEO & Fiscal Monitor (April 2026); BoJ flow of funds (Mar 2026); Japan MOF; Federal Reserve Z.1; OECD Pension Markets in Focus (prelim. 2025); MHLW 2024 actuarial valuation & Oguro/CIGS; France 24 (Dec 16, 2025); Eurostat nasa_10_pens1 (2021). Figures are each participant's own; horizon differences are flagged in place.
What this means
A large state fund is not the same thing as saved money. Norway banked an oil windfall and is a net creditor by 384% of GDP; Japan borrowed a portfolio into existence and remains a net debtor once the accounts are consolidated — a carry trade that worked, at rising cost now that rates are positive. The reading that matters for Europe: gross debt alone says little, and pension promises are debt that never appears in the statistics. Assets, debt and promises have to be read together — which is what the funded-vs-promised scatter on the main analysis does for every European country.
Savings, pensions and European innovation
Europe has the money. Very little of it reaches venture. What should change, and who can move first.
The Savings and Investments Union
The EU's plan to move household savings out of deposits — what has shipped, and the Q2 2027 test.
M · How it's built
How this is built
- The Japan debt charts pull live from a keyless public API at load (
/api/imf-debt-ratio: IMF DataMapperGGXWDG_NGDP, Fiscal Monitor net debt,ieinterest paid), with an inline snapshot fallback so the page never renders empty. - The consolidated Japan net-debt line is a different measure from the IMF series: Chien, Cole & Lustig, “What about Japan?” (St. Louis Fed WP 2023-028, rev. 2025), Table 1 snapshots, with the published Q2-2024 update of ~77–78% of GDP.
- The world pension map and country write-ups are hand-maintained against OECD Pension Markets in Focus / Pensions at a Glance 2025, national supervisors and fund annual reports — last reviewed July 2026.
- The six-system scorecard quotes each source's own published figures (IMF, OECD, Eurostat, NBIM, GPIF, SSA, Fed Z.1, BoJ, ONS, INSEE, Statistics DK/NO); the unfunded-promise and net-position rows use different national bases and are orders of magnitude, not like-for-like accounting.
- This page was split out of the main Savings, Pensions & Investment analysis so the main page can stay focused on the pension-to-innovation argument.
Who actually funds European tech?
The Dealroom platform tracks the LPs, funds and rounds behind these numbers — including which pension funds are finally moving.