Articles · Insights
How bubbles form: the stages of a mania, from boom to bust
The Minsky–Kindleberger stages of a bubble applied to 1929, Japan, dot-com, housing and Bitcoin, and what research says about spotting one in advance.
On this page
A bubble is a rise in the price of an asset far beyond what its income or use can justify, sustained mainly by the expectation that someone will pay even more later. Bubbles are easy to name after they burst and very hard to identify while they are inflating, because each one comes with a plausible story about why this time the old valuations no longer apply. Some of those stories turn out to be partly true. The internet did change the world; most internet stocks of 1999 still lost nearly everything.
This article is about the structure that episodes from the 1630s to the 2020s share: how a genuine change in the world turns into a self-feeding rise in prices, why borrowed money is usually at the center, and how the end tends to arrive. It sets five well-documented bubbles side by side, summarizes what research says about spotting one in advance, and is honest about where the models fail. The psychology of following the crowd is covered separately in Herd behavior and FOMO in investing: cascades, manias and panics; here the focus is on the cycle itself.
The Minsky–Kindleberger stages
Hyman Minsky's financial instability hypothesis, set out most compactly in a 1992 working paper, argues that stability itself breeds instability. He sorted borrowers into three kinds:
- Hedge borrowers can pay both interest and principal from their income.
- Speculative borrowers can pay the interest but must keep rolling over the principal.
- Ponzi borrowers can pay neither from income. They depend on the asset rising in value so they can sell or borrow more.
After years without a crisis, lenders relax and borrowers take more risk, so the economy drifts from hedge toward speculative and Ponzi finance. Then a modest shock, such as a rise in interest rates, forces the Ponzi borrowers to sell, prices fall, and selling spreads to everyone else. The moment when that happens is often called a Minsky moment.
The economic historian Charles Kindleberger built a narrative model on Minsky's work in Manias, Panics, and Crashes (1978), studying centuries of episodes. His stages are a useful map:
- Displacement. A real change opens new profit opportunities: a technology (railways, radio, the internet), a financial innovation, a change in policy, or the end of a war.
- Boom. Early investors make money. Credit expands to finance more buying: margin loans, mortgages, bank lending against rising collateral.
- Euphoria. Buying for the asset's income gives way to buying to resell at a higher price. Valuation rules are dismissed as outdated. New buyers, often inexperienced, arrive late.
- Distress. Insiders and early investors begin selling. Prices stop rising and wobble. Interest rates are often rising by now, making leveraged positions costly.
- Revulsion and panic. A trigger, sometimes small, starts forced selling. Lenders cut credit, collateral values fall, and the asset becomes impossible to sell at almost any price.
The stages are a description, not a clock. They tell you what the end of a bubble tends to look like, not when it will come.
Five bubbles compared
The table and chart set out five episodes with good price records. Older manias, such as the Dutch tulip episode of the 1630s (Tulip mania (1636–37): what really happened in the Dutch bulb bubble) and Britain's South Sea Company in 1720 (The South Sea Bubble of 1720: debt swaps, Newton and the crash), fit the same pattern, but their prices survive only in fragments; recent archive research on 1720 share prices, for example, shows a rise from January to June, a flat summer and a collapse from September.
| Episode | Displacement | Credit fuel | Peak to trough | Decline |
|---|---|---|---|---|
| US stocks, 1920s | Radio, cars, electrification | Broker loans to margin buyers | Dow 381.17 (Sep 3, 1929) → 41.22 (Jul 8, 1932) | −89.2% |
| Japan, 1980s | Export boom, financial deregulation | Bank lending against land and shares | Nikkei 38,915.87 (Dec 29, 1989) → 7,054.98 (Mar 10, 2009) | −81.9% |
| Dot-com, late 1990s | The commercial internet | IPO and venture money; some margin | Nasdaq 5,048.62 (Mar 10, 2000) → 1,114.11 (Oct 9, 2002) | −77.9% |
| US housing, 2000s | Securitization, low rates | Mortgages with small down payments | Case-Shiller national, Jul 2006 → Feb 2012 | −27.4% |
| Bitcoin, 2017 | A new digital asset | Exchange leverage; new token issues | about $19,800 (Dec 2017) → about $3,200 (Dec 2018) | about −84% |
Three things stand out.
Leverage decides the damage to the economy, not the size of the fall. Home prices fell least of the five, about 27% on the current Case-Shiller series, yet the housing bust did the most economic harm, because it was financed by mortgages that sat on bank and investor balance sheets around the world (see The 2008 financial crisis: housing, Lehman, TARP and the 57% fall). The dot-com crash wiped out more market value but was financed mostly with equity, and the recession that followed was mild.
The displacement was usually real. Radio, cars, the internet and global banking did transform their eras. What failed was the price paid for that future, and the assumption that every company in the new industry would be a winner. See The dot-com bubble: the 1995–2002 boom and the 78% Nasdaq fall for how much of the Nasdaq's fall came from companies that never earned a profit.
Recoveries can take a generation. The Nasdaq Composite regained its March 2000 high in April 2015. The Nikkei 225 did not close above its 1989 record until February 22, 2024. The Dow took until 1954 to regain its 1929 peak, as told in The crash of 1929: causes, the 1929–32 decline and lessons. Dividends shorten those waits for a reinvesting holder, but the price records show how long a bubble's overhang can last.
Can a bubble be spotted in advance?
This is one of the oldest arguments in finance. Eugene Fama, the economist most associated with efficient markets, has long argued that "bubbles" cannot be reliably identified in advance, because large price rises are not reliably followed by low returns.
A 2019 study by Robin Greenwood, Andrei Shleifer and Yang You, pointedly titled Bubbles for Fama, tested this on US industry stock returns from 1926 to 2014 and on international sectors. They defined a crash as a 40% fall within two years and found:
- After an industry rose 50% more than the market over two years, the chance of a crash was about 20%. After 100%, it was about 53%; after 150%, about 80%.
- Even so, average returns after such run-ups were not much lower than normal. Many industries kept rising for a while, and some never crashed.
- Run-ups that came with high volatility, rising share issuance, high turnover and an accelerating price path were the most likely to end badly.
So the evidence supports both sides in part: very large run-ups do sharply raise the probability of a crash, but betting on the timing has been a poor strategy. Robert Shiller, who warned about stocks in 2000 and housing in 2005, defines a bubble as a social epidemic in which price rises themselves spread the enthusiasm and the stories that justify it. He has also stressed that a bubble does not have to burst suddenly; it can deflate over years.
Warning signs, and their limits
None of these is proof, and each has given false alarms. Together they describe the euphoria stage:
- Prices far above long-run measures of value, such as earnings, rents or replacement cost, with the gap explained away by a new-era story.
- Rapid growth in borrowing to buy the asset: margin debt, mortgage terms, leverage offered to small investors.
- A surge of new supply: IPOs, new tokens, new housing starts. Insiders sell when buyers will pay the most.
- New, inexperienced buyers and a wave of media attention, which the herd-behavior lesson covers.
- Rising interest rates, which often mark the shift from boom to distress.
The limits are just as important. Alan Greenspan asked in a December 5, 1996 speech how anyone would know when "irrational exuberance" had pushed up asset values. The Nasdaq Composite, near 1,300 at the end of 1996, went on to close above 5,000 in March 2000. A warning can be right and still years early, and an investor who sells or bets against a bubble too soon can lose as much as one who stays too long.
How Finabase helps
You can see the shape of several episodes on real data. The Nikkei 225 chart and the Nasdaq Composite chart show the long climbs back from the 1989 and 2000 peaks. The Screener can sort stocks by valuation multiples and one-year performance, a way to see where the strongest run-ups are concentrated today, and the Scanner flags unusual moves. None of these will tell you whether something is a bubble; they help you see how far prices have moved relative to earnings and to history.
Frequently asked questions
What are the stages of a bubble?
In the Minsky–Kindleberger model: displacement (a real change), boom (rising prices and credit), euphoria (buying to resell), distress (insiders sell, prices stall) and revulsion or panic (forced selling and a credit squeeze).
What is a Minsky moment?
The point at which over-borrowed investors, who depend on rising prices, are forced to sell, starting a chain of falling prices and further forced selling. It is named after the economist Hyman Minsky.
Can you tell when a bubble will burst?
Not reliably. Research shows very large run-ups raise the chance of a crash, to about one in two after an industry doubles relative to the market in two years, but many keep rising for a long time, and the trigger is rarely predictable.
Are all big price rises bubbles?
No. Prices can rise a long way because earnings or the value of an asset genuinely grew. A bubble is defined by prices running far ahead of any reasonable estimate of value, usually with borrowed money and buying driven by expected resale rather than income.
Summary
Bubbles follow a recognizable arc. A real change sparks rising prices, credit expands to finance more buying, euphoria replaces valuation, insiders sell, and a trigger starts forced selling that feeds on itself. Minsky explained why long calm periods encourage the borrowing that makes the end so violent, and Kindleberger traced the pattern across centuries. Five well-recorded episodes lost between 27% and 89% from peak to trough, and the damage to the wider economy depended more on leverage than on the size of the fall. Research confirms that huge run-ups raise the odds of a crash but offers no reliable way to time one.
Check your understanding
4 questions on this lesson. Each answer comes with the reason, and nothing is scored.
In Minsky's terms, a property buyer who can pay neither the interest nor the principal from rental income and relies on prices rising is engaged in what kind of finance?
- Hedge finance
- Speculative finance
- Ponzi finance
Show the answer
C. Ponzi finance Ponzi finance depends on rising asset prices to sell or borrow more. When prices stop rising, these borrowers are forced to sell first.
US home prices fell about 27% from 2006 to 2012, far less than the Nasdaq's 78% fall after 2000, yet the housing bust did more damage to the economy. Why?
- Homes are more important than stocks
- Housing was financed largely with mortgage debt held across the financial system, so losses hit lenders and spread, while dot-com losses fell mostly on equity holders
- The housing decline happened faster
Show the answer
B. Housing was financed largely with mortgage debt held across the financial system, so losses hit lenders and spread, while dot-com losses fell mostly on equity holders Leverage decides how far a bust spreads. Losses on borrowed money become losses for banks and their creditors.
Research on US industries found that after a 100% two-year run-up beyond the market, about half later crashed 40%. What else did it find?
- Average returns after such run-ups were not unusually low, so timing a short sale was hard
- Every such industry crashed within six months
- Run-ups never predict crashes
Show the answer
A. Average returns after such run-ups were not unusually low, so timing a short sale was hard Big run-ups raise the odds of a crash, but many kept rising for a while, so the evidence supports caution more than market timing.
Which describes the "displacement" stage of a bubble?
- Insiders begin selling and prices stall
- A real change, such as a new technology or a policy shift, opens up new profit opportunities
- Lenders cut credit and forced selling begins
Show the answer
B. A real change, such as a new technology or a policy shift, opens up new profit opportunities Bubbles usually start from something genuinely new; the error comes later, in the prices paid for that future.
Sources and further reading
- Minsky — The Financial Instability Hypothesis, Levy Economics Institute Working Paper No. 74 (May 1992)
- Kindleberger — Manias, Panics, and Crashes: A History of Financial Crises (1978; later editions with Robert Aliber)
- Shiller — Irrational Exuberance, 3rd edition (Princeton University Press, 2015)
- Greenwood, Shleifer and You — Bubbles for Fama, Journal of Financial Economics 131(1), 2019 (NBER Working Paper 23191)
- Greenspan — The Challenge of Central Banking in a Democratic Society, December 5, 1996
- S&P CoreLogic Case-Shiller US National Home Price Index (CSUSHPINSA), via FRED
- Nippon.com — Nikkei index sets first record high since 1989 (February 2024)
- Fortune — Nasdaq surpasses internet-bubble peak (April 23, 2015)
- Fortune — Bitcoin’s all-time high, one year later (December 17, 2018)
- Frehen, Goetzmann and Rouwenhorst — New evidence on the first financial bubble, Journal of Financial Economics 108(3), 2013
- S&P Dow Jones Indices and Nasdaq — published index closes for the Dow (1929–1932) and Nasdaq Composite (2000–2002)
