Japan and Korea
Two societies that arrived at the future early - ageing, sub-replacement fertility, the zero interest rate, sovereign debt at levels thought impossible - and ran the experiments the rest of the world is now facing.
(the highest of any large country; the world average is 10%)
(sources differ on the measure; it is falling from a pandemic peak above 260%, and it has not produced the crisis the number implies)
(up from 0.72 in 2023, the lowest ever recorded anywhere in peacetime)
A note on framing. Japan and Korea appear on almost every other page of this site as an example inside someone else's story - the ageing case, the deflation case, the industrial-policy case. This page treats them as the subject rather than the illustration, because they share something no other pair of countries does: they hit a series of conditions decades before anyone else, and there are now results. Where the results contradict the familiar narrative, that is said plainly.
The country that arrived first
Almost every structural pressure that rich countries now discuss as a coming problem, Japan met between fifteen and thirty-five years ago. Its working-age population peaked in 1995 and has fallen every year since. Its asset bubble burst in 1990. It hit zero interest rates in 1999, before any other economy in the modern era. It began quantitative easing in 2001, seven years before the Federal Reserve tried it. Its public debt crossed 100% of GDP in the late 1990s and roughly 200% by 2010, at which point most of the world's economists expected something to break.
This makes Japan the closest thing the social sciences have to a natural experiment on the demographic and monetary questions everyone else is now asking. It is also, for the same reason, the country most often invoked and least often examined. "Japanification" has become shorthand for stagnation and decline in economic commentary, usually by people describing a country they have not looked at recently.
South Korea is the second instalment, running about twenty-five years behind Japan on demographics and considerably ahead of it on speed. Korea went from poorer than Ghana in 1960 to a member of the OECD within a working lifetime - the fastest sustained development in recorded history - and is now ageing faster than Japan did, from a lower birth rate, with a less developed migration policy. Whatever Japan's experience teaches, Korea is about to test it under harsher conditions.
What the "lost decades" were
The standard account is that Japan's bubble burst in 1990 and the country then stagnated for thirty years. The headline number supports it: Japanese GDP grew about 0.8% a year between 1991 and 2019, against roughly 2.5% in the United States. On that measure Japan fell steadily behind.
The measure is the problem. GDP counts total output, and Japan's working-age population has been shrinking for three decades while America's grew. Dividing by the number of people who could actually work gives a very different picture. Jesús Fernández-Villaverde and colleagues did exactly that: on GDP per working-age adult, Japan grew 1.44% a year from 1991 against the United States' 1.56% - and from 1998 to 2019, Japan grew slightly faster than the US, 31.9% cumulative against 29.5%. In the decade after the 2008 financial crisis, Japan had the highest growth per working-age adult in the G7.
This is not a technicality and it is not an apology for Japanese policy. It is a claim about what was actually lost. What Japan lost was not productivity growth. It was people of working age. A country whose labour force shrinks by roughly half a percent a year will post weak GDP figures no matter how well its economy works, and reading those figures as evidence of failure mistakes a demographic fact for an economic one.
The rest of the record is similarly mixed rather than dire. Japanese unemployment never exceeded 5.5% through the entire period. Life expectancy rose to the highest in the world. Violent crime fell to among the lowest anywhere. Public infrastructure was maintained rather than allowed to decay. What did go wrong is real and specific: two decades of mild deflation, corporate balance sheets repairing instead of investing, a labour market that split into protected permanent staff and a growing tier of insecure non-regular workers, and a generation that entered the job market during the worst of it and never fully caught up. That last one - the "employment ice age" cohort of the late 1990s - is the clearest human cost, and it is a cost of the policy response rather than of demography.
Debt that was supposed to break
Japan's gross government debt is somewhere between 205% and 230% of GDP depending on whose measure you take, roughly double the US level and higher than any other developed country has sustained in peacetime. It peaked above 260% during the pandemic and has been falling since - not because the debt shrank, but because the end of thirty years of deflation finally gave the denominator something to do. For thirty years, forecasts of a Japanese fiscal crisis have been made confidently and have not arrived. Traders who bet against Japanese government bonds on the debt figure lost money so reliably that the position acquired a nickname: the widow-maker.
Why it has held is genuinely contested, and the honest reading is that several things are true at once. Most Japanese debt is owed to Japanese savers and institutions rather than foreigners, so there is no run risk from external holders. It is denominated in yen, which Japan issues, so involuntary default is not a mechanism available to it. The Bank of Japan itself now holds well over half of outstanding government bonds, which makes a large part of the debt an obligation of one arm of the state to another. And Japan ran a persistent current-account surplus and high domestic savings throughout, which supplied the demand.
What this does and does not license is worth stating carefully, because the Japanese case is quoted in both directions. It does not show that debt levels are irrelevant - Japan's interest burden rises with every increase in rates, and the fiscal room it has is being spent on an ageing population rather than on anything else. It does show that the specific mechanism people feared, a bond-market revolt forcing default, requires conditions Japan did not have. Countries that borrow in someone else's currency, from foreign creditors, without a domestic savings pool behind them, are in a different situation entirely, and the lesson does not transfer to them.
Leaving negative rates, seventeen years on
For most of the last thirty years, Japan's monetary problem was the opposite of everyone else's. Prices fell or stayed flat, the central bank could not push rates below zero far enough to matter, and every tool tried - zero rates, quantitative easing, yield curve control, negative rates from 2016 - failed to produce sustained inflation. Economists studied Japan as the canonical case of a liquidity trap, and the working assumption was that it might never leave.
It left. Inflation returned after the pandemic and the 2022 energy shock, wage rounds began delivering real increases, and in March 2024 the Bank of Japan ended the world's last negative interest rate policy and abandoned yield curve control. Rates have risen in steps since, reaching 1% in June 2026 - the highest since 1995. Japan is now running an ordinary monetary policy for the first time in a generation.
The unresolved question is what did it. If the exit was caused by imported energy and food prices plus a weak yen, it is a shock working its way through and Japan may drift back. If it was caused by the labour market finally tightening as the workforce shrank - too few workers chasing the same jobs - then it is structural, and it is the first evidence that severe ageing is eventually inflationary rather than deflationary. That distinction matters well beyond Japan, because every rich country is heading into the same demographic condition and has been assuming it means the deflationary version.
The most expensive fertility experiment ever run
South Korea's total fertility rate fell to 0.72 in 2023. That is the lowest figure ever recorded for any country in peacetime, and barely a third of the 2.1 needed to hold a population steady. At that rate each generation is roughly a third the size of its parents'. The government's response has been the largest pro-natal effort any country has attempted. By its own accounting it has spent more than 280 trillion won since 2006 - somewhere around $200 to $300 billion, depending on what is counted - on cash payments, housing support, childcare, fertility treatment, and parental leave.
For eighteen years the rate fell anyway, which made Korea the standard evidence that money does not buy babies. Then it turned: births rose in 2024 and again in 2025, and the fertility rate reached about 0.80 - the first sustained increase in nine years, and the largest annual jump in births since 2007.
It is tempting to read that as the spending finally working, and most demographers do not. The more mundane explanation is that an unusually large cohort of women born in the early 1990s reached their early thirties at the same time, and that marriages - which in Korea still precede almost all births - jumped 15% in 2024 after being suppressed by the pandemic. Both of those are timing effects that will pass. The honest position is that Korea has demonstrated something narrower and more useful than either side claims: transfers of this size do not reverse a fertility collapse whose causes are housing cost, working hours, education spending, and the career penalty attached to motherhood. What might work is changing those, which is far harder and far less popular than writing cheques.
What one company inside one country looks like
Korea's industrial structure is unusual among rich democracies in how concentrated it is. The chaebol - large family-controlled conglomerates built through the state-directed industrialisation of the 1960s to 1980s - still dominate the economy. Estimates of their weight vary with the measure: Samsung's affiliated companies alone are put anywhere between 13% and 23% of GDP depending on whether you count value added or group revenue, and the five largest groups together are commonly estimated at something over 40% of GDP by revenue. Any of those figures describes a degree of concentration no other OECD economy approaches.
The model worked, in the narrow sense that it produced the fastest industrialisation in history. Directed credit, protected domestic markets, and explicit export targets moved capital into heavy industry, then electronics, then semiconductors, faster than a market process would have. This is the strongest single case for industrial policy in the historical record, and the countries now attempting it study Korea rather than anywhere else.
The costs are equally documented. Concentration this severe suppresses the formation of independent mid-sized firms, which is where most employment growth comes from in other rich economies. It produces a labour market split between a small number of highly paid conglomerate jobs and a large low-wage remainder, which is a substantial part of why Korean young adults compete so ferociously for so few positions - and, in turn, part of the fertility story above. And it has repeatedly generated the governance failures that concentrated family control tends to generate, including chairmen convicted of bribery and embezzlement and subsequently pardoned on the argument that the economy needed them. The Korean case is genuinely evidence for industrial policy. It is also evidence about what industrial policy costs.
Culture as industrial policy
One Korean export deserves separate mention because it is routinely misread as a spontaneous cultural phenomenon. The global spread of Korean music, television, and film - hallyu - was in significant part a deliberate state project. After the 1997 Asian financial crisis, the Korean government identified cultural export as a growth industry, created a ministry-level budget for it, funded production and overseas distribution, and built the institutional support that content industries in most countries do not receive.
It is worth being careful about how much credit the policy deserves. The state did not write the songs, and plenty of governments have funded cultural export with nothing to show for it. What the Korean case demonstrates is narrower: that a small country can convert cultural production into both revenue and diplomatic standing when the underlying work is competitive, and that soft power of this kind is at least partly buildable rather than purely emergent. For a country of 52 million sitting between China and Japan, with a hostile state on its border, that has turned out to be a strategically useful asset acquired remarkably cheaply.
Two answers to the same question
Both countries face the identical arithmetic: a shrinking workforce supporting a growing retired population. They have answered it differently, and the divergence is the most useful comparison on this page because it isolates policy from demography.
Japan chose adaptation over replacement. It raised the effective retirement age, brought older workers and women into the labour force at rates that now exceed most of Europe, automated aggressively in manufacturing and increasingly in services, and accepted a smaller economy rather than a changed society. Migration was kept low for decades - though this is shifting, and the foreign-resident share has roughly doubled since 2012 through technical-trainee and specified-skills routes that the government is careful not to describe as immigration.
Korea has been slower to adapt and is running out of time faster. Its retirement age is low, its older workers face steep income cliffs after leaving career employment, and old-age poverty is the highest in the OECD at around 40% - a fact that sits oddly beside the country's wealth and that connects directly to its elevated elderly suicide rate. Migration has begun to open, but from a much smaller base and against stronger resistance.
The lesson that transfers is not "do what Japan did." It is that the adjustment is unavoidable and the choice is only about which margin absorbs it: fewer workers each producing more, more workers brought in from outside, later retirement, lower benefits, or a smaller economy. Every country facing this arithmetic will pick some combination of those five. Japan picked early and deliberately. Korea is picking late and under pressure. Most of Europe and China have not yet picked at all.
How the two compare
The two countries are treated together on this page because they hit the same conditions, but they are not the same case. Setting the dimensions side by side shows where the shared arithmetic ends and the policy divergence begins.
The paths from here
Both countries are past the point where the demographic curve can be changed within a working lifetime. The open questions are about how the adjustment is distributed, and the answers matter well beyond these two countries because everyone else arrives at the same arithmetic later.
Managed decline that works
Output per person keeps rising even as the population falls, through automation, later working lives and higher participation. Living standards hold or improve while the headline economy shrinks. National weight in the world declines gradually and without crisis.
Will it happen? This is close to what Japan has already demonstrated for three decades, once the figures are measured per working-age adult rather than in aggregate. It is unglamorous, it is not what the "lost decades" framing describes, and it is the most likely path for Japan specifically.
The fiscal arithmetic finally binds
Rising interest costs on Japanese debt, or the arrival of Korea's ageing bill without the fiscal room it once had, forces sharp adjustment: higher taxes, lower benefits, or both, compressed into a shorter period than a gradual reform would have needed.
Will it happen? Predicted confidently and repeatedly for Japan since the 1990s, and wrong every time so far, which is a reason for caution in both directions. The exit from negative rates has made the interest-cost question live again for the first time in a generation. Korea's version arrives later and from a stronger starting position.
Immigration reframes the problem
Both countries move further toward accepting sustained inflows, as Japan has quietly begun to do for care work and construction. The labour arithmetic eases; the social and political questions that both societies have historically avoided become live.
Will it happen? Slowly and without being called immigration policy, which is how Japan has done it so far. Neither country has the political consensus for the scale that would actually offset the shortfall, and the sending countries are themselves ageing.
Korea's fertility recovers meaningfully
The rise since the 2023 trough continues and settles at a level that changes the projections rather than the headlines. The spending, the housing measures and the working-hours reforms turn out to have worked with a long lag.
Will it happen? The recent rise is real but small, and part of it is a cohort effect rather than a change in behaviour. Nothing in the international record suggests a country returns from 0.7 to replacement. A move to 1.0 or 1.1 would still be one of the larger fertility reversals on record and would not close the gap.
Concentration turns from strength to liability
Korea's dependence on a small number of conglomerates, and specifically on semiconductors, exposes it to a sector downturn or a technological shift it does not lead. The structure that produced the fastest development episode on record propagates a shock just as efficiently.
Will it happen? It is the risk Korean policymakers name most often themselves. Chinese memory-chip capacity is the specific pressure, and diversification has been an official priority for two decades without materially reducing the concentration.
The base case for Japan is continued managed decline with living standards intact and global weight falling. The base case for Korea is a harder version of the same adjustment, arriving faster, with less time to prepare and a more concentrated economy absorbing it. Neither is the catastrophe the standard telling implies, and neither is a solved problem.
Where serious analysts disagree
Was Japan's response to the bubble a policy failure? One camp says yes: rates cut late, banks allowed to carry bad loans for years, fiscal stimulus applied and withdrawn repeatedly. On that reading a faster response would have avoided the deflationary decade. The other camp says the deleveraging was going to take fifteen years regardless. Richard Koo's "balance sheet recession" describes a corporate sector paying down debt against a collapsed asset bubble, and monetary policy cannot shorten that. On that reading Japan's outcome, given the size of its bubble, was respectable. The 2008 experience gave the second camp more support than it had before, since the countries that responded faster still took years to recover.
Is severe ageing inflationary or deflationary? The long-standing view, built substantially on Japan's own record, is deflationary: older populations spend less, demand falls, prices soften. Charles Goodhart and Manoj Pradhan argued the opposite - that the deflationary era was caused by a one-off global labour-supply shock as China and the former Soviet bloc joined the world economy, that this is now reversing, and that shrinking workforces will produce persistent inflationary pressure and higher real rates. Japan's post-2022 experience is the first real test, and it is currently ambiguous enough to be cited by both sides.
Can a country decline gracefully? Japan is the live case for whether a society can shrink in population and relative economic weight while maintaining living standards, order, and institutional quality. Some observers read the last thirty years as a demonstration that it can - the country is safe, functional, healthy, and its people are not visibly worse off. Others argue the bill has been deferred rather than paid: the debt, the deferred pension adjustments, the rural depopulation, and the ice-age generation entering retirement without full careers behind them all come due in the 2030s. There is no way to settle this from the current data, and it is the single most consequential open question about the two countries.
What this means for reading everything else
The practical use of this page is as a correction applied elsewhere. When a commentator warns of "Japanification," check which thing they mean. Falling output per worker did not happen. A shrinking workforce did, and no policy prevents that once the births have not occurred. When someone cites Japan's debt to argue that debt levels do not matter, check whether the country in question borrows in its own currency from its own savers. And when someone cites Korea to argue that pro-natal spending is futile, note that Korea has now had two years of rising births - most likely from a demographic echo passing through, rather than from either the spending or its failure.
And when reading about the ageing of Europe, China, or North America, remember that two countries have already been where those are going, that they took different routes, and that the results are available rather than hypothetical. That is a rare thing in questions of this scale, and it is mostly not used.
The mechanics behind this
Several mechanisms from the Fundamentals pages are doing the work on this page. The productivity ceiling explains why output per working-age adult is the number that matters and total GDP is the number that misleads. Money covers why a government borrowing in a currency it issues faces a different constraint than one that does not. Firms explains why concentration at chaebol scale suppresses the mid-sized firm formation that employment growth usually comes from. And Norms covers why Korea's fertility problem has proved so resistant to money: the binding constraints are expectations about working hours, education spending, and who carries the career cost of a child, none of which a transfer payment changes.