Mark ở vũ trụ khác · Another UniverseBig Events Revisited #3 · Markets · Technology
The dot-com bubble
Nasdaq 1995–2002 · from 1,000 points up to 5,048 and back to 1,114
Whatinternet stock prices inflated, then burst
Scalethe Nasdaq lost nearly 78% from its 10 Mar 2000 peak to its 9 Oct 2002 low
Paradoxthe bubble burst, but the internet stayed and grew
Main sourcesNasdaq data, Greenspan's speech of 5 Dec 1996, the SEC, the NBER, company releases
Source: Nasdaq Composite closes of 5,048.62 (10 Mar 2000) and 1,114.11 (9 Oct 2002) per AP and Fortune, 23 Apr 2015 (secondary source); the 77.9% fall is Mark's calculation from those two figures (Reuters gives 78.4% using intraday figures).
One-minute recap
Essence
Real technologyThe internet changed how we buy, read the news and work. That part was not a bubble.
Inflated pricesThe Nasdaq went from 1,000 points (Jul 1995) to 5,048 (Mar 2000): five times higher in under 5 years.
Measuring the wrong thing1999 saw 446 IPOs, up 71% on day one on average. Traffic was treated like revenue.
The burstBy 9 Oct 2002 the Nasdaq was at 1,114, down nearly 78%. It took until 2015 to beat the peak.
New rulesStricter revenue accounting, fair disclosure, and research separated from investment banking.
Source: Nasdaq milestones from the fedprimerate.com history table, AP and Fortune (secondary source); 1999 IPOs from Jay Ritter's data (University of Florida) as republished by PBS Frontline (secondary source); SEC (SAB 101, Regulation FD, press release 2003-54; checked via excerpts, secondary source).
Part 1
The question
If the internet was real, why did the bubble still burst? Because a real technology does not make every price reasonable. People were right about the future, but paid too much, too early, and measured it in page views instead of money.
The answer is Mark's view, based on the sourced facts in Parts 2 and 4.
1995–1998ExcitementNetscape goes public. The Nasdaq passes 1,000, then 2,000. The Fed asks about "irrational exuberance".
→
1999–Mar 2000InflationIPO fever. The Nasdaq goes from 3,000 to 5,000 in four months.
→
Mar 2000–Oct 2002BurstCash runs out, companies close, the 2001 recession, new rules.
Source: the 3,000 milestone (3 Nov 1999) and 5,000 (9 Mar 2000) per AP and Nasdaq history tables (secondary source).
Part 2
Timeline
Closing milestones of the Nasdaq Composite, where most technology stocks were listed.
From 3,000 to 5,000 took four months; from 5,000 back to 1,114 took two and a half years. Only on 23 Apr 2015 did the Nasdaq close above the old peak, at 5,056.06 (not adjusted for inflation).Source: 1,000 points (17 Jul 1995, 1,005.89) and 2,000 (16 Jul 1998, 2,000.56) per the fedprimerate.com history table and Benzinga; 3,000 (3 Nov 1999, 3,028.51) per AP; 4,000 (29 Dec 1999, 4,041.46) per CNN Money, 29 Dec 1999; peak, low and the 2015 mark per AP, Fortune, 23 Apr 2015, and Reuters (all secondary sources); about 1,300 on 5 Dec 1996 per Empiricus and MarketScreener (secondary source).
17 Jul 1995 · Nasdaq passes 1,000First close above 1,000, at 1,005.89.
9 Aug 1995 · Netscape goes publicOffer price raised from $14 to $28. Trading delayed nearly two hours by a flood of buy orders. Opened at $71, closed at $58.25. The company had no profits yet.
5 Dec 1996 · Greenspan's questionThe Fed chairman asks: how do we know when "irrational exuberance" has unduly escalated asset values? The next day Tokyo fell about 3% and the US opened about 2% lower.
Late 1998 · The Fed cuts rates three timesThen from 30 Jun 1999 to 16 May 2000 it raises them six times, from 4.75% to 6.5%.
1999 · IPO fever446 IPOs, up 70.9% on average on the first trading day.
3 Nov and 29 Dec 1999 · 3,000 then 4,000Two thousand-point milestones eight weeks apart.
10 Mar 2000 · The peakThe Nasdaq closes at 5,048.62, a day after first passing 5,000.
20 Mar 2000 · "Burning Up"Barron's cites a study: at least 51 of 207 internet companies will run out of cash within 12 months.
Nov 2000 · Pets.com closesListed in Feb 2000 at $11, it announced it would shut down about 268 days later.
Mar 2001 to Nov 2001 · RecessionThe NBER dates an 8-month recession. In 2001 the Fed cut rates 11 times, to 1.75%.
9 Oct 2002 · The lowThe Nasdaq closes at 1,114.11.
Source: Netscape per AP via the Tampa Bay Times and the Virginian-Pilot, 10 Aug 1995 (secondary source); Greenspan's question per his speech at the American Enterprise Institute, 5 Dec 1996, market reaction per DeLong and Magin (2006), NBER Working Paper 12011 (via excerpts, secondary source); rates per Fed rate history as reported by CNN Money and fedprimerate.com (secondary source); IPOs per Jay Ritter's data via PBS Frontline (secondary source); Barron's, 20 Mar 2000, archived on Prof. Damodaran's NYU site (secondary source); Pets.com per summary articles, with the announcement dated 6 or 7 Nov 2000 depending on the source (secondary source); recession per NBER announcements of 26 Nov 2001 and 17 Jul 2003 (via excerpts); 2001 rates per AP via CNN Money, 11 Dec 2001 (secondary source).
Part 1
Timeline
Netscape 19951,000 → 5,048446 IPOs1,114
Part 2
Key decisions
IPO pricethe Fed's questionget big fastrole-play
Part 3
What the world learned
real revenueReg FDmoney, not views
Part 3
Key decisions
Four decisions, seen through the eyes of the time: what the decision makers knew and did not know.
1 · Raising the offer price for an unprofitable company 9 Aug 1995
Who decided
Netscape and its underwriting banks.
What they knew
The web browser was spreading fast, revenue was reportedly doubling every quarter, demand for shares was huge. The company had no profits.
What they chose
Raise the offer price from $14 to $28 and still list early.
What followed
Opened at $71, closed at $58.25, a value of about $2.9 billion. The door opened for unprofitable internet companies to go public.
Source: AP via the Tampa Bay Times, 10 Aug 1995; Virginian-Pilot, 10 Aug 1995; Bloomberg (Ritholtz, 9 Aug 2017) (secondary source).
2 · A warning as a question, with no tools used 5 Dec 1996
Who decided
Federal Reserve chairman Alan Greenspan.
What he knew
Low inflation, a growing economy, sharply rising stock prices. The Nasdaq was at about 1,300. No measure could say for sure what was "too high".
What he chose
Ask publicly about "irrational exuberance", but not use interest rates or regulation to restrain stock prices.
What followed
Markets fell for a day and then rose for more than three more years. In 2002 Greenspan explained that it is very difficult to identify a bubble definitively until after it bursts.
Source: speech "The Challenge of Central Banking in a Democratic Society", 5 Dec 1996 (checked via excerpts); DeLong and Magin (2006), NBER WP 12011; Jackson Hole remarks of 30 Aug 2002 per CNN Money (secondary source).
3 · Funding "get big fast", measured by traffic 1998–1999
Who decided
Venture capital funds, investment banks, stock market investors.
What they knew
New listings jumped on day one. Internet users were growing very fast. Whoever grabbed market share first seemed likely to win.
What they chose
Fund and list loss-making companies, scored on user growth and traffic.
What followed
1999 had 446 IPOs, up 70.9% on day one on average. By March 2000 at least 51 of 207 internet companies were forecast to run out of cash within 12 months.
Source: Jay Ritter's data via PBS Frontline; Barron's, 20 Mar 2000, "Burning Up", citing Pegasus Research International (secondary source).
4 · Raising rates against inflation, then cutting fast Jun 1999–Dec 2001
Who decided
The Fed's Federal Open Market Committee (FOMC).
What they knew
A hot economy, low unemployment, worry that inflation would return after three cuts in late 1998.
What they chose
Six rate rises, from 4.75% (Jun 1999) to 6.5% (May 2000). When the economy weakened, 11 cuts in 2001, down to 1.75%.
What followed
Money got dearer just as many companies needed to raise more. The 2001 recession lasted 8 months, shorter than the post-World War II average according to the NBER.
Source: Fed rate history via fedprimerate.com, CNN Money, 24 Aug 1999, and BCA Research (one source dates the last rise to Jul 2000, the others to 16 May 2000); AP via CNN Money, 11 Dec 2001; NBER, 17 Jul 2003 (via excerpts) (secondary source).
Mark does not grade these decisions with hindsight from after 2000. The right question is: with what was known then, was there a more reasonable option? The judgement is Mark's.
Part 4
What the world learned
What changed after the bubble burst: accounting rules, disclosure rules, how research is done, and the habits of the people who fund companies.
Rules follow the lessons; the technology keeps going. The first three milestones fixed how companies are measured and talked about. The last shows that what people bet on in 1999 was real.Source: SEC, Staff Accounting Bulletin 101 (Dec 1999; Federal Register, 9 Dec 1999); SEC, Regulation FD (Release 33-7881, effective 23 Oct 2000; per FindLaw, WilmerHale); SEC press release 2003-54 (28 Apr 2003); Census via E-Commerce Times, Mar 2000 (0.64%) and Census, Nov 2020 (16.1%). All checked via excerpts and counted as secondary sources.
Revenue must be real money
SAB 101 gave specific guidance for e-commerce companies; EITF 99-19 separated gross from net revenue reporting. Many companies switched to net, and revenue "shrank" to the part they actually kept.
Everyone hears it at once
Regulation FD bars companies from telling analysts material news privately before telling the public.
Research does not work for sales
The 2003 settlement: $875 million in penalties and disgorgement, $432.5 million for independent research, $80 million for investor education.
Source: SEC staff speech (spch495) and Journal of Accountancy, Jun 2001, on SAB 101 and EITF 99-19; FindLaw on Regulation FD; SEC press release 2003-54. Checked via excerpts (secondary source).
Survivors
Amazon posted its first net profit in Q4 2001: $5 million, though it still lost $567.3 million for the full year 2001.
The internet kept growing
By the end of 2018, 51.2% of the world's population (3.9 billion people) used the internet, per the ITU.
Central banks
The Fed kept the view that it should not fight bubbles with interest rates because they are hard to spot in advance; the question is still debated.
Source: Amazon release of 22 Jan 2002 (via excerpts); ITU press release 2018-PR40 (via excerpts); Greenspan at Jackson Hole, 30 Aug 2002, per CNN Money (secondary source). "Still debated" is Mark's view.
Part 5
Fact check: what's true and what's myth
13 things often said about the dot-com bubble. Click each line to see why.
6 SOLID
4 CAUTION
3 FALSE
The bubble was in the prices, not the technology. At the height of the bubble, online shopping was under 1% of US retail. Twenty years later it was one sixth.Source: the Census Bureau's first e-commerce report (Q4 1999) per E-Commerce Times, Mar 2000; Census, "Share of online retail sales soaring", Nov 2020 (the seasonally adjusted series gives 0.6% for Q4 1999) (secondary source).
SOLIDThe Nasdaq peaked on 10 Mar 2000
It closed at 5,048.62, a day after first passing 5,000 (AP, secondary source). The intraday high was higher, about 5,132.
SOLIDThe Nasdaq lost nearly 80% of its value
On closing prices: 5,048.62 down to 1,114.11 (9 Oct 2002), a fall of 77.9% (Mark's calculation). Reuters gives 78.4%, using the intraday peak and low (a low of 1,108.49 on 10 Oct 2002). Different measures, neither is wrong.
SOLIDIt took 15 years for the Nasdaq to regain its peak
On 23 Apr 2015 it closed at 5,056.06 (Fortune, Reuters; secondary source). That is not adjusted for inflation; adjusted, it took longer.
SOLIDPets.com listed and closed within a year
IPO in Feb 2000 at $11; announced its shutdown in Nov 2000, about 268 days later. Sources give different announcement dates, 6 or 7 Nov 2000 (secondary source).
SOLIDRight at the peak, many internet companies were close to running out of money
Barron's, 20 Mar 2000, "Burning Up": per Pegasus Research International, at least 51 of 207 internet companies would run out of cash within 12 months (archived on Prof. Damodaran's NYU site; secondary source).
SOLIDAnalysts praised stocks in public and trashed them in private
New York Attorney General's affidavit (Apr 2002): a Merrill Lynch analyst privately called a stock "a piece of junk" while it carried the firm's top public rating (CNN Money, secondary source). The 28 Apr 2003 settlement with 10 firms came to about $1.4 billion; Henry Blodget paid $4 million and was barred for life (SEC 2003-54, CNN Money; via excerpts). The firms settled without admitting or denying wrongdoing.
CAUTION"The Fed blew the bubble with cheap money"
The Fed cut rates three times in late 1998, but raised them six times from Jun 1999 to May 2000, to 6.5%, exactly when the bubble was inflating fastest (secondary source). Rates were one factor, along with expectations about technology, IPO fever and wrong metrics. There is no agreed single cause.
CAUTION"The burst caused the 2001 recession"
The NBER dates the recession from Mar 2001 to Nov 2001, 8 months. The NBER only dates turning points, it does not assign causes; it also excluded September 2001 (the 9/11 attacks) from its trough analysis (via excerpts). The burst contributed, but was not the only cause.
CAUTION"$5 trillion vanished"
The figure depends on the index and the dates: summary articles give more than $5 trillion for tech/Nasdaq stocks; CNN Money (Dec 2002) estimated about $8.5 trillion for the whole market by the Wilshire 5000. All secondary sources, with no single official number.
CAUTION"1999 IPOs doubled on day one"
The average was 70.9% across 446 deals (Jay Ritter via PBS Frontline, secondary source): very high, but not double. An average also hides the spread: a few deals that multiplied pulled the number up while many others rose little.
FALSE"The internet was just a bubble"
Stock prices were a bubble; the technology was not. Online shopping went from 0.64% of US retail (Q4 1999) to 16.1% (Q2 2020) per the Census Bureau; by the end of 2018, 51.2% of the world's population used the internet per the ITU (via excerpts).
FALSE"Greenspan warned right before the bubble burst"
The "irrational exuberance" line was on 5 Dec 1996, with the Nasdaq at about 1,300, more than three years before the peak. And it was a question, not a statement that "this is a bubble". Anyone who sold on it missed a rise of almost four times.
FALSE"Every dot-com company died"
Many died, but not all. Amazon reported a net profit of $5 million in Q4 2001, its first ever (release of 22 Jan 2002, via excerpts), though it still lost $567.3 million for 2001.
Source: as noted in each line; full list in Part 9. "Almost four times" is Mark's calculation from about 1,300 to 5,048.62.
Part 6
Assessment: how the world responded
Scores are Mark's assessment based on the sources in Part 9, not data from a study.
Early detection
3/5
Transparency
2/5
Coordinated response
3/5
Learning lessons
4/5
Got right
Public warnings in 1996 and in March 2000
The Fed cut rates quickly as the economy weakened (11 times in 2001)
Rules fixed: revenue, disclosure, research
Got wrong
Warnings with no tools or metrics anyone could act on
Research steered by investment banking interests
Traffic scored as if it were money
The world was not wrong to believe in the internet. It was wrong to pay for that belief with a yardstick that was not money.
More: why a bubble is hard to see from the inside
Everyone involved had a sensible reason: banks earned fees, funds had to keep up with the index, employees were paid in stock options, and anyone who sold early looked foolish for three years.
So the useful question at work is not "is this a bubble?" but "what am I measuring with, and who benefits when I believe that number?".
The "More" section is Mark's commentary, based on the sourced events in Parts 2 to 5.
Part 7
Work lesson: separate real technology from inflated prices, and measure in real money
Use it when your team assesses a new project, a supplier, a tool everyone is excited about, or your own product.
Each habit patches one failure of 1999. Mixing up technology and price, scoring on traffic, not counting cash left, and trusting research with strings attached.Source: the failures come from Parts 2 to 5 (Barron's, 20 Mar 2000; SEC 2003-54); the habits are Mark's suggestions.
Two separate questions"Is this technology real?" and "Is this price reasonable?". A yes to the first does not answer the second.
Measure in real moneyNet revenue (the part you keep), gross margin, customers who come back and pay a second time. Views and downloads are only signals.
How many months of cash?Cash divided by monthly burn. In 2000 this simple sum showed a quarter of the surveyed companies were about to run dry.
Who benefits if you believe it?Does the person praising a project earn a fee, commission or position from it? If so, find an independent opinion too.
"A quarter": 51 of 207 companies per Barron's, 20 Mar 2000 (secondary source).
Check your own team: 5 questions
1. What metric is your team's hottest project reported in? Is that metric money?
2. Strip out views, downloads and free sign-ups: how many customers have paid and come back?
3. At the current spending rate, how many months of budget are left? Who knows that number?
4. Who is the strongest supporter, and what do they gain if the project is approved?
5. Has the team written down its stopping conditions: which metric, which threshold, which review date?
How do we know when irrational exuberance has unduly escalated asset values?Alan Greenspan, American Enterprise Institute, 5 Dec 1996 (shortened; original sentence checked via excerpts, secondary source).
Part 8
Further reading
Robert J. Shiller, Irrational Exuberance (Princeton University Press, 2000)A book by an economist (later a Nobel laureate) on why asset prices can drift far from real value.
DeLong and Magin (2006), "A Short Note on the Size of the Dot-Com Bubble"NBER Working Paper 12011: an attempt to measure how big the bubble was.
Barron's, "Burning Up" (20 Mar 2000)The article that counted the cash left at 207 internet companies, just after the peak.
Big Events Revisited #04: The Cuban Missile Crisis (Oct 1962)Buy time and leave the other side a way back; invite dissent.
Part 9
Sources
Primary documents first: Nasdaq data, Fed speeches, SEC documents, NBER announcements, company releases. The working environment blocked the primary sites (federalreserve.gov, sec.gov, nber.org, census.gov), so every document below was checked only through search-result excerpts and is therefore counted as a secondary source.
Alan Greenspan, "The Challenge of Central Banking in a Democratic Society", American Enterprise Institute, 5 Dec 1996. federalreserve.gov
J. Bradford DeLong and Konstantin Magin (2006), "A Short Note on the Size of the Dot-Com Bubble", NBER Working Paper 12011 (market reaction on 6 Dec 1996). nber.org
NBER, Business Cycle Dating Committee announcements, 26 Nov 2001 and 17 Jul 2003. nber.org · nber.org
Jay Ritter (University of Florida), IPO statistics by year, as republished by PBS Frontline. pbs.org
Barron's, Jack Willoughby, "Burning Up", 20 Mar 2000, archived on Prof. Aswath Damodaran's site. stern.nyu.edu
Other secondary sources: AP via the Tampa Bay Times and the Virginian-Pilot, 10 Aug 1995 (Netscape); Bloomberg (Ritholtz 2017); fedprimerate.com (Nasdaq milestone table, Fed rate history); Benzinga; CNN Money, 24 Aug 1999, 29 Dec 1999, 11 Dec 2001, 15 Apr 2002, 30 Aug 2002, 18 Sep 2002, 6 Dec 2002, 25 Apr 2003; Fortune, 23 Apr 2015; Reuters via Business Standard, 24 Apr 2015; BCA Research; E-Commerce Times, Mar 2000; Journal of Accountancy, Jun 2001; Empiricus and MarketScreener (the 1,300 level); Wikipedia and summary articles (Pets.com).
VI
Mark học AI