Extend a moat from four years to fifteen and its value roughly triples

What the terminal value takes on faith

A discounted cash flow splits a company's worth into two parts: the cash flows an analyst forecasts year by year, and a single terminal value standing in for everything after the forecast ends. The course's discounted-cash-flow lesson showed that this terminal value usually holds most of the total. It rests on an assumption almost no model states out loud, that the company keeps earning more on new investment than the investment costs. The value-of-growth lesson showed why the assumption matters: growth adds value only while the return on invested capital beats the cost of capital. Terminal value quietly assumes that gap stays open for good.

Capitalism does not leave such gaps open. A business earning far more on its capital than the capital costs is advertising an opportunity to everyone who might take it. Warren Buffett put the pressure plainly in his 2007 letter to Berkshire shareholders: “The dynamics of capitalism guarantee that competitors will repeatedly assault any business ‘castle’ that is earning high returns.”¹ Left alone, the assault pulls a high return back down toward the cost of capital, and the gap the terminal value counted on closes.

Michael Mauboussin frames value creation as two separate things: the size of the spread between the return on invested capital and the cost of capital, and how long a company can hold that spread open.² The spread is visible in this year's filing. The single-year habit of setting this year's return beside this year's cost of capital invites an analyst to check the spread once and stop looking. The number that decides most of the value never appears in a single year's accounts.

Some moats fade in a year, others in a generation

Returns on invested capital drift back toward a common level. Across US public companies from 1963 to 2004, McKinsey found the median return on invested capital, excluding goodwill, averaged close to 10 percent and stayed relatively constant over four decades.³ That is the gravity a high-return business is pulled toward. Mauboussin describes the same pull as a life cycle, in which “high returns attract competition, generally causing economic returns to move toward the cost of capital.”²

Gravity is not a timetable. Its speed varies so widely that the average buries the point. Mauboussin measures the speed with the year-to-year correlation of returns inside an industry. A correlation near 1.0 means this year's leaders are next year's leaders. A lower number means the ranking reshuffles fast.

A packaged-food maker's returns barely move from year to year; an apparel maker's erode faster, and a technology-hardware business faster still. Two companies can post the same return this year and sit on completely different clocks, because the industry each occupies fades at its own rate.

Some of even that persistence is luck. Mauboussin's own reading of the data is that sustained value creation from skill alone is rare, and that some companies still on top after a decade are there by chance.² The fade is real and its direction reliable; its speed for any one company is a matter of judgment.

When only the fade changes

Put two real businesses on those different clocks. Copart runs online salvage auctions, the marketplace where wrecked cars, mostly from insurers, are sold to about a million registered buyers.⁴ Crocs sells molded foam footwear.⁵ Both earn strikingly high returns on the capital they employ: Copart around 30 percent, Crocs roughly 27 percent in fiscal 2024, before a large write-down the following year.⁴⁵ On this year's numbers, the two read as equally good businesses.

They are not the same business or in the same industry, so a fair comparison strips out everything except the one thing in question. Hold the return equal at 30 percent for both. Hold their growth equal. Change only how long each keeps earning the excess return.

Start with the excess return in dollars, the profit a business earns above what its capital costs: its invested capital times the spread between its return and its cost of capital. A firm with $1,000 of invested capital earning 30 percent against a 10 percent cost of capital earns $300, of which $100 merely pays for the capital, leaving $200 of excess profit this year. The value it adds is the present value of that stream for as long as the advantage lasts, growing while it does:

Every input but one is fixed. The first year's excess profit is $200, it grows 6 percent a year, and the cost of capital is 10 percent. The advantage period, N, the years the excess return survives, is the only thing that moves.

The business whose advantage lasts fifteen years is worth a little over three times the one whose advantage lasts four, on identical returns, growth, and capital. Nothing in this year's income statement separates them. The whole difference is the number in the first column, and that number is a forecast about the future, not a fact from the past.

Nobody observes an advantage period. The 4 and the 15 are judgments, defended well or badly by what is known about each business. On almost any high-return stock, the argument between a bull and a bear comes down to that column and little else. It is the same forecast that sits, unlabeled, inside a discounted cash flow's terminal value. Assume the spread lasts and the terminal value is large; assume it fades and most of that value goes away.

Which advantages actually last?

Everything rides on the advantage period, which turns the abstract question into a practical one: what actually holds a return open, and can it be seen in advance. Michael Porter's account of barriers to entry names the sources, read from the incumbent's side.⁶ Economies of scale force a rival to enter big or accept higher costs. Demand-side benefits of scale make a product worth more to each buyer as more buyers join. Switching costs make leaving a supplier expensive. Porter's own example of the demand-side kind is an online auction, where bidders go where the other bidders already are.

Copart is that example in the metal. Its virtual auction, the company says, “increases the pool of available buyers for each sale, which brings added competition and an increase in the amount that buyers are willing to pay for vehicles.”⁴ More buyers raise prices, higher prices attract the insurers who supply the cars, and more cars draw more buyers. The advantage also stands on physical ground: salvage yards need local zoning approval, which the company says makes new sites “more challenging and expensive” to develop.⁴ About 70 percent of the US vehicles Copart sells go to buyers in another state or country.⁴ A new entrant would need the yards, the permits, and both sides of the marketplace at once.

Crocs has no comparable wall. Its returns rest on a foam clog staying in fashion, and taste is not a barrier to entry. Operating income reached $237.8 million at the 2007 clog-fad peak, then collapsed to a $188.3 million loss in 2008 and another loss in 2009.⁷ The excess return did not erode over a decade. It was gone in two years. The company later tried to buy durability, paying $2.3 billion for the HEYDUDE brand in 2022.⁸ By 2025 it had cut its own forecast for HEYDUDE and written the brand down by $737 million.⁵ A purchased moat can be marked down within three years.

If high current returns settled the question, Crocs at its 2007 peak would have looked like a fortress. Buffett, who made the moat famous, misjudged one himself: Dexter Shoe, bought in 1993 for $433 million of Berkshire stock, which he had assessed as durable.

His test for telling a real moat from an illusory one turns on durability, not on the current return: “A moat that must be continuously rebuilt will eventually be no moat at all.”¹ A brand defended only by this season's taste is rebuilt every season. A marketplace that grows stronger as it grows larger is not. The current return cannot tell the two apart. Only the source of the advantage, and whether anything renews it for free, can.