The COVID crash, 2020 — the fastest fall and the fastest recovery

The answer

The S&P 500 peaked on 19 February 2020 and fell about 34% by 23 March 2020, the fastest fall of that size in its history. The Sensex fell from around 42,000 in January 2020 to about 25,981 on 23 March. Both markets had recovered to new highs within roughly 6 to 12 months. The speed of the fall carried no information about the size of it, or about how long it would last.

Why this costs you money

The COVID crash is the most personally useful episode in this cluster, because most readers were there.

Here is what it cost people. Not the fall — the fall cost everybody the same, and it reversed. What cost money was the decision made during the fall, and the specific decision was selling into it and then not returning.

The pattern is well documented and it is not about intelligence. It goes: sell in the third week of March because the news is genuinely frightening; feel relief; watch the market rise in April and decide it is a temporary bounce; watch it rise in May and decide it is irrational; wait for a re-test that does not come; return in October at a higher price than you sold at.

That sequence converts a temporary paper loss into a permanent realised loss, and it does so through a series of decisions that each felt sensible.

The second, quieter cost is the opposite error. Somebody who held through March 2020 concluded that holding always works, increased their position size, and discovered in a later fall that their tolerance had been measured on a decline that recovered in 5 months.

Both errors come from not knowing your own actual behaviour under stress. You have data on that. Almost nobody looks at it.

How it works

A market fall has 2 separate dimensions and people constantly confuse them.

Speed is how fast the price moves. Depth and duration are how far it falls and how long it stays there. They are close to unrelated.

Compare 3 episodes.

The 2020 fall was the fastest and the shallowest. The 1929 fall was slow and catastrophic. If speed predicted severity, the table would run the other way.

Why was 2020 different? Three reasons.

1. The cause was outside the financial system. In 1929 and 2008 the damage was inside the market's own machinery: leverage and credit. In 2020 the shock was a pandemic, and banks were better capitalised than in 2008 because of rules written after 2008.

2. The policy response was immediate and enormous. The Federal Reserve cut its policy rate to near zero on 15 March 2020 and announced large-scale asset purchases, then emergency lending facilities. Congress passed the CARES Act, about $2.2 trillion, on 27 March 2020. In India, the RBI cut the repo rate on 27 March 2020 and announced liquidity measures and a loan moratorium .

3. The market's forecast changed before the news did. Equity prices began rising in late March 2020, while infections and unemployment were still rising sharply. That is not irrationality. A share price is a claim on profits over many years, so it responds to the expected path, not to today's conditions. This is the most misunderstood feature of 2020 and it will be misunderstood again.

One mechanical detail is worth knowing. US market-wide circuit breakers, which had not been triggered since 1997, fired 4 times in 2 weeks: on 9, 12, 16 and 18 March 2020. In India, trading was halted for 45 minutes on 13 March 2020 . If you were watching, you saw the machinery from article 5 in operation.

What it tells you, and what it does not

What it tells you. Speed of decline is not evidence about outcome. Reacting to speed alone is reacting to the one variable that carries no information.

It also tells you that a recovery normally begins while the news is still bad. If your rule is "I will buy back when things look better", you have written a rule that guarantees you buy back higher, because prices move before conditions do.

What it does not tell you. It does not tell you that markets always recover quickly. That belief is the single most dangerous thing a person could take from

1. The recovery was fast because the shock was external and the policy

response was extraordinary. Neither condition is guaranteed. Japan's Nikkei 225 peaked in December 1989 and took more than 30 years to regain that level. The US market took 25 years after 1929. "It always comes back quickly" is a statement about 2 recent American decades, not about markets.

And it does not tell you that holding is always right. Holding through 2020 was right. Holding an individual company through a fall is a different decision from holding an index, because an index replaces its failures and an individual company does not.

The decision rule

Decide your response to a fall of a given size before it happens, in writing, and use your own history to set the size you can survive.

The conditional form: if the fall is market-wide and driven by something outside the companies you own, then the correct default is to do nothing and continue any regular investing. If the fall is specific to a company and the reason is a permanent change in what that business earns, then it is not a drawdown, it is a repricing, and holding is a fresh decision rather than patience.

The distinguishing question is the one from cluster article 4 on why prices move: did the index move too? A 30% fall in your stock on a day the index fell 8% is 2 different events, and only the second one is about your company.

Try this now

Fifteen minutes, using your own trade history. This is the most personal exercise in the cluster and readers consistently find it uncomfortable.

1. Open your broker app and go to your trade history or order history. Set the range to the longest period available. If your broker limits it, most Indian brokers can email you a full statement, and every US broker keeps several years of confirmations.
2. Find the worst month your portfolio has been through. March 2020 for many readers. For others it might be a later fall. Note the approximate percentage your portfolio was down from its high at that moment. This is your maximum drawdown.
3. Now the important part. Look at what you did in the 8 weeks after that low point. Write down 3 numbers:

• the value you sold in those 8 weeks,
• the value you bought in those 8 weeks,
• whether any regular investment, such as a systematic investment plan, was paused or cancelled.

1. If you sold, find the price you sold at and the price of the same instrument 3, 6 and 12 months later. Write down the difference in money. Not percentage. Money.
2. Write 1 sentence, dated, stating the largest drawdown you now know you have actually lived through, and what you did.

What you should see. Three things, in order of how uncomfortable they are.

Your remembered behaviour will not match your recorded behaviour. Most people remember holding through 2020 more calmly than the trade history shows.

Your drawdown tolerance is now a number, not a feeling. If your worst lived experience is a 25% fall that recovered in 5 months, you have not been tested by a 50% fall that takes 4 years, and you should not assume you would behave the same way.

And if you paused a regular investment, calculate what those missed instalments would be worth now. That is usually the largest single cost of the episode, and it is invisible because it never appeared as a loss.

Keep the dated sentence. Before the next fall, read it.

Three real cases

1. The United States, 19 February to 23 March 2020 — the fastest 30% The S&P 500 fell about 34% in 33 days, with 4 market-wide circuit breaker halts. The Dow's largest single-day point falls in history occurred in this period, including about 12.9% on 16 March 2020. What it turned out to be: a fall that was fully recovered by August 2020. Anybody who acted on the speed of the decline made a decision that was reversed within months.

2. India, January to March 2020 — a deeper fall and a new generation The Sensex fell from above 42,000 in January to about 25,981 on 23 March 2020 , a fall of roughly 38%. Trading was halted for 45 minutes on 13 March 2020, the first such halt in 12 years. What it turned out to be: the start of an enormous expansion in Indian retail participation. The number of demat accounts in India rose several times over in the following years. Millions of investors began after March 2020, meaning their entire experience was of a rising market.

3. Oil futures, 20 April 2020 (United States) — the part nobody predicted The May 2020 contract for West Texas Intermediate crude settled at about minus $37 per barrel. A negative price means a seller paying somebody to take delivery, because storage was full and the contract required physical delivery. What it turned out to be: a reminder that in an unprecedented event, instruments can do things their holders believed were impossible. Several retail platforms had to change how they handled the contract, and some investors lost more than their account balance.

The question that resolves it

A novice watching a fall asks: how fast is it dropping?

An expert asks: has the amount of cash these businesses will produce over the next 10 years actually changed?

In March 2020 the honest answer for most large listed companies was: 2020's profit is destroyed, 2021 is uncertain, and years 3 through 10 are largely unaffected. Since most of a company's value sits in those later years, a 34% fall was a much larger repricing than the change in expected cash justified. That was arguable at the time, with public information, and it did not require predicting the vaccine.

What would make this wrong

The falsification condition. The claim is that speed of decline does not predict depth or duration. If a study of market declines showed that faster falls were systematically deeper or longer-lasting, the article would be wrong. The historical record points the other way: the fastest large falls, 1987 and 2020, were both followed by relatively quick recoveries, while the slower falls of 1929 and 2000 to 2002 were the deepest and longest.

The honest limit. Two falls is a very small sample. Nobody should conclude that fast falls are safe. The correct conclusion is narrower: speed on its own is not evidence, so it should not be the input to a decision.

The hindsight trap, and this one is recent enough that people misremember their own view. It is now common to hear that the recovery was obvious because the central banks would obviously act. Consider what was actually known in the third week of March 2020.

Nobody knew how lethal the virus was; early case fatality estimates varied widely. Nobody knew whether a vaccine was possible, and no vaccine had ever been developed that quickly. Entire industries had revenue of zero, not reduced revenue. Corporate bond markets stopped functioning for several days in mid-March 2020, which is a genuine systemic warning. And the policy response, while fast, was untested at that scale.

The strongest argument for holding was also being made in 2008, when it was wrong for another year and a half.

So what could a careful person genuinely have seen?

Two things.

First, the balance sheets of what they owned. A company with net cash, low fixed costs and a product people would still need could be identified from its last annual report. That was a real distinction between businesses and it did predict which ones survived the shutdowns.

Second, the difference between a liquidity problem and a solvency problem. In mid-March 2020 even US Treasury bonds, the safest asset in the world, traded poorly. When the safest asset is being sold, the selling is not a judgement about value; it is people raising cash. That signal was visible in real time to anybody watching bond markets, and it is the clearest evidence that the equity selling was not primarily about company prospects.

What a careful person could not have seen was the size or the speed of the policy response, or the arrival of effective vaccines in under a year. Anybody who says they knew is describing a guess that happened to be right.

Carry this

• Speed of a fall predicts nothing about depth or duration.
• The recovery starts while the news is still bad. "I will return when it looks better" guarantees a higher price.
• Your real drawdown tolerance is in your trade history, not in your opinion of yourself.

Knowledge check

Related

• Black Monday 1987 and the machines
• The 2008 global financial crisis
• GameStop and the meme-stock mania, 2021
• ← Market history: manias, crashes and manipulation