Japan's lost decades
A rich country cut rates to zero, printed money on a scale never tried before — and stayed stuck for a generation. What did the case demand from macroeconomics?
The case the news-reader saw
In the late 1980s Japan looked like the future of capitalism. The land under the Imperial Palace was said to be worth more than all of California; Mitsubishi bought Rockefeller Center; “Japan Inc.” was the cliche every American business school taught. Then, on the last trading day of 1989, the Nikkei touched 38,915 — and the line you see above started down.
You need almost no theory to follow what happened next — only four numbers a newspaper reader was already watching: real GDP growth (how much the economy produces), CPI inflation (whether prices rise or fall), asset prices (stocks and land), and the policy rate (the interest rate the central bank sets). Keep those four in view and the chronology tells itself. If you want the formal definitions, they sit in Ch 7 §7.1 (Gross Domestic Product).
What actually happened
The bubble had a trigger. The 1985 Plaza Accord pushed the yen sharply higher; to cushion the export economy, the Bank of Japan cut rates and kept them low, and the cheap money poured into stocks and land. Through 1989 prices climbed to numbers that, in hindsight, no cash-flow could justify. Then the BOJ tightened, and the air went out. The Nikkei peaked in December 1989; land prices peaked in 1990–91 and then fell for sixteen consecutive years.
For the first few years nobody was sure it was structural. Recessions end; markets recover. Through 1992–95 the working assumption was that this was a deep cyclical dip that would pass. It did not pass. A strong yen and the 1995 Kobe earthquake added strain. In April 1997 the government raised the consumption tax from 3% to 5% to repair the budget — and the economy, just as it was finding its feet, slid back into recession, with the Asian financial crisis hitting later that year.
1997 was the year the banking damage became visible. Hokkaido Takushoku Bank failed; Yamaichi Securities, one of the country's “Big Four” brokerages, collapsed, its president bowing in tears on national television. The banks were carrying mountains of bad loans against collateral whose value had evaporated, and rather than write them off they kept dying borrowers alive — the “zombie firms” that tied up credit without producing growth. By the late 1990s the press had a name for it: the lost decade.
The 2000s did not redeem the name; they extended it. Junichiro Koizumi's government pushed structural reforms — the forced resolution of Resona Bank in 2003, postal privatization, labor-market liberalization. Growth flickered but never caught. By the late 2000s the phrase had quietly become the lost two decades. Then the global financial crisis of 2008 knocked Japan down again, and in March 2009 the Nikkei bottomed near 7,054 — roughly a fifth of its 1989 peak, twenty years on.
The big political bet came in December 2012, when Shinzo Abe's LDP returned to power. In April 2013 he announced the “Three Arrows” — aggressive monetary easing, flexible fiscal spending, and structural reform — and the BOJ under Haruhiko Kuroda launched Quantitative and Qualitative Easing, eventually expanding its balance sheet past the size of the entire economy. “Abenomics” lifted the stock market and weakened the yen. But the consumption tax rose again in 2014 (5%→8%, triggering yet another recession) and 2019 (8%→10%), inflation stayed stubbornly near zero through the decade, and then COVID arrived.
Only very recently has the line on the chart looked different. In the 2024 shunto — the annual spring wage round — large firms agreed to the strongest pay rises in three decades; inflation settled near 2%; and in March 2024 the BOJ raised its policy rate for the first time since 2007, finally lifting it out of negative territory. After thirty-four years the Nikkei also retook its 1989 high. Whether the case is finally over is the question the last stage returns to. (To feel how strange this was, set it against the neighbours: through the same years South Korea quintupled its income per head. East Asia was booming. Japan, the region's giant, was not.)
Where this leaves us
The reader who lived through this had the chronology but no framework. Every conventional prescription was tried: rates were cut toward zero, fiscal packages worth tens of trillions of yen were spent, then rates were cut further, then the central bank turned to unconventional monetary expansion on a scale no textbook had imagined. And the case persisted past every prescription. The discipline did not yet have ready language for what it was watching. The next stage steps inside the policy makers' chairs — and meets the first economist who did have the language, arguing it in real time.
By 1999 the most powerful central bank in Asia had cut its policy rate to zero. The standard textbook said: that's the floor. So what do you do next?
The case the BOJ and the government saw
“The Bank of Japan will provide ample funds and encourage the uncollateralized overnight call rate to move as low as possible.”
— Bank of Japan, Monetary Policy Meeting decision, February 12, 1999 (the announcement that introduced the Zero Interest Rate Policy)
“As low as possible” meant zero. Japan became the first major rich-country central bank to drive its policy rate to the floor, and the acronym — ZIRP, the zero interest rate policy — entered the macroeconomic lexicon here. The standard playbook had run out of pages.
Put yourself in the BOJ's chair in 1999. The transmission story you were trained on runs through the interest rate: cut the rate, borrowing gets cheaper, firms invest, demand recovers. That is the standard IS–LM monetary channel, and it had worked for every postwar downturn. The problem was arithmetic. By 1999 the rate was already at zero, and you cannot cut below the floor. The lever you were holding had reached the end of its travel.
That floor has a name: the zero lower bound (ZLB). Once the policy rate hits zero, the central bank's standard reaction function is pinned and can no longer respond to a weakening economy by cutting further.
Write the policy rule with the floor made explicit. The central bank would like to set the rate from its reaction to inflation $\pi_t$ and the output gap $\tilde{y}_t$, but it cannot go below zero:
$$i_t = \max\{\,0,\; r^*_t + \phi_\pi \pi_t + \phi_y \tilde{y}_t\,\}$$When the desired rate inside the braces turns negative — a deep enough slump — the $\max$ binds at zero and monetary policy can no longer deliver the stimulus the rule is calling for.
It is a thermostat pinned at its lowest setting while the building is still freezing. The dial says “colder” on the other side of zero, but there is no other side. You can flip every other switch in the house — and the BOJ did — but the one control you were trained to trust has stopped responding.
The obvious alternative was fiscal — let the government spend where the central bank no longer could. Here the Ministry of Finance had a real objection, not a foolish one. Japan's gross public debt was already climbing fast through the 1990s, and every yen of deficit spending added to a path that looked, on the standard government-budget arithmetic, unsustainable. The orthodox view inside the BOJ and MOF was that monetary accommodation must eventually do the job through the credit channel, and that piling on fiscal debt risked a future crisis worse than the present stagnation. That worry was sincere, and it is why the fiscal response was repeatedly started and then pulled back.
And then there was the economist who, away from the BOJ and the MOF and the big academic centres, was already naming the thing. Richard Koo, chief economist at the Nomura Research Institute, argued through the 1990s that Japan was not in an ordinary recession at all. After the asset crash, he said, Japanese firms were not behaving like the profit-maximizing borrowers the textbook assumed; they were minimizing debt — pouring every yen of cash flow into paying down loans against collateral that had collapsed, repairing their balance sheets rather than expanding. With the borrowers' side of the credit channel gone, cutting the rate was pushing on a string: there was no one to lend to, however cheap the money. In a regime like that, Koo argued, fiscal stimulus is not one option among several — it is the only counter-cyclical lever left, and the policy makers were using it too timidly. The same balance-sheet wreckage showed up in the banks as the zombie-firm problem: lenders rolling over loans to insolvent borrowers rather than realizing the losses.
For the depth on the unconventional tools the BOJ reached for once the rate hit zero — quantitative easing, forward guidance, the whole machinery beyond the rate — see Can central banks control the economy? (Stage 5); for the question of whether fiscal stimulus actually closes the gap at the floor, see Does government spending help the economy? (Stage 3).
“When a debt-financed bubble bursts, asset prices collapse while liabilities remain, leaving millions of private-sector balance sheets underwater. In order to regain their financial health, the affected companies are forced to repair their balance sheets by paying down debt — even at zero interest rates.”
— Richard Koo, The Holy Grail of Macroeconomics: Lessons from Japan's Great Recession, 2008 (expanding Balance Sheet Recession, 2003)
Was Japan in a different kind of recession?
Koo's claim is that the standard model had the firms wrong. When the corporate sector is repairing its balance sheet, it minimizes debt instead of maximizing profit — and cutting rates does nothing, because there are no willing borrowers.
Push fiscal, or hold the line?
“In a balance sheet recession, the government must borrow and spend the excess savings of the private sector, for if it does not, those savings leak out of the economy and the contraction deepens. Monetary policy is largely powerless because there are no borrowers.”
— Richard Koo, The Holy Grail of Macroeconomics, 2008
Koo is voicing the case at full strength. The mechanism he is naming is old — the trap where rate cuts stop working because money sits idle — and its origin is Keynes's 1936 account of liquidity preference, the idea that at low enough rates the demand to hold cash becomes effectively infinite (History of Economic Thought Ch.8 §8.2, Keynes's General Theory). Koo, with Krugman close behind, was retrieving that tradition for a live case — the post-1990 revival wing of the mainstream that History of Economic Thought Ch.17 §17.1 places at the centre of post-2008 macro.
“Japan's economy is showing clear signs that the downward pressure on prices stemming from weak demand has markedly receded. The conditions that warranted the zero interest rate policy are no longer present.”
— Masaru Hayami, Governor of the Bank of Japan, on the August 2000 decision to lift the zero interest rate policy
This is the orthodox case in its own register, and it deserves to be heard at its strongest. Hayami genuinely believed the economy had stabilized enough to begin normalizing — that holding rates at zero indefinitely carried its own costs (zombie firms kept alive by free money, savers punished, market discipline eroded) and that a central bank that never exits the floor loses credibility. The August 2000 hike was the considered judgment of serious people operating with the framework they had. It was also, in hindsight, a mistake: within months Japan slid back toward deflation and the BOJ had to reverse. The exit is the cleanest single illustration of the gap between the apparatus the policy makers were using and the apparatus the case actually demanded.
Where this leaves us
The policy makers were working with the apparatus they had, and the apparatus they had ran out. Koo was the first to name what was happening in language that fit the data — and he was not at the BOJ, the MOF, or any of the major academic centres when he named it. The case was demanding apparatus the standard mainstream had retired a generation earlier. The next stage asks what the discipline then had to rebuild to catch up.
In 1998 a Princeton economist named Paul Krugman opened a Brookings paper with two words: “It's Baaack.” The liquidity trap, retired as a Keynesian curiosity in the rational-expectations 1970s, was back — because of Japan.
The apparatus the case demanded
“If we want to fight that economic war, we had better understand it. And to understand it we must overcome our adult sophistication and revisit the seemingly childish concerns of an earlier era — the era of the liquidity trap.”
— Paul Krugman, “It's Baaack: Japan's Slump and the Return of the Liquidity Trap,” Brookings Papers on Economic Activity, 1998:2
A generation of macroeconomists had treated the liquidity trap as a museum piece — the kind of thing that might happen in a textbook example but not in a real economy with a functioning central bank. Krugman put it back on the mainstream's agenda with a two-word title. Japan was why.
The liquidity trap, revived. Krugman's 1998 move was to take Keynes's 1936 trap seriously again with modern tools. When the natural real rate of interest falls below zero — as it does when everyone wants to save and no one wants to invest — the central bank, stuck at the zero bound, cannot deliver the negative rate the economy needs. What is left is the expectations channel: the bank can still stimulate if it credibly promises to keep money loose and let inflation run in the future. The catch is the word “credibly.” The BOJ's stop-start signaling — the August 2000 exit above all — destroyed exactly the credibility the mechanism needs. The case did not just illustrate the liquidity trap; it showed the apparatus's hardest problem is making promises about the future stick.
Balance-sheet recession, formalized. Koo's diagnosis needed microfoundations, and Gauti Eggertsson and Krugman supplied a clean version in 2012. Split households into the patient and the impatient; let an asset crash force the impatient (the borrowers) to deleverage suddenly. The standard intertemporal-optimization machinery — the Euler equation that links consumption today to consumption tomorrow — breaks for them, because they are jammed against a binding debt constraint rather than smoothly trading off across time. Two results fall out, both visible in Japan: the paradox of deleveraging (everyone's individually-sensible saving aggregates into a demand collapse) and the paradox of toil (in this regime, working or producing more can make the slump deeper, not shallower).
Who, exactly, is constrained. Representative-agent models — one average household standing in for everyone — quietly assume the deleveraging crowd does not exist. Heterogeneous-agent New Keynesian (HANK) models, the line associated with Kaplan, Moll and Violante's 2018 work, put the wealth distribution back in: a large “hand-to-mouth” group with little liquid wealth responds to income and binding borrowing limits, not to the interest rate the textbook fixates on. More broadly, occasionally-binding-constraints modeling lets the macro apparatus treat the zero bound (and other corners) as a regime the economy can fall into and stay in, rather than a transient edge condition that resolves itself by next quarter. Japan was the exhibit that made “stay in” impossible to wave away.
Wiring finance into the macro. The slump also forced finance back into models that had largely abstracted it away. Bernanke, Gertler and Gilchrist's 1999 financial accelerator — the idea that weak balance sheets amplify shocks because collateral and net worth gate credit — and the unconventional-monetary-policy models that followed (intermediary-based frameworks in the Gertler-Karadi line) built the bridge between the financial system and the real economy that became canonical after 2008. Whether 2008 is best read through that financial-frictions lens or a monetary one is its own question, taken up in Was 2008 a financial crisis or a monetary crisis?.
The long-run name for it: secular stagnation. Stretch the picture out and you reach the oldest framing of all — Alvin Hansen's 1939 worry that a mature economy might simply not generate enough investment demand to absorb its savings at a positive interest rate. Eggertsson, Mehrotra and Robbins gave that intuition a formal overlapping-generations model in 2019, with Japan's two-decade residency at the zero bound as a leading empirical target and an aging, high-saving population as the driver. Whether rich-world slow growth is a pathology to be cured or a new normal to be lived with is the reframe taken up in a sibling walkthrough on secular stagnation.
For the depth on two of these pieces, defer to the walkthroughs that own them: the fiscal multiplier at the zero bound is Does government spending help the economy? (Stage 3), and the unconventional-tools machinery is Can central banks control the economy? (Stage 5). The predecessor chronology — the stagflation crisis that built the pre-1990 mainstream this case challenged — is History Ch.16 (Stagflation and the neoliberal turn); the post-2008 macro-finance integration is the spine of Ch.19 (The 2008 crisis and after).
“Japan's slump — and the impotence of conventional monetary policy in the face of it — cannot be understood without recognizing that the economy is in a liquidity trap.”
— Paul Krugman, “It's Baaack,” Brookings Papers on Economic Activity, 1998:2
Did Japan need a Keynesian apparatus the mainstream had retired?
Krugman's claim is that the liquidity trap is not a curiosity but the operative regime for Japan — and that the monetarist confidence “just print more money” mistakes the rate channel for the quantity channel.
A new apparatus, or an old failure of nerve?
“The way to make monetary policy effective is for the central bank to credibly promise to be irresponsible — to commit to higher future inflation, so that the real interest rate can be negative even though the nominal rate cannot.”
— Paul Krugman, paraphrasing the core result of “It's Baaack,” 1998
Krugman is retrieving a tradition the counter-revolution had explicitly retired. Keynes's 1936 liquidity preference (History of Economic Thought Ch.8 §8.2) was the origin; Eggertsson and Krugman's 2012 deleveraging model became the formal extension, placed by History of Economic Thought Ch.17 §17.1 in the post-2008 revival wing. What makes the move dramatic is that it runs directly against the rational-expectations / New Classical mainstream of the 1970s–80s (History of Economic Thought Ch.10 §10.2, Lucas), which had treated this whole apparatus as obsolete.
“The Bank of Japan can end deflation whenever it chooses. Monetary expansion and a policy of buying real and foreign assets are available. The problem is not that the BOJ lacks tools; it is that it has lacked the will to use them.”
— Allan Meltzer, in the monetarist tradition's reading of Japan, early 2000s
This is the pre-1990 mainstream voiced at its strongest, and it was not a fringe view. Meltzer — the great historian of the Federal Reserve — held, with Friedman behind him, that a central bank can always reflate if it is willing to buy enough of the right assets, and that Japan's deflation was therefore a choice, not a constraint. A monetarist reading this should feel it is being represented fairly: the quantity channel is real, and a bank that prints enough should eventually move prices. The reason the case sits with Krugman is not that this argument is silly — it is that Japan ran the experiment to its extreme and the prices still did not move on schedule.
Where this leaves us
Each piece of the apparatus — the liquidity-trap revival, the formalized balance-sheet recession, the deleveraging paradoxes, HANK, the macro-finance bridge, secular stagnation — was built because the case demanded it. The discipline did not converge on these tools in the abstract and then go looking for an example; it converged on them because Japan's experience falsified the simpler tools the previous generation had thought were enough. The case is the why; the apparatus is the what. The last stage asks what that apparatus then explained beyond Japan.
December 16, 2008. The Federal Reserve announces a target range for the federal funds rate of 0 to ¼ percent. The most powerful central bank in the world has joined the BOJ at the zero lower bound. The apparatus built for Japan is suddenly the apparatus the rich world needs.
What the apparatus then explained
December 16, 2008: the Fed joins the BOJ at zero. The ECB follows, going below zero in 2014. The apparatus built for one country's lost decades was suddenly the apparatus the whole rich world needed. Japan was the canary.
The rich world at the zero bound. The Fed sat at the floor from December 2008 to December 2015 — seven years — and the ECB went to zero and below from 2014. This was not bad luck repeated three times; it was the pattern Japan had previewed, and the apparatus explained why it recurred. Estimates of the natural rate of interest (the Holston-Laubach-Williams series is the standard reference) showed it drifting down across the rich world for decades, so that a normal-sized recession could now push the required rate below zero and pin the central bank. The same secular-stagnation logic that named Japan's case named the rest.
The case for active fiscal at the floor. The lesson Japan had been demanding since the 1990s became the operative result everywhere: when the central bank is pinned at zero, the fiscal multiplier rises — estimates in the 1.5–2.0 range — because the spending does not crowd out private borrowing the way it would when rates can move. The US ran the experiment one way with the 2009 stimulus; the euro area ran it the other way with austerity, then a belated fiscal pivot — and the apparatus read both outcomes. The depth on the multiplier is Does government spending help the economy? (Stage 3).
And the limits of monetary policy. The Fed and ECB reached past the rate exactly as the BOJ had — quantitative easing, forward guidance, negative deposit rates — and these worked, but through indirect channels (portfolio rebalancing, expectations management) that were weaker, slower, and more uneven than the old rate channel. The apparatus built from Japan's case had predicted precisely this asymmetry: at the floor, the unconventional tools help at the margin but do not substitute for the lever that has stopped working. The machinery is the subject of Can central banks control the economy? (Stage 5), and whether 2008 is best read as a financial or a monetary crisis is its own walkthrough. The full post-2008 chronicle is History Ch.19 (The 2008 crisis and after).
And, finally, Japan in 2024–25. The most recent data is the first in a generation that looks like an exit. The 2024 shunto wage round delivered pay rises of around 5% — the strongest in three decades — inflation has settled near 2%, and in March 2024 the BOJ ended negative rates and its yield-curve-control regime, raising the policy rate for the first time since 2007. The apparatus held; the policy response was too cautious for too long; the case may, at last, be resolving. The caution in that “may” is deliberate — the turn is recent, and the demographic deceleration that drove the savings glut has not reversed.
Where this leaves us
The apparatus built up from Japan's case generalized. Balance-sheet recessions are real — when a sector is deleveraging from an asset crash it minimizes debt, not maximizes profit, and the textbook causal chain breaks. Monetary policy alone cannot fix them — ZIRP, QE, and QQE past the size of the whole economy produced near-zero inflation for two decades, the strongest empirical test the “rates always work” prior has ever faced, and the prior failed. Fiscal stimulus is the right tool at the floor. And structural reforms matter but act more slowly than the demand-side problem — Koizumi's reforms were the right kind and Abenomics' third arrow ran weakly, but neither could do the work of demand-side stimulus during a deleveraging regime. The post-2008 rich world confirmed all of this at continental scale. Japan-2025 is finally showing escape signs. The verdict on the lost decades is settled even if the lost decades are not quite over: this was the case that previewed the post-2008 macro the rich world has been living with for a generation — and the policy makers had the apparatus to read it earlier than they were willing to use it.
Where this leaves us
We started inside the case, with the line on the Nikkei chart going down and a newspaper reader who had the chronology but no framework. Every conventional prescription was tried and the case persisted past all of them. The policy makers, working with the apparatus they had, watched that apparatus run out — while Richard Koo, away from the official centres, was already naming a balance-sheet recession that the textbook firms could not exist in. Then the discipline had to rebuild: Krugman retrieving the liquidity trap, Eggertsson formalizing the deleveraging dynamics, HANK putting the constrained households back in, the macro-finance bridge, secular stagnation as the long-run name. None of it was theory looking for an example. It was a case that demanded apparatus the previous generation had retired.
The point of the whole sequence is the direction of causation. Japan did not illustrate balance-sheet-recession theory; it forced the theory into being, and then 2008 made it everyone's theory. The case demanded an account of why a rich economy can sit at the zero bound for decades; the apparatus delivered one, and the post-2008 rich world confirmed it. What remains open is narrower than it once was: whether Japan's 2024–25 turn is a durable exit, and whether the demographic forces underneath the lost decades will let the escape hold. The discipline now has the tools to read the case. The policy makers had them earlier than they used them.