How Stock Market Crashes Happen

Stock market crashes follow a recognizable mechanical sequence. Prices rise above what earnings can justify, often with borrowed money. A shock breaks confidence, and lenders and exchanges force owners to sell regardless of their opinions. Falling prices then trigger more forced sales while buyers step back, which is why the Dow Jones Industrial Average could fall 22.6% in a single day in 1987 with no single catastrophic headline. This article walks through each link of that chain, with the numbers from 1929, 1987, 2000, 2008, and 2020.

This article is for educational purposes only and is not investment advice.

What a crash actually is

A stock market crash is a sudden, severe fall in prices across a broad index. There is no official threshold. Market participants usually call a 10% fall from a recent high a correction and a 20% fall a bear market, and those are conventions rather than rules. "Crash" is looser still, and tends to be used when the decline is fast, steep, and disorderly.

What separates a crash from a normal correction is the feedback loop. In a correction, prices fall and buyers step in because valuations start to look attractive. That buying puts a floor under the market, and the decline fades. In a crash the loop reverses: falling prices force some owners to sell and scare others into selling, which pushes prices lower still. Lower prices produce more selling instead of more buying.

Think of a crowded theater with one exit. Each person's decision to leave is individually sensible, but the stampede that results helps nobody, and the narrower the exit, the worse it gets. In markets, the "exit" is the number of buyers willing to take shares at a given price, and it narrows precisely when everyone wants to use it.

How asset bubbles build

Every major crash begins with a period in which prices rise well above what the underlying earnings, rents, or cash flows can justify. Bubbles are not irrational at the start. A new technology, a policy change, or a burst of economic growth can legitimately push prices higher. The problem is overshoot.

Picture a rubber band stretched between two hands. One hand is genuine optimism about future earnings. The other is leverage, which means borrowed money used to buy assets. Leverage amplifies returns on the way up and losses on the way down. As prices climb, more investors borrow to participate, and lenders extend credit willingly because the collateral, the rising assets themselves, looks solid.

In the late 1920s the stretch was extreme. According to the Federal Reserve's history of the episode, purchasers of stock "put down a fraction of the price, typically 10 percent, and borrowed the rest." The Dow peaked at 381.17 on September 3, 1929. In August 1929 the Fed raised the New York discount rate to 6% to cool speculation, and the Fed's own later account says that the tightening of 1928 and 1929 slowed the U.S. economy.

The pattern is not unique to 1929. Housing in the mid-2000s and internet shares in the late 1990s followed the same outline, as the data below show: a plausible story, rising prices, wider access to credit, and a growing share of buyers who owned the asset because it was going up.

The margin call cascade: why crashes accelerate

This is the mechanical core of how stock market crashes happen, and it is counterintuitive. When you buy stock on margin, you borrow part of the purchase price from your broker. In the United States, Regulation T sets the initial requirement at 50% of the purchase, and the minimum maintenance requirement set under FINRA rules is generally 25% of the market value of the position. Brokers often impose stricter house requirements, commonly 30% to 40%, according to the Securities and Exchange Commission's investor guidance on margin.

If the value of your securities falls until your equity drops below the maintenance level, the broker issues a margin call: a demand to deposit cash or securities. If you cannot, the firm sells your holdings to restore the ratio, and the SEC notes the firm may do so without consulting you first. The selling is a contractual obligation, not a decision.

A worked example shows how fast the arithmetic moves. Suppose an investor buys $20,000 of stock using $10,000 of their own money and $10,000 borrowed. The table shows when the call arrives.

Maintenance requirement	Position value that triggers a call	Fall from $20,000
25% (the FINRA minimum)	$13,333	33%
40% (a common broker house rule)	$16,667	17%

Calculation: equity is position value minus the $10,000 loan, and a call arrives when equity falls below the requirement times position value. The 25% and 40% requirements come from the SEC's investor guidance on margin.

Now compare the 1929 setup, where the investor put down 10%. A $10,000 position carried $9,000 of debt, so a 10% fall in price wiped out the investor's entire stake. There was almost no room for error. The Fed's history records that the Dow fell nearly 13% on October 28, 1929 and nearly 12% the next day, which is enough to erase a 10% cushion twice over.

Here is the feedback loop in sequence:

1. Prices fall, triggering margin calls for leveraged investors.
2. Those investors, or their brokers, sell to meet the calls.
3. The additional selling pushes prices down further.
4. Lower prices trigger margin calls for a new set of investors.
5. The cycle repeats, faster each time.

Each round of forced selling is unrelated to anyone's view of what the shares are worth. Selling is driven by the loan contract, and the lower prices that result become the next trigger. Our explainer on margin trading and leverage covers the mechanics of borrowing against securities in more detail. The central bank's role is the other half of the story. The Fed's own history of the Great Depression notes that the money supply fell by nearly 30% from the fall of 1930 through the winter of 1933 and that the Fed failed to act as a lender of last resort during the banking panics of those years. How central banks respond to these events is essential context, because it helps explain why a bad crash became the Great Depression.

Why liquidity disappears

A price is not a fact about the world. It is the result of the most recent trade between a willing buyer and a willing seller. In calm markets, specialists and trading firms stand ready to take the other side of trades, and they earn a small spread for doing so. When prices move sharply, those firms face a choice: keep buying a falling market and risk heavy losses, or step back.

Stepping back is usually rational for each firm, and fatal for the market. Bid orders thin out, the gap between the best buy and sell prices widens, and a modest sell order moves the price by far more than it normally would. This is called a liquidity vacuum, and it is why crashes often look like prices "gapping" down through levels where no trades happened at all. Our article on why liquidity matters in markets explains how the depth of buyers and sellers shapes price stability.

Think of liquidity as the shock absorbers on a car. On a smooth road you never notice them. Hit a pothole at speed with worn absorbers and every bump goes straight through the frame. Crashes are, in large part, the moments when the market's shock absorbers wear out at the same time.

Herd behavior, loss aversion, and panic

Forced selling from margin calls is only part of the story. Discretionary sellers, who are not compelled to sell, amplify a crash through herd behavior. The logic is rational-seeming: if everyone else is selling, maybe they know something I do not. Each person discounts their own information and copies the crowd, and when the crowd is panicking, the copying accelerates the panic. Economists call this an information cascade, and our article on social proof covers the same dynamic outside finance.

Loss aversion makes it worse. In the 1992 version of prospect theory, Amos Tversky and Daniel Kahneman estimated that a loss weighs roughly 2.25 times as heavily as an equal gain, in the experimental choices they studied. In market terms, a $10,000 loss hurts about twice as much as a $10,000 gain pleases, so investors become hypersensitive to falling prices. Behavioral economics documents this pattern, and our piece on why losses feel stronger than gains looks at the psychology.

The result is that rational individual decisions combine into an irrational collective outcome. No single seller causes a crash. The crash is the sum of thousands of defensible choices made in the same direction at the same time.

Black Monday, 1987: the crash with no obvious cause

On Monday, October 19, 1987, the Dow Jones Industrial Average fell 22.6%, or 508 points, in a single session, the largest one-day percentage drop in its history. The Federal Reserve's history of the event emphasizes market structure rather than a single piece of bad news.

One factor was portfolio insurance, a hedging technique that used futures and options to limit losses by selling as prices fall. If many investors use the same rule, a falling market triggers selling from all of them at once, which pushes prices lower and triggers more selling. The Fed's account also points to mismatched settlement cycles, three days for stocks against one day for options and futures, and to exchanges that had no ability to halt large sell-offs.

The Fed's response on October 20 was a one-sentence statement from Chairman Alan Greenspan that the central bank "affirmed today its readiness to serve as a source of liquidity to support the economic and financial system." Behind the scenes, it encouraged banks to keep lending on normal terms. The Dow regained 288 points, about 57% of the Black Monday loss, in the next two sessions, and U.S. stock markets surpassed their pre-crash highs in less than two years. Unlike many earlier financial crises, the crash was not followed by a recession or a banking crisis.

The contrast with 1929 is instructive. The 1987 crash was as steep as any in history for a single day, but the credit system stayed open. The Fed supplied liquidity quickly, banks kept lending, and the spiral stopped at the stock market's edge. Whether a crash becomes an economic catastrophe often depends less on how far prices fall than on whether the banking system keeps working.

Flash crashes: when machines amplify the fall

Modern markets added a faster variant. On May 6, 2010, U.S. equity prices fell sharply and then rebounded within minutes. The Dow dropped by roughly 1,000 points (about 998.5), about 9% from its open, at its low before recovering much of the loss the same afternoon.

The official explanation came from the SEC and the Commodity Futures Trading Commission in a joint report dated September 30, 2010. It described a large fundamental trader that started a program to sell 75,000 E-mini S&P 500 futures contracts, valued at about $4.1 billion, using an automated algorithm set to target a share of recent trading volume without regard to price or time. As prices dropped, other firms' trading algorithms and the withdrawal of liquidity helped turn a large sale into a plunge.

A second thread came later. In 2016 the CFTC announced a consent order against Navinder Sarao, a London-based trader who admitted manipulating E-mini S&P 500 futures through spoofing, which means placing large orders with no intention of letting them execute in order to create a false picture of supply and demand. According to the CFTC, his "dynamic layering" program ran for a cumulative 4 hours and 25 minutes on May 6, 2010, modified its orders more than 81,000 times, and executed only 81 lots. At times its orders represented 20% to 29% of the sell side of the order book, and the CFTC said the resulting imbalance contributed to a sudden loss of liquidity, together with other events. Sarao also pleaded guilty to spoofing and wire fraud in a related criminal case.

So the cause was a combination, not a single villain: a very large automated sale, thin liquidity, and one trader who was pressuring prices. The underlying mechanism is the same as in 1929 or 1987. Falling prices triggered selling, which triggered more falling prices. What changed is the speed, with the cycle now completing in minutes. Our article on algorithmic trading and neural networks explains how automated strategies work and where they can go wrong.

Circuit breakers: the modern safety valve and its limits

After 1987, regulators introduced trading halts to interrupt the feedback loop and give markets time to find buyers. The current U.S. market-wide circuit breakers are keyed to declines in the S&P 500 from the prior day's close:

Level	Decline	Result
Level 1	7%	15-minute trading halt
Level 2	13%	15-minute trading halt
Level 3	20%	Trading halted for the rest of the day

Level 1 and Level 2 halts apply only between 9:30 a.m. and 3:25 p.m. Eastern time. A 7% or 13% decline at or after 3:25 p.m. does not halt trading, while Level 3 applies at any time during regular hours. Each level can trigger only once per day. Market-wide halts cover all U.S. equity symbols, plus stock options and certain index-related futures, according to the trading firm Optiver's explainer, but not extended-hours trading.

The breakers have been used less often than their reputation suggests. A Level 1 halt was triggered four times in 2020, on March 9, 12, 16, and 18, during the first wave of the COVID-19 pandemic. No Level 2 or Level 3 halt has occurred to date. An earlier version of the rule, tied to the Dow rather than the S&P 500, triggered on October 27, 1997, during the Asian financial crisis.

Do they work? The stated purpose is to pause trading so participants can "stop and assess the situation," and a reopening auction restores an orderly price. They also have limits. A 15-minute halt does not change the leverage positions that are forcing sales, so when trading resumes, forced selling can resume with it. Markets outside the United States are not covered, and no circuit breaker can stop a crash that unfolds over weeks rather than hours, which describes most of the large ones.

Five crashes compared

Looking across the major crashes reveals a consistent pattern rather than a collection of unique disasters. The table summarizes five episodes, with the size of the fall measured in each market's own terms.

Episode	Size of fall	Time to recover	What drove it
1929 to 1932	Dow down 89% from September 3, 1929 to July 8, 1932	Dow regained its 1929 high on November 23, 1954 (about 25 years, nominal)	10% margin buying, bank panics, a Fed that did not act as lender of last resort
October 1987	Dow down 22.6% in one day	Prior highs surpassed in under two years	Portfolio insurance, settlement mismatches, no trading halts
2000 to 2002	Nasdaq down about 78% from March 10, 2000 to October 9, 2002	About 15 years for the price index	Valuation bubble in technology shares
2007 to 2009	S&P 500 down about 57%	About five and a half years	Housing credit, high leverage at banks, funding runs
February to March 2020	S&P 500 down 33.9% in 33 days	New record close on August 18, 2020	Pandemic shock, then large policy support

Sources: Federal Reserve History essays on the 1929 and 1987 crashes; DQYDJ Nasdaq drawdown history (a total-return measure showing a 73.9% decline and nearly 15 years underwater); The Reformed Broker and S&P Dow Jones Indices data as reported in the financial press for the S&P 500 in 2007 to 2009 and 2020. Different data providers use closing prices or total returns and report slightly different percentages.

The dot-com crash illustrates that not every crash needs a leverage cascade to be severe. The Nasdaq Composite closed at 5,048.62 on March 10, 2000, and its closing low was 1,114.11 on October 9, 2002. Prices of technology shares had run far ahead of earnings, so the repricing of speculative assets, together with herd behavior, played the central role. The index did not regain its 2000 peak for about 15 years.

The 2008 financial crisis combined an asset bubble in housing with extreme leverage and a crisis in short-term funding. By 2007, large investment banks were running assets of roughly 30 times their equity. For Lehman Brothers the reported figure was about 31 to 1, which would mean that a decline of about 3% in the value of its assets could erase its equity. Margin-call-style dynamics moved through bank lending markets rather than individual brokerage accounts. The Fed's history reports that the recession lasted 18 months from December 2007 to June 2009, output fell 4.3% from peak to trough, and unemployment more than doubled from under 5% to 10%. The mechanics of the recession that followed show how a market crash transmits into the real economy.

The 2020 crash was the fastest decline of its size in modern U.S. history. The S&P 500 fell 33.9% from its February 19 peak to its March 23 low, 33 days later. It closed at a record on August 18, 2020, about six months after the peak, helped by large fiscal and monetary support. The speed of that rebound reflected a specific policy response, and past recoveries do not guarantee future ones.

What every major crash shared: valuations that depended on rising prices, borrowed money or fragile funding, a trigger that shifted sentiment, and a mechanical loop of forced selling that took prices far below any sober estimate of value.

What happens to the real economy after a crash

A stock market crash is not just a paper loss for investors. It reaches the real economy through two main channels.

The wealth effect is the more direct. When household and institutional portfolios shrink, consumption and business investment tend to fall. The channel is stronger when stock ownership is widespread. Gallup's annual survey found that 62% of Americans reported owning stock in 2025, the same as in 2024, counting individual shares and holdings in mutual funds and retirement accounts. Ownership averaged 62% from 2001 to 2007, fell to 52% in 2013 and 2016, and has been above 60% since 2023, so a large share of households feels a crash in some form.

Credit tightening is often more damaging. Banks and other lenders that hold securities as capital find that capital impaired, and they pull back lending to businesses and consumers. The cost and availability of credit then change for everyone, not just investors. This is how the 2008 stock decline became a credit crisis, and it explains why inflation and interest-rate policy matter in the aftermath, when central banks face conflicting pressures between supporting growth and keeping prices stable.

The two most recent crashes had very different effects. In 2020, U.S. real GDP fell at an annual rate of 31.4% in the second quarter, according to the Bureau of Economic Analysis, after a 5.0% annualized fall in the first quarter. Annualized means the quarterly change is expressed as if it continued for a full year, so the actual drop in output across the quarter was much smaller than 31.4%. Output then rebounded quickly. The 2008 crisis produced a slower, grinding contraction, and home prices fell more than a fifth on average nationwide between early 2007 and mid-2011, per the Fed's history. What a recession is and how it is dated is covered separately.

Common confusions about crashes

A crash is not the same as a recession. The 1987 crash did not lead to a recession, while the 2008 crash was part of one. The link depends on whether the banking system and credit keep functioning.

A fall is not a verdict on the economy. Prices can fall because of a sudden shift in sentiment or forced selling even when current earnings have not changed. That is one reason crashes often reverse partially within days, as the Dow did after Black Monday.

Circuit breakers do not set a floor. They pause trading, but they do not set a price. A market that reopens after a halt is free to keep falling.

Prediction is a different task from explanation. Researchers can identify conditions that have historically preceded crashes, such as elevated valuations, high leverage, and rising interest rates. Timing remains unpredictable. Overvalued markets have stayed overvalued for years, and many forecast crashes never came. Understanding how the loop works is far easier than knowing when it will start.

Recovery is not guaranteed in the same way every time. U.S. indexes have recovered from every crash discussed here, but recovery times ranged from about six months to about 25 years. Individual companies and other countries' markets have fared differently. The yield curve and other indicators attract attention as warning signs, though none of them has a perfect record, and for the broader picture of how markets function, see how the stock market works.

Frequently asked questions

Q: What causes a stock market crash? A: Crashes come from the interaction of three things: an overvalued market, often financed with leverage; a trigger that shifts sentiment; and a feedback loop of forced selling, especially from margin calls, that pushes prices far below fundamental value. No single factor is enough. In 1929, margin loans left investors with a 10% cushion. In 1987, mechanical selling met a market with no trading halts.

Q: How quickly can a stock market crash happen? A: Very quickly. Black Monday in 1987 was a 22.6% single-day fall. In 2010 the Dow dropped roughly 1,000 points in minutes before recovering much of it the same afternoon. In 2020 the S&P 500 lost 33.9% in 33 days. Other crashes, such as the 2000 to 2002 technology decline and the 2007 to 2009 financial crisis, unfolded over more than two years.

Q: What is a margin call and why does it make crashes worse? A: A margin call is a broker's demand for more cash or securities when the equity in a leveraged account falls below the maintenance requirement, which is generally at least 25% of the position's value under FINRA rules. If the investor cannot pay, the broker sells holdings, sometimes without notice. That forced selling lowers prices and triggers calls on other accounts.

Q: What is a circuit breaker in the stock market? A: It is an automatic trading halt triggered when the S&P 500 falls by a set percentage from the prior close. At 7% and 13%, trading halts for 15 minutes if the decline happens before 3:25 p.m. Eastern time. At 20%, trading stops for the rest of the day. The Level 1 halt was triggered four times in March 2020, and the higher levels have never been used.

Q: Can a stock market crash be predicted? A: Not reliably. Elevated valuations, high leverage, and rising interest rates have preceded past crashes, but timing is hard to know. Bubbles can last years longer than analysts expect, and many predicted crashes never arrived. The loop that turns a decline into a crash begins with triggers that are, by nature, difficult to forecast.

Q: Does the stock market always recover after a crash? A: U.S. broad indexes have recovered from every crash covered here, but the timing varied widely. The Dow regained its 1929 high in November 1954, about 25 years later in nominal terms, after falling 89%. The Nasdaq took about 15 years to recover from the 2000 to 2002 decline, and the S&P 500 took about five and a half years after 2008. Recovery for an index does not mean recovery for every company, and past results do not guarantee future ones.

The takeaway

A crash is best understood as a failure of the market's shock absorbers, not as a sudden change in what companies are worth. Leverage turns a price decline into a legal obligation to sell. Lost liquidity turns that selling into a price collapse. Human psychology then turns a collapse into a stampede. Each stage is mechanical, which is why crashes across nearly a century share a recognizable shape even though the triggers differ.

The difference between a bad week and a lost decade usually lies in what happens next. In 1987, the Fed signaled it would supply liquidity, the banks kept lending, and the damage stayed within the stock market. In 1929 to 1933, the money supply shrank by nearly 30% and the banking system broke. The lesson is that the speed of a fall tells you less about its consequences than the health of the credit system that sits behind it.

Sources

• Federal Reserve History, The Stock Market Crash of 1929 and The Stock Market Crash of 1987.
• Federal Reserve History, The Great Depression and The Great Recession and Its Aftermath.
• U.S. Securities and Exchange Commission, Investor guidance on margin risks.
• Optiver, Market-wide circuit breakers explained.
• Commodity Futures Trading Commission, Sarao consent order, press release 7486-16.
• DQYDJ, Nasdaq drawdown history.
• Gallup, Percentage of Americans who own stock.
• Bureau of Economic Analysis, GDP third estimate for the second quarter of 2020.
• SEC and CFTC, Findings Regarding the Market Events of May 6, 2010 (September 30, 2010).
• Tversky and Kahneman, "Advances in Prospect Theory: Cumulative Representation of Uncertainty," Journal of Risk and Uncertainty 5 (1992).

---

Read next:

• How Inflation Works
• How Recessions Happen
• How Central Banks Work