Case: Dot-Com Valuations, 2000

Reverse the price into a forecast

Take the price and ask what has to happen, rather than forecasting and then checking the price. At a 100× multiple with a 16× terminal multiple, the earnings have to grow roughly sixfold just to hold the price flat, and more than that to justify it. If the business is going to earn a mature operating margin of 20% — generous for a competitive industry and above most hardware and services companies of the era — the revenue implied is a multiple of gross domestic product. The arithmetic does not require scepticism about the internet to fail; it fails on its own. Now add the second half of the account, which is what the survivor stories miss: the companies that were going to be large did become large, and the shareholders still lost money. Cisco, the canonical example, grew revenue for years and the shares took more than a decade to recover, because the multiple fell from over a hundred to the market range and the earnings growth was spent on making the multiple plausible rather than on adding value. That is not a failure of the business. It is the arithmetic of paying today for decade-out earnings, and it is why the entry multiple is the largest single determinant of a long-horizon return. Two routes to a flat price — Earnings grow 6×, multiple falls 100→16: price ≈ unchanged — and it took a decade ← · Earnings grow 2×, multiple falls 100→16: price falls about two thirds · Earnings fall, multiple holds: the price follows earnings, but the multiple was the risk · The lesson: the multiple is the assumption with the most leverage on the outcome A high multiple is not automatically wrong: it can be justified by growth that has already been contracted, by network economics that make margin durable, or by a genuinely larger opportunity. The test is whether you can state the growth and the margin that make it work, in numbers, and then say what would falsify them.

Why the assumptions could not all hold

Three assumptions were needed at once, and each was individually defensible. Growth had to be very fast for a long time. Margins had to end high and stay high. And the terminal multiple had to stay in the range that the growth implied. The problem is that the three are linked by competition: fast growth invites entry, entry compresses margins, and compressed margins remove the justification for the multiple. That interaction is the mechanism behind every bubble’s mathematics, and it is why the collapse is not a sentiment event that could have been avoided with better information — the information was there, in the requirement that a sector eventually earn more profit than the economy in which it operates. What made it rationally persuasive was a real observation: the internet did change the world. That part was correct, and the value created was enormous — it simply accrued to consumers and to a later generation of companies rather than to the buyers of 2000. This is a general pattern worth remembering, because it recurs: a true technology, priced as though the true part implies the valuation part. Railroads, radio, airlines and the internet all produced large social returns and repeated the same shareholder experiment. The accounting completed the picture. Revenue-recognition aggressiveness, advertising barter, and the treatment of gross versus net revenue made reported growth faster than the economics, and the companies that later restated were precisely those whose multiples had relied on the metric. The practical residue of the episode is the habit of reading the cash flow statement alongside the income statement (F4, F8) and asking what the business actually earns per dollar of capital, rather than what the narrative says it will. • Fast growth invites entry; entry compresses margins; compressed margins remove the multiple. • A true technology does not imply the valuation — the value can accrue to consumers instead of shareholders. • Reported growth was flattered by revenue recognition and gross-versus-net presentation. • The survivors grew and their shareholders still lost: the entry multiple decided the outcome. The dot-com peak is often described as a moment of insanity. Read the contemporary research and it is better described as an unusually coherent set of assumptions about a genuinely transformative technology, which is what makes it a repeatable lesson rather than a curiosity.

Fifteen years, and the ones who survived

The dot-com bust is usually told as a crash, and the part that matters for a long-term owner is the recovery: the Nasdaq Composite peaked in March 2000 and did not close above that level again until **April 2015**. Fifteen years is the number to hold on to, because it is what multiple compression does when you are right about the growth. A business could have doubled its revenue and earnings several times over across that period and still delivered a flat or negative return to the person who bought at the top, since the price paid was a claim on decade-scale growth that had already been capitalised. This lesson’s multiple-compression lab is that arithmetic, run forward. The counterexample inside the same period is equally instructive, and it is why the lesson is not simply a warning about bubbles. Several of the companies that fell the furthest in the bust — names whose shares lost well over ninety percent of their value from the 2000 peak — went on to become among the best investments of the following two decades, because the survivors were real businesses whose revenue kept compounding while their share prices had been reset. The same price that made the buyer of March 2000 wait fifteen years made the buyer of 2002 a fortune. The valuation did the damage; the business did not. Holding both halves together gives the practical rule. Buying an exceptional business at an exceptional price can be a worse decision than buying an ordinary business at an ordinary one, and the reason is not that growth disappoints but that expectations are a second variable, priced alongside the cash flows and capable of moving more than they do. A long holding period is not protection from buying at an unreasonable multiple; it is time for the multiple to matter. The reflex to check: when a valuation today requires the company to become one of the largest businesses in the world, that is the March-2000 assumption, whoever is making it. Reverse the price into the forecast before the optimism becomes a discounted cash flow.

Who was selling, and to whom

Valuation arithmetic explains what the prices required; it does not explain why they could be reached, and the second question has a supply-side answer. A share price is set by the balance between the shares available to trade and the money trying to own them, and in 1999 and early 2000 the available supply was unusually small and shrinking at the worst possible moment. New issues came to market in small floats, deliberately: a company would list a fraction of its shares and let the thinly traded remainder set the price, so a modest inflow of money could move the quotation a long way. The lock-up agreements that kept insiders from selling expired on a published schedule, and each expiry was a moment when the float expanded. The mechanics are worth naming because they recur. An initial public offering is priced by the underwriters and then trades, and the difference between the offer price and the first print is the *pop*, which is money the selling shareholders do not receive. The shares that do trade are the float — a minority of the total — and index inclusion is a further mechanical step, because a fund tracking the index must buy whatever weight the company is given, regardless of the price. Add a lock-up expiry that admits insider supply and you have three separate moments where the balance of shares and money changes for reasons unrelated to the business. Those moments answer a question the earlier analysis leaves open: how a demand-driven re-rating happens without anyone forecasting the earnings. Money arrives, the float is small, the price rises, the rise is written about, more money arrives — a loop that runs on its own momentum and is broken either by supply arriving or by the money stopping. The collapse of 2000–2002 was both. Lock-ups expired through 2000 and early 2001, adding tradable shares to names whose prices had been set in thin markets, while the flow of new capital into the sector slowed and then reversed. Two lessons follow, and they are practical rather than historical. First, when a valuation looks stretched, ask what fraction of the company trades and what is scheduled to. A small float with a lock-up ahead is a supply event with a date on it, and it is visible in the filings before it happens. Second, treat index inclusion and large passive flows as mechanical demand rather than as an opinion: they change the balance of buyers to shares without changing a single forecast, which is exactly how a price can detach from the arithmetic that was supposed to anchor it. • A price is a balance of float against money: small floats amplify flows, and flows run on their own momentum. • IPO mechanics: the underwriters set the offer, the first print is the pop, and the selling shareholders do not receive it. • Lock-up expiries add tradable supply on a published schedule — a dated supply event visible in advance. • Index inclusion creates buyers who must buy regardless of price, which is mechanical demand rather than opinion. • The 2000 re-rating was financing and flow as much as narrative; the collapse was supply arriving as the flow stopped. None of this makes the flow story a substitute for the valuation arithmetic; it answers a different question. The multiple told you what the price assumed and why it could not hold. The float and the flow tell you why it could be reached at all, and when the mechanism that produced it would reverse.

Sources

• Valuation of the internet sector at the peak, and the implied growth ratesDamodaran, "Investment Valuation", internet-company chapters; "The Dark Side of Valuation"
• Multiple compression and long-horizon returnsStandard empirical evidence on the sources of long-run returns; Shiller, "Irrational Exuberance"
• Why high-multiple growth and low-multiple value convergeFama & French (2000); mean reversion in profitability and valuation
• The technology bubble’s financing structure and the companies that survivedSEC filings and contemporaneous market data, 1998–2002