Planted at Yesterday’s Price

IThe hogs found it first

In the late 1920s a young agricultural economist named Arthur Hanau, working at the Institut für Konjunkturforschung in Berlin, studied German hog prices and found a swing that nobody was steering. Pork prices rose and fell on a steady cycle of three to four years. No weather pattern explained it. No cartel ran it. The cause was time itself. A farmer decides to breed hogs when the price is high — but the hogs arrive at market a year or more later, and every other farmer saw the same high price and made the same sensible decision. The hogs arrive together. Glut. The price collapses. Farmers quit together. Shortage. The price spikes, and the wheel turns again. The Germans named it the Schweinezyklus — the hog cycle.

Nicholas Kaldor gave the mechanism its name in 1934, from the picture it draws. Plot supply against demand, then trace what happens when each season’s supply answers the previous season’s price: the path spirals around the crossing point like a spider’s web. Mordecai Ezekiel’s 1938 paper, “The Cobweb Theorem,” made it canon.

Nobody in the cycle is foolish and nobody is lying. The swing is not an error in the people. It is a law of the structure.

That is the part worth carrying from hogs to humans. Every farmer acts reasonably on the best number available. The oscillation comes from the shape of the situation: wherever supply must be decided long before it is delivered, and the deciders expect today’s price to hold, the market must swing. The length of the lag sets the period. Hogs take a year or so to raise, and pork swings in three or four. Cattle take far longer, and the cattle cycle runs about a decade — the longest in agriculture. A profession takes four to six years to grow. Careers swing by the generation.

Figure 01

The web: each step answers the price before it

IIThe human cobweb

You have lived inside this diagram. The “learn to code” decade is the textbook case: record numbers of students chose computer science in the early 2020s, acting sensibly on the prices of the late 2010s, and those cohorts arrived in the mid-2020s into the thinnest junior software market in memory. Not because the advice was dishonest — because a million families took it at once, with a four-year delivery lag. The same web spun law school after 2008, and nursing, and petroleum engineering before that. Different fields, one diagram.

Why did it never damp? Because expectation is brought, not chosen. The forecast inside each of those decisions was trained on the only record available — headlines, cousins, counselors — and all of them are archives of old prices. Look at a fast world through a slow window and you do not see a blur; you see a coherent world that is wrong. A seventeen-year-old deciding once, with folklore for a forecaster, cannot do otherwise. The structure never gave anyone a better instrument.

IIIThe cure that stayed in the journals

Economists wrote the prescription in 1961. John Muth’s “Rational Expectations and the Theory of Price Movements” — the paper that introduced rational expectations — was built on exactly these markets: an isolated commodity with a fixed production lag. His point was simple and devastating. If producers forecast the harvest price with the model itself, instead of extrapolating the planting price, the cobweb dies. Theory took the cure and rebuilt macroeconomics around it. Kitchen tables never got it, because the cure requires a forecasting machine, and for the entire century between Hanau and now, the only forecasting machinery a family owned was the folklore that caused the cycle.

That is what just changed. Two inputs of the law are moving at once. The lag is collapsing: when a capability can be rebuilt in months — when the stack carries the first eighty percent on day one — the delivery delay that set the period shrinks from half a decade toward a season, and a wrong turn becomes a correction instead of a sentence. And the forecast has arrived: for the first time, the person at the kitchen table can ask a machine that holds the model, and the machine can answer with the price at the arrival date, not the price on the sign. By the time you get there, this lane is full. This adjacent lane is short-handed, and nobody is walking toward it.

The non-obvious part is what happens next. The cobweb does not simply die. It moves. It drains out of everything whose lag is made of learning and pools in everything whose lag is made of matter. Skills damp; cattle still take a decade. Houses, power plants, ships, mines, transmission lines — supply there is still committed years before delivery, by committees reading today’s price. The oscillation retreats into the physical, and the famous “cycle” becomes, increasingly, a construction phenomenon.

And the cure carries a new instability of its own. When every farmer consults the same oracle, the oracle’s answer becomes the new price on the sign. If a million households are told on the same morning that welding is short-handed, and a million households answer at once, the shortage is filled and overshot by the very forecast that named it — the web spinning around the forecast instead of the price. The only honest counselor in that world is one that subtracts its own crowd: that answers conditioned on everyone else hearing the same answer.

IVWhat I expect

1. The early-2020s cohorts are the last full-amplitude swing

   The students who chose at the 2021 price and arrived at the 2026 price rode the oldest human cobweb end to end. After them, corrections come mid-flight — in months, not degrees — and “I trained four years for a market that left” retires as a sentence. If the class of 2033 says it, I was wrong.

2. “Hot field” advice dies as a genre

   Advice that names a destination is a planting-price snapshot, and it will be visibly, repeatedly wrong. Advice shifts to structures that hold at any price — positions short on lag, long on adjacency. Within a few years, telling a teenager to “go into” a named field will sound the way stock tips at a barbecue already do.

3. Equilibrium-aware guidance becomes the product

   The first advisers — human or synthetic — that condition their answers on their own audience will measurably outperform the sign-readers, and “how many people are being told this?” becomes a question ordinary families know to ask of any forecast handed to them.

4. The cycle retreats into matter

   Gluts and shortages increasingly trace to things that must be bred, poured, permitted, or built — cattle, housing, generation capacity, ships — while cognitive-supply whiplash fades. When you hear “cycle” in the 2030s, it will mostly mean concrete.

5. The first oracle glut

   Within a few years, a visible shortage-turned-glut will be traced in the postmortem not to stale prices but to a shared forecast: everyone was told the same scarce thing, and everyone answered at once. It will be reported as an AI failure. It will actually be the oldest law in agricultural economics, running at machine speed on a new signal.

VMy own planting

I spent part of my career inside this diagram. E2open, the company I co-founded, built supply-chain software — a business that exists because a demand signal distorts as it climbs a chain of lagged decisions. The trade calls that the bullwhip effect; it is the hog cycle wearing a badge. And when the twenty-four-year-olds I advise ask me what field to enter, I have stopped answering with a destination, because a destination is a price on a sign. What I tell them instead — watch which hours refuse to compress, let your first employer be your first customer, keep your needs low enough that you can turn — is lag advice, not price advice. It is a position that survives any harvest. The cheapest insurance against a market that swings is the ability to answer the new price before the crowd that planted beside you even knows it changed.

VIThe claim ledger

Already true

The mechanism is century-old record, not argument: Hanau’s hogs, Kaldor’s name, Ezekiel’s theorem, Muth’s cure. And the first human-scale swing has already landed — record computer-science cohorts, chosen at the 2021 price, arrived into the weakest junior market in memory. That part is history, not forecast.

What has to happen

Harvest-time forecasting has to actually be consulted at decision time by ordinary families, not merely be available to them. And retraining lag has to keep collapsing, so that mid-flight correction is real rather than theoretical. If the sign stays more trusted than the forecast, the cycle keeps its food.

Where I am probably wrong

Shared forecasts may damp rather than swing — if the oracles are built equilibrium-aware from day one, my fifth prediction never happens and I have underestimated the engineers. Matter may not hold the cycle as long as I think; robots pour concrete too. And the deepest risk: people may simply prefer the sign. Folklore has survived every almanac ever printed.

VIIThe sign in the window

It is a Tuesday night in 2031, and the seventeen-year-old at the kitchen table is yours. The sign in the window still shows a price, and it is still yesterday’s — that part never changes; a price is a record the moment it is printed. For a hundred years the law said you had to plant by it anyway, because the lag was long and the forecast was folklore, and so whole generations delivered themselves into markets that had already turned.

She won’t. Not because the world got kinder, but because the two numbers that powered the wheel — the length of the lag and the age of the forecast — are both falling toward zero at the same time, for the first time since the hogs.

The web is coming down. Watch where the spider moves.