Why do commodity price spikes usually end themselves?

There is an old trading-desk line: the cure for high prices is high prices. It sounds like folklore. It is actually the elasticity arithmetic run forward in time — and it is the structural reason commodities behave differently from equities over long horizons.

Three cures, on three clocks

A high price sets three processes in motion at once:

• Demand destruction (months). Elasticity grows with the horizon. Buyers who could not cut this week reschedule, re-route, insulate, run the plant at lower rates. The −0.05 of lesson 1 drifts toward −0.2.
• Substitution (months to years). Users switch to whatever is now relatively cheap — a different fuel in a power plant, a different metal in a heat exchanger, a different oil in a food product.
• Supply response (years). The high price makes marginal production economic: restarted mines, higher-cost fields, extra acres planted. Then, much later, brand-new capacity arrives.

Each cure is weak on day one and strong on year three. That is precisely why the spike is violent and why it tends not to last.

It runs downhill too

The symmetric statement is just as important and much less quoted: the cure for low prices is low prices. Below the marginal cost of production, high-cost producers lose money, cut capital spending, shut wells and mines, and stop planting. Supply falls, and the surplus closes from the other side. Both directions are self-correcting; neither is fast.

Why this makes commodities structurally different from equities

A share in a growing company has nothing forcing its price back down: earnings can compound for decades, and there is no physical mechanism that punishes a high share price. A barrel of oil has exactly such a mechanism. Its price is tethered — loosely, slowly, but genuinely — to what it costs to produce the marginal unit.

This is why research on very long samples finds that real commodity prices have wandered around a slow-moving level rather than compounding upward the way equity indices have. A commodity has no earnings to grow. It has a price, and a cost of production that price gravitates back toward.

A worked example

A market prices at $50 in a balanced year; the marginal producer's all-in cost is about $45. A supply disruption removes 3%, and with a short-run elasticity of −0.05 the price runs to $80.

• Year 1: high-cost supply that was uneconomic at $50 restarts. Consumers begin trimming.
• Year 2: demand is down 2% versus trend; the shortfall is covered.
• Year 3: projects sanctioned at $80 begin delivering into a market that no longer needs them, and the price undershoots to $40 — below the $45 cost that was supposed to be the floor.

Notice the ending. Self-correction is not a gentle return to fair value; it is an overshoot in both directions. Understanding why the overshoot happens is the subject of the entire next unit.

The floor is two numbers, not one. The cash cost is what an existing mine, well or field spends to keep producing; only below it does running supply shut in, and even then owners run at a loss for a long time to avoid closure and restart costs. The incentive price is what a new project needs over its whole life to be sanctioned, and it sits far above cash cost because it has to repay the capital. Between the two the market can sit for years: no supply leaves and none is being built. The $45 in the example is an all-in figure, closer to the second number than the first, so the fall to $40 stops the next project long before it shuts an existing one.

The honest limits

Two caveats, stated plainly. "Eventually" can mean a decade — a mechanism that works over ten years is no guide to next quarter. And the level it reverts toward moves: the marginal cost of production changes with technology, energy costs, ore grades and regulation, so there is no fixed number to revert to. This is a lens for understanding why spikes fade, never a claim about when.

In the data

The long physical price histories do not come at the same frequency. Crude's is published daily as well as monthly; the global copper and corn prices exist only as one number a month, stamped on the first day. The newest month of each of the two is below.

So an energy-versus-metal comparison over the full history has to be run monthly on both legs: fine for seeing the shape, spike then slow decay, and useless for counting days that moved more than 5%.

Try it now

1. The full history this page holds for one energy commodity and one industrial metal is below. Switch both to Monthly bars — at this length daily bars are noise — and look for the shape the lesson describes: a spike that goes up in months and comes down over years.

2. Measure one spike on each chart twice: once from the base up to the peak, once from the peak back down to wherever it settled. Compare the two bar counts. Up-fast and down-slow is the whole asymmetry, and it is a number, not an impression. Then say in one sentence why an equity index does not have this shape: one wanders around a level it keeps returning to, the other compounds away from where it started.
3. Pick any past spike on either chart and ask which of the three cures you can see arriving first.