Investing Classics: Energy & CyclicalsEnergy, Commodities, and the Logic of Cycles

The most counterintuitive feature of cyclicals: the valuation signal is inverted. Peter Lynch warns in One Up on Wall Street that when a cyclical's P/E is very low and earnings are at a record, it is often the tail end of the boom — because peak earnings make the P/E look "cheap" just as earnings are about to revert to the mean. This is especially true of oil: when prices are high, shale and deepwater projects pile in, and two or three years later the supply gushes out and prices collapse. The correct anchor isn't the current P/E, but where the commodity price sits relative to the long-run marginal cost of production.

In June 2014 WTI crude was about $107/barrel; the US shale revolution surged output, OPEC refused to cut, and by February 2016 it had fallen to about $26 (−76%). More extreme still was April 20, 2020: lockdowns evaporated demand and storage ran out, and the WTI May contract closed at a negative price for the first time in history — −$37.63/barrel, meaning longs had to pay to take delivery. Both crashes proved the same point: however "cheap" oil looks, it can get cheaper.

Cycles can be broken by structural change — if demand is permanently displaced, low prices need not rebound, and a "cycle" becomes a "secular decline." Misuses: ① running a DCF on peak earnings as if they were normal; ② buying a low P/E to bottom-fish a cycle that's actually topping; ③ ignoring the balance sheet — a company that can't survive the trough has no "rebound."

• Relative to long-run marginal cost of production, is the commodity price high or low?
• Am I being misled by a "low P/E" while earnings sit at the cycle peak?
• Can this company's balance sheet survive one full price trough?
• Is demand cyclically dipping, or structurally and permanently shrinking?

Marathon Asset's "capital cycle" framework upends common sense — most people stare at demand forecasts, while the real money comes from the supply side. The loop: high prices → high profits → capital pours in → glut → price crash → losses and bankruptcies → capital flees, capacity is shaken out → supply shrinks → prices recover. The investor's job is to run against the capital cycle: build positions when capex has dried up and everyone has left; step aside amid expansion mania. The key metric is industry-wide capex and new capacity, not a single quarter's demand.

2003–2011, the "China super-cycle": urbanization devoured the world's metals, and iron ore climbed from about $30/tonne to roughly $190 by early 2011, as the big three miners and countless new projects expanded furiously on the logic that "China demand never stops." But supply finally caught up — by end-2015 iron ore had fallen back to about $40 (−79%). Capital that expanded at the top was almost wiped out; those who entered after the 2015–2016 shakeout caught the subsequent rebound.

The capital cycle demands patience and can't be timed precisely — a shakeout may drag on for years, and being early is agonizing. It also assumes the industry is ultimately rational, but subsidies, state-owned firms, and strategic-industry status can keep capacity that "should die" alive, distorting the whole cycle.

• Is this industry expanding furiously right now, or cutting capex and shutting capacity?
• Am I following capital in, or positioning against it as capital flees?
• Are there subsidies or policies that keep inefficient capacity from exiting, distorting the normal cycle?

The energy transition is almost certainly one of the largest trends of the coming decades — but "the trend is right" and "the investment is right" are two different things. Buffett uses autos and airlines to make the point: of the roughly 2,000 American carmakers of the early 20th century, nearly all vanished; aviation changed the world yet cumulatively destroyed enormous shareholder value. The reason is that these industries grow fast but sell commodities, lack pricing power, and have bottomless capex. Solar modules are the textbook case: technical progress crashed costs, capacity became massively oversupplied, and manufacturing degenerated into a thin-margin "commodity," with the trend's dividend captured by consumers and latecomers rather than early shareholders.

Amid the 2010s solar boom, SunEdison expanded via aggressive debt-funded acquisitions and briefly became the world's largest renewable-energy developer; in April 2016 it filed for bankruptcy under roughly $16 billion of debt — one of the largest US bankruptcies of that year. Countless early solar manufacturers fell amid overcapacity and price wars. Solar generation grew dozens-fold over the decade, yet most early investors in manufacturing got nothing — the growth was real, the returns were not.

This isn't to say transition industries are uninvestable — rather, find the links in the chain with pricing power and cost barriers (specialized equipment, key materials, concession-type operating assets), and avoid commoditized manufacturing. Misjudgment usually comes when "the story is too compelling" overrides a cool look at unit economics.

• Does this business's growth come with real pricing power and returns on capital (ROIC)?
• Is the product differentiated, or will it become a commodity in a price war?
• Is its capex "widening the moat," or burning cash because competition forces it to?
• Will the trend's dividend land with shareholders, or with consumers and latecomers?

Buffett divides investments into three kinds: currency-denominated assets (cash, bonds), assets that produce nothing (gold, commodities), and productive assets (businesses, farms, real estate). The entire return of the second comes from "someone paying more later" — that is speculation, not investment. An ounce of gold a century later is still that ounce; a farm or a great company keeps producing cash. This doesn't mean commodities should never be held — in inflation or as a crisis hedge they have tactical value, but you must be clear-eyed: you're making money on price, not on compounding, and there is no intrinsic-value anchor to tell you whether it's "dear or cheap."

In the 2011 letter Buffett did the arithmetic: the world's roughly 170,000 tonnes of mined gold would form a cube about 21 meters per side, worth about $9.6 trillion. For the same money you could buy all US cropland (producing about $200 billion a year) plus about 16 ExxonMobils, and still have about $1 trillion left over. A century on, the cropland still grows grain and the oil firms still pay dividends, while the gold cube "produces nothing at all."

Buffett's critique has its limits — Berkshire itself bought the gold miner Barrick Gold in 2020 (then sold it). In specific macro regimes (high inflation, negative real rates) commodities do have a phase-by-phase role, and a pure "productive-asset" framework is not all-powerful.

• Am I buying this asset because it produces cash, or because I'm betting the price rises?
• With no cash-flow anchor, on what basis do I judge it "dear" or "cheap"?
• Is this a core holding (compounding) or a tactical hedge (timing)? Have I kept them straight?