The standard advice is to invest where demand is exploding: find the industry with the biggest growth runway and ride it. Edward Chancellor, who edited two decades of Marathon Asset Management’s investment letters into the book Capital Returns, argues the opposite is closer to the truth. Returns in an industry are driven less by demand than by the supply side— how much capacity exists and how fast it is being added. And the cruel part is that high returns contain the seed of their own destruction. A booming industry is, by construction, an industry about to be flooded with competitors.
The supply side is the variable nobody watches
Here is the mechanism, named plainly. When an industry earns unusually high returns on capital, that profit is a signal flare. It attracts capital: new entrants appear, incumbents expand to defend their share, and bankers happily raise the money to build it all. Capacity rises. Eventually supply outruns demand, pricing power evaporates, and returns fall back toward — or below — the cost of capital. Then capital flees, weak players merge or die, capacity is withdrawn, and the survivors enjoy fat returns again. The whole loop turns over a span of years.
The insight is that supply is observable and demand is not. You can count capex announcements, plant openings, and merger activity in real time. Demand forecasts, by contrast, are guesses dressed as analysis, and they are systematically too rosy at exactly the wrong moment. That is the edge Chancellor describes: watch where capital is gushing in, and avoid it; watch where capital is fleeing, and look there. The observation isn’t new. Graham and Dodd noted in 1934 that a high return on capital draws in competition. The capital cycle just turns that one sentence into a system.
How an industry eats itself
A toy version first. Imagine a town with one profitable bakery. The owner earns a fat margin, so a neighbor opens a second bakery, then a third, each watching the first one’s success. Soon the town has five bakeries fighting over the same customers, all cutting prices, all earning nothing. A few close. The survivors raise prices again. No single decision was foolish; collectively they were ruinous. Warren Buffett described this in his 1985 Berkshire letter with the image of a crowd at a parade: each person stands on tiptoe to see better, and once everyone does, nobody sees better and everyone’s legs ache.
The real case is the telecom and fiber-optic boom of the late 1990s. Convinced that internet traffic would double every few months, operators raised enormous sums and laid fiber-optic cable across continents. Capacity expanded many times over. When the traffic forecasts proved wildly optimistic, the result was a glut so severe that, by 2002, only a small fraction of the installed fiber was actually carrying traffic — the rest sat unused as “dark fiber.” Wholesale bandwidth prices collapsed. Global Crossing, which had built a network spanning more than 100,000 miles of cable, filed for bankruptcy in January 2002 with over $22 billion in assets — one of the largest US bankruptcies up to that point. The demand was real; the industry still drowned, because the supply side overshot.
Why next quarter’s earnings tell you nothing
This is why the cycle is nearly invisible to the way most investors look. A typical analyst is trained on the next quarter’s earnings per share. But the capital cycle plays out over five to ten years. Chancellor points to the lag: capital pours into an industry, factories take years to build, the goods take more time to reach the market, and only then do the depressed financial results show up in the numbers. By the time the earnings turn ugly, the capacity has already been built and the damage is locked in.
So the variables that matter are not on the earnings headline. They are capex versus depreciation (is the industry investing far more than it is wearing out?), aggregate capacity additions, and consolidation (are competitors merging and exiting, or multiplying?). Above all, watch how management behaves: does it expand hardest when returns are high and valuations are stretched, or does it harvest cash then and invest when everyone else has fled? The counter-cyclical operator is rare, because pay packages and peer pressure reward chasing growth at the top. They are also the ones worth owning.
Where the cycle shows up in the value rank
You don’t have to track capex spreadsheets by hand to see the cycle’s shadow. It shows up in the value rank. An out-of-favor, capital-starved industry tends to screen cheap on peer-relative value — price against sales, earnings, book and dividends — precisely when the supply side is about to tighten and returns are about to recover. A capital-flooded darling screens expensive at the very top of its cycle, when the building is at its peak. The same dynamic erodes competitive position over time, which is why a falling moat is just the company-level face of the same supply-side pressure; how to read a moat trend tracks it from the other side.
One honest limitation: timing this is genuinely hard. An industry can stay over-supplied far longer than you can comfortably wait, and “capital is leaving” is necessary but not sufficient. A starved industry only rewards you if there are survivors with the balance-sheet strength to outlast the shakeout, which is why cheapness on the value rank has to be paired with the safety rank before you act. The principle is simple even if the patience isn’t: distrust the industry everyone is funding, and look hard at the one everyone is abandoning. The concrete action is to add capex-to-depreciation and consolidation to your checklist, and to read a cheap value rank in a hated sector as a question worth investigating, not a trap to avoid.