Japan's Asset Price Bubble: Peak, Crash and a 34-Year Recovery

Key Takeaways

• The all-time high was set immediately after a rate rise. The Bank of Japan raised the official discount rate to 4.25 percent on 25 December 1989. The Nikkei 225 closed at 38,040.37 on 22 December, then rose in each of the four remaining sessions of the year to its record 38,915.87 on 29 December.
• The equity fall was fast at the start and then very long. The index was 48.0 percent below its peak by 1 October 1990, nine months in, and then took another eighteen years to reach its worst close of 7,054.98 on 10 March 2009, a total decline of 81.9 percent.
• Recovering that decline required a gain of about 452 percent, which is why the round trip took thirty-four years. The Nikkei first closed back above its 1989 high on 22 February 2024.
• Land peaked after stocks, not with them. The Bank of Japan's study records that the Urban Land Price Index peaked in September 1990, nine months after the equity high, at almost four times its September 1985 level.
• Consumer price inflation gave no warning. Japanese consumer prices rose 0.6 percent in 1986, 0.1 percent in 1987 and 0.7 percent in 1988, the three years in which the boom accelerated hardest.
• Bank capital was a function of the stock market. The combined capital base of city banks, long-term credit banks and trust banks rose from 35 trillion yen at end-September 1988 to 46 trillion a year later, partly through Tier II capital reflecting unrealized gains on the banks' own equity holdings.
• The banking failures arrived almost eight years after the peak, not with it. Sanyo Securities, Hokkaido Takushoku Bank, Yamaichi Securities and Tokuyo City Bank all failed inside November 1997, and the cleanup from April 1992 to March 2000 consumed 86 trillion yen, equal to 17 percent of gross domestic product.

What Happened in Japan's Asset Price Bubble?

Japan's bubble is usually told as a story about a stock index, and the index is the smallest part of it. Equity prices, urban land prices and bank credit rose together for four years, each supporting the other, and then unwound over a period long enough that the unwinding produced several distinct crises rather than one.

The Bank of Japan's Institute for Monetary and Economic Studies published a study of the episode in February 2001, written by Kunio Okina, Masaaki Shirakawa and Shigenori Shiratsuka, which remains the central bank's most direct account of its own conduct. It characterises the period by three simultaneous features: a rapid rise in asset prices, an overheating of economic activity, and an expansion of money supply and credit. Its analytical centre of gravity is what the authors call intensified bullish expectations, a shared belief about future asset values that became dominant among households, firms, financial institutions and government at roughly the same time.

That study declines to fix a single start date, and the refusal is informative. It puts the beginning of the rise in asset prices around 1982, the acceleration in 1985 and 1986, and notes that those two years coincided with the recession caused by the appreciation of the yen, which is why few count them as bubble years. Its own reading is that many treat 1987 as the starting year, since that was when the economy was expanding, money supply and credit growth turned back up, and asset prices rose fastest all at once. Equities led: the Nikkei 225 stood at about 12,598 at the time of the September 1985 Plaza Agreement and reached 38,915.87 at the end of 1989, which the authors put at 3.1 times the earlier level. Land followed with a lag, spreading outward from Tokyo to Osaka and Nagoya and then to smaller cities, and the Urban Land Price Index did not peak until September 1990. Credit expanded alongside both. Money supply on the M2 plus certificates of deposit measure bottomed at 8.3 percent annual growth in the fourth quarter of 1986 and passed 10 percent by the second quarter of 1987, while fund-raising by the domestic corporate and household sectors grew at close to 14 percent year on year by 1989. The Bank of Japan's authors are explicit that credit growth was the more conspicuous of the two.

Chronology of the Japanese bubble and its unwinding

Selected dated events with the Nikkei 225 close on that session. Index levels are daily closes from the published Nikkei 225 series; where a listed date was not a trading day, the nearest prior close is shown and noted.

Two things about that sequence resist the usual telling. The first is how long it is. The gap between the equity peak and the banking failures is nearly eight years, and the gap between the equity peak and the worst equity close is nineteen. Nothing in this library has that shape. The second is that the worst close arrived in March 2009, driven by a crisis originating in American mortgage credit, which is to say that the low of Japan's bubble was set by an event with no connection to it. The comparison worth drawing is with the 2008 financial crisis, whose own trough fell in the same week and took five and a half years rather than thirty-four to recover.

What Did Japan Look Like Before the Nikkei Peaked?

Four conditions defined the setup, and the striking feature is that none of them looked like a warning through the lens most investors were using.

An exchange rate shock and the policy response to it. The Bank of Japan's study describes the first phase of the bubble period as running from the September 1985 Plaza Agreement through the spring of 1987, during which monetary easing was promoted to counter a recession caused by the rapid appreciation of the yen. The scale of that appreciation is visible in the exchange rate series: the yen averaged 236.53 to the dollar in September 1985 and 123.61 by December 1988, roughly doubling in value in a little over three years. The Bank of Japan cut the official discount rate five times between January 1986 and February 1987, from 5.00 percent to 2.50 percent. Its own authors note that only the first of those five cuts was at the pure instigation of the Bank; the rest were shaped by the framework of international policy coordination, under which surplus countries such as Japan and Germany were asked to boost domestic demand.

A record-low policy rate held for an unusually long time. The 2.50 percent discount rate, then the lowest in Japanese history, was in place from 23 February 1987 until 31 May 1989. The Bank of Japan's account of that middle phase is candid: it sought an appropriate moment to tighten and could not easily find one. Those are precisely the years in which the Urban Land Price Index was climbing toward four times its 1985 level.

Deregulation that changed what banks needed to do to earn a living. The Bank of Japan's study dates the gradual deregulation of interest rates on deposits to 1985, which puts it alongside the boom rather than before it, and sets it in a wider sequence: the Foreign Exchange and Foreign Trade Control Law was revised in 1980, restrictions on yen conversion were abolished in 1984, and restrictions on firms raising money in the securities market were removed from around 1980 while banks were allowed only phased entry into the securities business. The study is direct about the consequence: as deposit rates were deregulated, banks gave up the economic rent they had earned from accepting deposits at regulated rates, and pursued aggressive lending to small firms backed by property and property-related loans instead. The study compares seven banks that later failed against their peers in the Second Association of Regional Banks and finds that the failures were already less profitable in the first half of the 1980s and expanded property-related lending most aggressively from the mid-1980s. The vulnerability was visible in the loan book years before it was visible in the price.

A narrative about scarcity that came with an official forecast attached. The most quoted anecdotes about Japanese land are usually unsourced, so here is one that is not. The Bank of Japan's study records that the National Land Agency forecast in 1985 that demand for office space in Tokyo would increase to a level equivalent to the office space contained in 250 skyscrapers, and observes that such a forecast could have had a significant effect on expectations about future land prices. A bullish story carrying a government agency's projection is considerably harder to argue with than one that does not.

Meanwhile the indicator most investors and central banks were actually watching stayed quiet. Japanese consumer prices rose 0.6 percent in 1986, 0.1 percent in 1987 and 0.7 percent in 1988. Inflation only reached 2.3 percent in 1989 and 3.1 percent in 1990, by which time the tightening was underway and the equity peak had passed. For the general version of why a stable price index and a stable financial system are different questions, see financial conditions, credit spreads and liquidity, and for the collateral loop that Japanese land and Japanese bank lending formed with each other, the credit cycle and refinancing risk.

How Far Did Japanese Stocks and Land Actually Fall?

The equity decline is easy to state and easy to misread, because a single peak-to-trough number compresses nineteen years into one figure. The table below breaks it into the lows that mattered at the time.

Nikkei 225 closes measured against the 29 December 1989 peak of 38,915.87. Nikkei 225 daily closes, measured close to close, dividends excluded.

Read the last column rather than the third. Each of those lows was, in its moment, the bottom. An investor who bought the 63.2 percent decline of August 1992 on the reasoning that two thirds was surely enough spent the next six years watching a further loss and the next seventeen waiting to be right. That is the practical content of an 81.9 percent drawdown: recovering it requires a gain of about 452 percent, which is a different order of arithmetic from the roughly 132 percent needed to undo the 56.8 percent fall of 2008. Our drawdown and recovery calculator makes that asymmetry concrete for any decline you want to test.

The pace also confounds the usual crash imagery. Japan did not have a Black Monday. Measured close to close, the worst single session between the 1989 peak and the end of 1992 was 2 April 1990, at 6.60 percent, followed by 19 August 1991 at 5.95 percent and 23 August 1990 at 5.84 percent. Compare that with Black Monday 1987, where a single day did more visible damage than any day of Japan's entire unwinding. The Japanese decline was made of ordinary sessions accumulating for years, which is precisely why it was so hard to recognise as an event while it was happening.

Land, which never came back

Two land measures are worth keeping separate, because they are different indexes with different peaks. The Bank of Japan's study uses the Japan Real Estate Institute's Urban Land Price Index and reports that it peaked in September 1990 at almost four times its September 1985 level, that it fell continuously afterwards, and that by 1999 it stood some 80 percent below the 1990 peak and about 20 percent below where it had been in September 1985. In other words, fourteen years of Japanese land ownership produced a loss even before adjusting for anything.

The Bank for International Settlements residential property series for Japan, a separate index published quarterly, peaked in the first quarter of 1991 and reached its trough in the second quarter of 2009, 46.5 percent lower in nominal terms and 48.9 percent lower after deflating by consumer prices. The number that matters for an investor is not the trough but the latest reading: as of the fourth quarter of 2025, that series was still about 21 percent below its 1991 nominal peak and about 36 percent below in real terms. The Nikkei took thirty-four years to recover. Japanese residential property, on this measure, has not recovered after thirty-four. For anyone treating property as the stable half of an allocation, that is the counterexample worth carrying, and real estate and REITs covers how property exposure behaves inside a portfolio more generally.

Which Warning Signs Were Visible in Advance, and Which Only in Hindsight?

Japan is the episode where hindsight does the most damage, because the price chart is so extreme that it feels self-evidently unsustainable in retrospect. It was not self-evident at the time, and the specific reasons why are more instructive than the general observation that bubbles are hard to call.

Six signals a Tokyo investor could in principle have watched during 1987, 1988 and 1989, and what each was worth before the December 1989 peak.

The single most uncomfortable fact in this episode is the December 1989 sequence. The Bank of Japan raised the discount rate to 4.25 percent on 25 December, the third increase in seven months. The Nikkei had closed at 38,040.37 on 22 December. It then closed higher on 26, 27, 28 and 29 December, ending the year at its all-time high of 38,915.87. An investor watching for the market to acknowledge tightening was, in the last week of 1989, watching the market do the opposite, on the way to the exact top. Any account of Japan that implies the turn was obvious has to explain that week.

The credit signal was the better one, and it was better for a reason that generalises. Asset prices tell you what people are willing to pay. Credit growth tells you what they are borrowing to pay it with, and borrowing is the part that creates forced sellers later.

What was genuinely unavailable was the condition of the loan books. Japanese supervision at the time operated on what the Bank for International Settlements account calls the convoy system, under which regulation was conducted so as not to undermine the viability of the weakest banks, and under which depositors and other stakeholders took it for granted that banks would never be allowed to fail. A regime designed to prevent weakness from becoming visible is, by construction, a regime in which an outside investor cannot measure weakness. That is not a Japanese peculiarity so much as a general property of implicit guarantees.

Hindsight check. Ask of any Japanese bubble warning sign: would acting on it in 1987 or 1988 have been rewarded, and would the person acting still have had a job in 1989? An investor who called the top in 1987 watched the Nikkei almost double before being proved right, and the credit data supporting the call kept getting worse rather than resolving. Two years of underperformance was not a survivable position for anyone managing outside money, which makes Japan a governance problem as much as an analytical one, and the two-year lag is why so few Japanese institutions acted on evidence they could see. Cognitive biases in trading sets out the machinery behind that.

Why Did an Asset Bubble Become a Banking Crisis Eight Years Later?

This is the question that makes Japan worth studying separately from every other bubble. The equity peak was December 1989. The month in which major Japanese financial institutions failed almost weekly was November 1997, seven years and eleven months later. Understanding that gap requires three mechanisms operating together.

Bank capital was itself built out of the bubble. The Bank of Japan's study reports that the combined capital base of city banks, long-term credit banks and trust banks stood at 35 trillion yen at the end of September 1988 and 46 trillion yen at the end of September 1989. It identifies three sources of that increase: higher profits from the bubble-era expansion, an increase in Tier II capital reflecting unrealized capital gains on the banks' own stockholdings, and equity issuance made easy by favourable market conditions. Every one of those channels runs on asset prices. A falling Nikkei therefore did not merely damage bank shareholders. It shrank the regulatory capital that determined how much the banking system could lend, at the same moment that falling land prices were shrinking the collateral behind loans already made. Two constraints tightened from one shock.

The lending had been steered into property by deregulation, not by a mania alone. As deposit rates were liberalised, the reliable spread that Japanese banks had earned on regulated deposits eroded. The Bank of Japan's study describes the response plainly: banks pursued aggressive lending to small firms against property collateral, and property-related lending directly. Its comparison of seven failed regional banks against their peers shows the pattern was not evenly distributed; the institutions that later failed were the ones already earning least and lending most aggressively into property from the mid-1980s. The credit was concentrated exactly where the collateral was about to fall.

The non-bank layer failed first and poisoned the politics of the response. The jusen were housing loan corporations founded by banks and other financial institutions in the 1970s to complement bank housing lending. During the 1980s they shifted toward lending to real estate developers, a business in which the Bank for International Settlements account says they had little expertise. A Ministry of Finance inspection in the summer of 1995 found aggregate losses across the seven jusen companies of 6,410 billion yen, far beyond what their founder banks could absorb. The resolution allocated 3,500 billion yen to founder banks, 1,700 billion to lender banks, 530 billion to agricultural financial institutions, and left a residual 680 billion to be covered by taxpayers.

That last figure is small, and it is the most consequential number in the entire aftermath. Public anger at using taxpayer money to resolve non-depository institutions with no role in ordinary Japanese life was severe enough, in Nakaso's account, that it became almost politically taboo to discuss further use of public funds for the banking problem, and the taboo held until the failures of autumn 1997. Public money committed to the jusen totalled 685 billion yen; public money eventually committed to the banking system reached 60 trillion, roughly ninety times as much. The first intervention was too small to fix anything and large enough to make the second one politically impossible for two years.

And the safety net was not built for this. Japan had experienced no major bank failure in the postwar period. The Deposit Insurance Corporation held an insurance fund of 300 billion yen at the end of March 1987 and, as late as early 1996, had a staff of sixteen and 390 billion yen of funds, operating from a corner of the Bank of Japan's premises. By 1999 the same institution together with the Resolution and Collection Corporation had more than 2,000 staff and 60 trillion yen of public funds. The capacity to resolve a banking crisis was constructed during the crisis, which is a large part of why the crisis lasted.

When the failures came, they came through funding. Sanyo Securities was ordered to suspend business on 3 November 1997 and defaulted on 8.3 billion yen of unsecured call money, a trivial sum against interbank turnover. Nakaso records that the market stayed calm for a few days, and then participants absorbed that this was the first default in the history of the Japanese interbank market. Lenders preferred to place money with the Bank of Japan rather than with each other, foreign banks cut credit lines to Japanese banks generally, and the Bank of Japan's outstanding credit supply through money market operations reached 22 trillion yen in December 1997. Nakaso's summary is the sentence to keep: a small default paralysed the entire interbank market, because when the overall system is fragile, a default by any one institution can become a general disruption.

The failures that followed were not small. Hokkaido Takushoku Bank, a city bank with 9.5 trillion yen of assets, failed in November 1997. Yamaichi Securities, one of the four largest securities houses in Japan with 22 trillion yen of client assets, announced on 24 November 1997 that it had suspended new contracts. The Long-Term Credit Bank of Japan, with 26 trillion yen of assets, saw its problems surface in June 1998 and was nationalised on 23 October 1998, by which point it was, in Nakaso's telling, barely able to fund itself for another day. The mechanism by which a solvency problem becomes a liquidity problem is the one described in how correlations change across regimes: exposures assumed to be independent turned out to share one factor, and the factor was Japanese land.

How Did the Bank of Japan and the Ministry of Finance Respond?

The rate path is the cleanest record of the whole episode, because the Bank of Japan publishes every change with its effective date. Read the tightening half first.

Bank of Japan official discount rate, from the Bank's own record of basic discount rates. Effective dates as published.

Three features of that table matter. The tightening was late and then fast: five increases in fifteen months, from 2.50 percent to 6.00 percent, with the last of them on 30 August 1990, by which point the Nikkei had already fallen 34.0 percent from its peak. The easing was slower than the collapse: the first cut came on 1 July 1991, ten months after the final increase and eighteen months after the equity high, with the index already 38.0 percent below its peak on the day of that cut. And the destination was somewhere no major central bank had been. By 8 September 1995 the official discount rate was 0.50 percent, which is not a stimulus setting but the end of the conventional instrument entirely.

What followed was the invention of the toolkit that every other central bank later borrowed. The Bank of Japan's own reference record of unconventional measures lists a zero interest rate policy running from February 1999 to August 2000, a quantitative easing policy from March 2001 to March 2006 in which the operating target was changed from an interest rate to the outstanding balance of current accounts at the Bank, comprehensive monetary easing from October 2010 to April 2013, and quantitative and qualitative monetary easing from April 2013 to March 2024. Look at the closing date on that last entry. Japan's unconventional policy era ended in March 2024, one month after the Nikkei regained its 1989 level in February 2024. The two clocks finished together, thirty-four years after the peak.

The fiscal and supervisory response ran on a different track and arrived later. The Bank for International Settlements account records that from April 1992 to March 2000, the total spent dealing with the non-performing loan problem was 86 trillion yen, or 17 percent of gross domestic product, covering write-offs and provisioning by banks, Deposit Insurance Corporation transfers to cover losses at failed institutions, and capital injections. Over the same period 110 deposit-taking institutions were dissolved. After the 1998 legislation, available public funds were doubled from 30 trillion to 60 trillion yen, allocated 17 trillion to loss coverage, 18 trillion to institutions under public administration or nationalised, and 25 trillion to capital injections.

The judgement worth forming is not that Japanese policymakers were slow, which is the standard verdict and is not very useful. It is that the response was sized against the problem the authorities believed they had. Nakaso is explicit that in the early stage, the expectation that asset prices and therefore collateral values would eventually recover allowed the authorities to adopt a wait-and-see policy. That expectation was not absurd in 1992. It was simply wrong, and every year it remained the working assumption was a year in which the resolution capacity was not built. For how a policy rate and a market bottom relate to each other in general, policy rates and forward guidance covers the transmission channel.

How Long Did It Take the Nikkei to Get Back to 1989?

The headline answer is 8,379 trading sessions, or a little over thirty-four years, from 29 December 1989 to 22 February 2024. That is the strictest possible measure and the one most people mean, and it needs four qualifications before it means anything useful.

It is price only. The Nikkei 225 is a price index, so the 2024 date ignores thirty-four years of Japanese dividends. Reinvesting them would have pulled the break-even date meaningfully earlier, and across a window as long as Japan's the gap between the price date and the total-return date is wider than in any other episode here. This page names no total-return date for the Nikkei, because no source consulted for it supplied one.

It is nominal, and Japan is the one case where that cuts the other way. Nominal recovery normally overstates the investor's result, because inflation erodes the purchasing power of the recovered sum. Japanese consumer prices fell 0.1 percent in 1995 and the following decades included extended periods of very low and sometimes negative inflation, so the usual correction is smaller here than it would be for a comparable American drawdown.

Recovery of an index is not recovery of a portfolio, and thirty-four years is long enough for that to matter absolutely. An investor who was forty in 1989 and holding Japanese equities was seventy-four when the index got back. Someone contributing steadily throughout bought the entire decline at progressively lower prices and recovered decades sooner. Someone drawing down through the 1990s never recovered at all, which is the mechanism laid out in sequence of returns risk. The index chart is the same for all three and describes none of them.

Other clocks ran on their own schedules, and slower. Japanese unemployment was 2.1 percent in December 1989 and 2.0 percent in March 1990, its cycle low, so the labour market gave no signal at all as the market peaked. It reached 3.4 percent by December 1995, 4.4 percent by December 1998, and first touched its series high of 5.5 percent in June 2002, more than twelve years after the equity high. Bank balance sheet repair took longer still, and the unconventional monetary policy that accompanied it did not end until 2024.

Residential property never recovered on the series available here at all, which is the sharpest divergence in the whole episode. An investor tracking the Nikkei back to its 1989 level in 2024 could reasonably call the round trip complete. Someone who had bought Japanese residential property at the same time was still, in late 2025, roughly a fifth below their nominal purchase level and about a third below in real terms. One asset closed the loop, the other did not, and both were part of the same bubble.

What Was Specifically Different About Japan?

Four features of Japan are close to unrepeatable, and three of them sit in the supervisory and accounting rules rather than in the market.

Bank regulatory capital contained unrealized equity gains. This is the single most important structural difference and it has no analogue in the other case studies here. When the Nikkei fell, Japanese bank capital fell with it mechanically, independent of any credit loss. That coupling turned an equity bear market directly into a contraction in lending capacity. A market that behaves this way has no clean separation between an asset price shock and a credit shock, because they are the same shock arriving through two doors.

Supervision was designed to make weakness invisible. The convoy system worked as long as the economy grew, and its cost only appeared when it had to be dismantled under pressure. A resolution framework built during a crisis is slower and more expensive than one built before it, and the sixteen-person Deposit Insurance Corporation of early 1996 is the concrete measure of how far behind Japan started.

Land and equities peaked nine months apart, so there was no single moment of recognition. The Urban Land Price Index was still rising through most of 1990 while the Nikkei was falling hard. Anyone whose wealth was mostly in property spent the first year of the equity decline seeing their own balance sheet improve. Compare that with the dot-com bubble, where the overvalued asset and the falling asset were the same thing and the reversal was legible within weeks.

Policy started from an unusually low rate and had unusually little room. The discount rate reached 6.00 percent in August 1990 and 0.50 percent by September 1995, so the entire conventional cushion was spent within five years of the peak while the banking problem was still ahead. Contrast the 2020 COVID crash, a far shallower fall compressed into weeks, met with immediate and overwhelming policy support and recovered inside a year. The relevant variable is not the size of the fall but whether the institutions that finance the economy are still able to function afterwards.

Common Myths About Japan's Lost Decades

"The Bank of Japan popped the bubble." The tightening began on 31 May 1989 and the Nikkei rose another 13.6 percent to its peak seven months later, setting the all-time high in the four sessions after the third increase. The rate that arguably mattered more was the 2.50 percent held from February 1987 to May 1989, during which the Bank's own study says it looked for a chance to tighten and could not find one. Blaming the puncture obscures the inflation of the balloon.

"It was one lost decade." The equity index made distinct lows in 1992, 1998, 2003 and 2009. The banking crisis peaked in 1997 and 1998, nearly a decade after the market did. The unconventional policy response ran from 1999 to 2024. The phrase compresses at least three different adjustments with different causes into one label, which is why so many analogies drawn from Japan are wrong about which decade they mean.

"Japan just refused to write off the bad loans." Between April 1992 and March 2000, 86 trillion yen was spent on the non-performing loan problem and 110 deposit-taking institutions were dissolved. That is not inaction. The accurate criticism is about sequencing: the early response was sized to a problem the authorities expected asset prices to solve, and the political capital was spent first on the jusen, where 685 billion yen of public money bought a taboo on discussing the far larger banking problem.

"Buying after an 80 percent decline works." The Nikkei first closed 80 percent below its peak in the spring of 2003. It then fell further, to 81.9 percent below in March 2009, and did not recover the 1989 level for another fifteen years after that. The decline percentage tells you nothing about the time still to run.

"The 2024 recovery means Japanese equities were a fine long-term hold." A round trip is not a return. Thirty-four years of nominal flatness against a starting point is a real outcome for a person with a finite horizon, and the index eventually getting back does not convert that into a successful holding period. Anyone citing 2024 to argue that patience always pays should state the required holding period alongside the claim.

What a Reader Can Actually Carry Forward

Japan is most often invoked as a warning about central banks or about ageing populations. Neither is the useful lesson for someone managing a portfolio. The useful lessons are about what determines how long an unwinding takes, and about which questions are answerable in advance.

What generalizes

• Ask what the lenders own. The length of Japan's aftermath was set by the fact that bank capital and borrower collateral were exposed to the same two assets. Before assuming a market decline is a market event, work out whether the institutions financing it lose capacity at the same moment their borrowers lose collateral. Where the answer is yes, the recovery clock is a balance sheet clock.
• Credit growth is a better tell than price. Japanese fund-raising by the corporate and household sectors reached almost 14 percent growth in 1989 while consumer inflation was still under 3 percent. Borrowed money is what creates the forced sellers later.
• A quiet inflation reading is not a clean bill of health. Japan's boom ran with consumer prices between 0.1 and 0.7 percent for three years. Any framework that treats stable consumer prices as evidence that financial conditions are appropriate will miss this class of episode, and it missed this one.
• Deep drawdowns are an arithmetic problem before they are a psychological one. Recovering 81.9 percent needs about 452 percent. Set position sizes with that arithmetic in front of you rather than after the fact, which is the purpose of the sizing discipline in risk management.
• Diversify across countries as well as across assets. A Japanese investor holding domestic equities and domestic property held two positions and one risk. That is the concrete case for international investing, and it is stronger than the usual correlation argument because it is about the jurisdiction of the balance sheet rather than about price series.

What does not generalize

• The thirty-four year recovery. It required an 81.9 percent decline, a banking system whose capital fell with the index, and a resolution framework that had to be constructed from almost nothing. Remove any one of those and the duration changes completely.
• The specific policy sequence. The 2.50 percent rate held for two years and three months, the five increases in fifteen months, and the arrival at 0.50 percent within five years of the peak are a path shaped by the Plaza Agreement and by international policy coordination.
• The supervisory regime. Deposit insurance with a 390 billion yen fund and sixteen staff in 1996 is not a plausible starting point for a modern banking system, and the reforms Japan built during its crisis are now largely standard.
• "Real estate is safe." That belief did specific work in Japan because deregulation had pushed banks into property lending and because the National Land Agency had put an official forecast behind Tokyo office demand. The fault line was in one particular collateral class, and the next one will be somewhere else.

The one question worth asking now

Instead of asking whether any current market resembles Japan in 1989, ask the question a Japanese investor could have answered in 1988 without predicting anything: if the assets I own fell by half, who would be forced to sell, and would my own lender still be standing? In Japan the answer was that the banks financing the whole structure would themselves be capital-impaired by the same fall, which is why the unwinding took a generation rather than a cycle. That is not a forecast. It is a structural question about the financing behind a position, answerable today from public information, and it is the part of Japan's bubble that genuinely travels. A Japanese household in 1988 could have asked it about its own bank without forecasting anything, and stress testing and scenario analysis is the modern apparatus for putting a number on the answer.

References

Every figure on this page was verified against the following sources, each retrieved on 26 August 2026:

• Bank of Japan Institute for Monetary and Economic Studies: The Asset Price Bubble and Monetary Policy, Japan's Experience in the Late 1980s and the Lessons: by Kunio Okina, Masaaki Shirakawa and Shigenori Shiratsuka, February 2001. Source for the intensified bullish expectations framework, the paper's discussion of when the bubble period began, the Nikkei level at the Plaza Agreement, the Urban Land Price Index peak and its 1999 position, the money supply and fund-raising growth rates, the deregulation sequence and the effect of deposit rate deregulation on property lending, the bank capital base figures and their three sources, the comparison of seven failed banks with their peers, and the National Land Agency's 1985 Tokyo office demand forecast.
• Bank of Japan: The Basic Discount Rates and Basic Loan Rates: every official discount rate change and effective date quoted, from the Bank's published series covering 2 October 1972 through 8 September 1995.
• Bank for International Settlements: The financial crisis in Japan during the 1990s, how the Bank of Japan responded and the lessons learnt: BIS Papers No 6 by Hiroshi Nakaso, October 2001. Source for the convoy system and the wait-and-see policy, the Deposit Insurance Corporation's size in 1987, 1996 and 1999, the 110 dissolved institutions and the 86 trillion yen cleanup cost, the jusen loss figures, the November 1997 failures and Sanyo's call money default, the asset sizes of the failed institutions and the October 1998 nationalisation of the Long-Term Credit Bank of Japan, and the doubling of public funds to 60 trillion yen.
• Bank of Japan: Unconventional Monetary Policy Measures from the Late 1990s: the start and end dates of the zero interest rate policy, quantitative easing, comprehensive monetary easing and quantitative and qualitative monetary easing.
• Federal Reserve Bank of St. Louis FRED: Nikkei Stock Average, Nikkei 225: every Nikkei 225 closing value, decline, low, session count and recovery date quoted, computed by Swoopr Investment from the published daily series.
• Federal Reserve Bank of St. Louis FRED: Residential Property Prices for Japan: the nominal index peak in the first quarter of 1991, the 46.5 percent decline to the second quarter of 2009 and the level roughly 21 percent below the peak in the fourth quarter of 2025. The real counterpart, Federal Reserve Bank of St. Louis FRED: Real Residential Property Prices for Japan, supplies the 48.9 percent real decline and the level about 36 percent below peak in 2025. Both series come from the Bank for International Settlements.
• Federal Reserve Bank of St. Louis FRED: Consumer Price Indices (CPIs, HICPs), COICOP 1999, Consumer Price Index, Total for Japan: the annual Japanese consumer price changes quoted for 1986 through 1995.
• Federal Reserve Bank of St. Louis FRED: Infra-Annual Labor Statistics, Monthly Unemployment Rate Total, 15 Years or over for Japan: the monthly unemployment readings quoted, including the 2.0 percent March 1990 low and the 5.5 percent series high first reached in June 2002.
• Federal Reserve Bank of St. Louis FRED: Japanese Yen to U.S. Dollar Spot Exchange Rate: the monthly average of 236.53 yen in September 1985 and 123.61 in December 1988.

Figures deliberately not stated. This page does not give a price-to-earnings ratio for the Nikkei at its 1989 peak, a Japanese real gross domestic product path across the 1990s, a total-return recovery date, a figure for aggregate Japanese bank non-performing loans at any single date, or any version of the widely repeated claim comparing the value of the Imperial Palace grounds with the value of a United States state. No source verified in this session supplied them, and an unverified figure is worse than a missing one.

Method note: Nikkei 225 peak, low, decline, session-count and recovery figures labelled as computed were derived by Swoopr Investment from the daily closing values published in the FRED series NIKKEI225, retrieved on 26 August 2026. Declines are measured close to close, not intraday. The recovery date is the first session on which the index closed at or above 38,915.87, dividends excluded throughout. Where a listed calendar date was not a trading session, the table states which prior close is shown. The Urban Land Price Index and the Bank for International Settlements residential property series are different indexes with different peak dates and are never combined into a single figure.

This is educational content about a historical episode. It is not investment advice, it is not a forecast, and nothing here should be read as a claim about how any future market decline will behave.

Related Reading

• Market History Case Studies: where Japan's thirty-four year round trip sits against episodes measured in months.
• The Dot-Com Bubble: the closest comparison for a bubble whose recovery took more than a decade, and the contrast in how quickly the reversal became legible.
• The 2008 Financial Crisis: the S&P 500 bottomed on 9 March 2009 and the Nikkei on 10 March, the same week reached from opposite ends of a twenty-year gap, and recovered in five and a half years rather than thirty-four.
• Black Monday 1987: a single-session collapse, which is the shape Japan's decline conspicuously did not have.
• The 2020 COVID Crash: what an overwhelming and immediate policy response does to recovery time.
• Drawdown and Recovery Calculator: work out the gain any decline requires, including the 452 percent that an 81.9 percent fall demands.
• International Investing: why holding domestic equities and domestic property is one risk rather than two.
• The Credit Cycle and Refinancing Risk: the collateral loop that Japanese land and Japanese bank lending formed with each other.
• Sequence of Returns Risk: a Japanese retiree who began withdrawing in 1990 never saw 1989 again, whichever year the index did.
• Risk Management: sizing a position against the arithmetic of a deep drawdown rather than after it.

Frequently Asked Questions