The Commodity Cycle: Why Supply Responds Slowly

A commodity cycle is what a market does when its two sides answer at different speeds. Buyers can cut what they use inside one period. A new mine or field takes about four periods to find, fund, build and connect. So a price moves at once, the answer to it arrives four periods later, and by then the demand that called for it has already gone.

A commodity cycle rests on two things already established. The first is how a price forms when what people want and what is available do not match: the price moves until the two agree. The second is that a cycle is a repeating shape rather than a single move, so it has stages that come round again. One ingredient joins those two, and only one, and the ingredient is time. Once each side of the market has its own answering speed, with everything else exactly as it was, a market that would otherwise settle in one step starts to swing instead.

Every figure below belongs to the Republic of Sankhya, an invented country. Sankhya burns 10,00,000 barrels a day of crude at a base price of Rs 4,000/- a barrel, and both figures stay fixed throughout.

Why can buyers answer in one period when new supply takes four?

Start with the side that can move. When crude gets dearer, a Sankhya household combines two errands into one trip, a factory reschedules a shift, a haulier drops a run. None of that needs a building, a licence or a bank. A cut needs a decision, and the decision takes effect almost immediately. Cutting consumption costs nothing but the inconvenience of using less, so the demand side of a commodity market can answer a price inside a single period.

Now the other side. A producer who wants to sell more crude cannot simply decide to. The rock has to be found, and finding it takes surveys. The project has to be costed and financed, and a board and a lender both have to say yes. The project has to clear whatever consents apply. Then the field has to be built and connected to a pipe or a port that can carry the barrels away. That whole run is set at four periods from the moment a project is sanctionedThe point where a project stops being a study and becomes a spending commitment: the approval is given, the contracts go out and money starts leaving the account. to the moment a barrel from it reaches a buyer. Every step in that four period run is a physical or a legal one, so nothing in the run is a failure of intent and no amount of wanting shortens it.

The two answering speeds sit side by side, and everything else falls out of them. One period on the demand side. Four periods on the supply side. The two sides of this market answer the same question at speeds that differ by a factor of four. The shape is that of a shower with a very long pipe. The tap turns and nothing changes, so it is turned further, and further again, and then all of the hot water arrives at once. Nobody was careless. The pipe was long, and the pipe is the whole story.

Why does the price overshoot instead of settling at the new balance?

Follow the sequence one step at a time rather than taking the conclusion. Sankhya opens new power stations and puts more freight on the roads, so national use steps up by 1,60,000 barrels a day. Production that period is still 10,00,000 barrels a day, so the market is short by 1,60,000 barrels a day and the price lifts to Rs 4,800/- a barrel. In period 2 buyers do what buyers can do and cut 80,000 barrels a day. The cut helps, but supply has not moved at all yet, so it closes only half the gap. So the price lifts again. The price keeps lifting through periods 3 and 4 for exactly one reason: the only side that can answer has already given everything it has, and the other side has not arrived.

Then period 5 comes. The field sanctioned back in period 1 finishes its build, passes its commissioningThe stage where a finished facility is tested and brought into normal service. A plant that is built but not commissioned is producing nothing. checks and starts delivering 80,000 barrels a day. Those 80,000 barrels are exactly the right quantity for the shortage of period 1. The same 80,000 barrels are the wrong quantity for period 5, where buyers have already cut back to 10,10,000 barrels a day. Supply is now 10,80,000 barrels a day against use of 10,10,000, and the market is 70,000 barrels a day over. The barrels that arrive in period 5 were the correct answer to a question asked four periods ago, and the question has changed since it was asked.

Notice what has and has not been assumed. Nobody misjudged anything. Nobody was greedy, panicked or slow. Every party did the sensible thing with the information in front of them at the time. The overshoot appeared anyway, out of the arithmetic of two different answering speeds. The overshoot is produced by the lag and not by anybody's error, and a market with these two lags cannot sit still even when everyone in it is behaving well.

What does one full cycle look like, period by period?

Three rules generate every row of the eight period path below, and all three are simple enough to check on paper. Demand this period equals 10,00,000 barrels a day plus the step of 1,60,000, less 100 barrels a day for every Rs 1/- the price sat above Rs 4,000/- one period ago. Supply this period equals 10,00,000 barrels a day plus 100 barrels a day for every Rs 1/- the price sat above Rs 4,000/- four periods ago. The price moves from its own last value by Rs 1/- for every 200 barrels a day of gap, upward when demand is ahead and downward when supply is.

Before reading the table, work out where it has to end up. Both sides have the same responsiveness of 100 barrels a day per rupee. So at the price that finally balances the market, half the extra 1,60,000 barrels a day is met by new production and half is rationed away by the price. The balancing price is Rs 4,800/- a barrel and the balancing quantity is 10,80,000 barrels a day. Rs 4,800/- is the level this market is trying to reach, and the path spends every one of its eight periods somewhere other than there.

The supply column explains the shape, so read it first. Supply sits at 10,00,000 barrels a day for four straight periods and then starts climbing. The flat stretch is the four period build, drawn as a number. While supply is flat the price has only one direction it can take. By period 4 the price stands at Rs 5,500/- a barrel, 14.58 per cent above the Rs 4,800/- that balances the market. Then supply arrives, and it keeps arriving. Periods 5 to 8 deliver fields sanctioned against the prices of periods 1 to 4, and those were the four dearest periods on the path. By period 8 the price is Rs 4,293.75/- a barrel, 10.55 per cent below the balance, and every barrel causing the fall was ordered when the market was short.

Why does the cycle repeat instead of dying out?

Look at the last two rows of the published table again. In periods 7 and 8 the Sankhya price sits at Rs 4,487.50/- and Rs 4,293.75/- a barrel, both below the Rs 4,800/- that balances the market. Now ask what a producer sitting in period 8 does with a project that would deliver in period 12. The price in front of that producer is below the balance. The project gets deferred. So does the next one. Four periods later, in period 12, no new barrels arrive. The low price meanwhile stopped rationing national use, so use has been quietly recovering the whole time. The shortage that starts the next cycle is manufactured by the surplus that ended the last one, and nobody has to intend any part of that.

Feeding the end of one swing into the start of the next is what makes the pattern a cycle rather than a bump. A bump happens, gets absorbed and is over. Here the absorption itself sets up the next event. The signal producers read to decide whether to build is the same price the previous round of building pushed down. There is no villain and no plan. The pattern is a loop, and the loop has a length: roughly the build time plus the time buyers take to drift back.

The cleanest way to see that the lags are doing all of this is to remove the difference between them and watch the swing disappear. Set both sides to answer inside one period and run the same step up of 1,60,000 barrels a day. In period 1 nothing has answered yet, so period 1 still lifts the price to Rs 4,800/- a barrel. From period 2 onward, both sides have answered, demand is 10,80,000 barrels a day, supply is 10,80,000 barrels a day, the gap is nil, and the price sits at Rs 4,800/- for the rest of the run. With both sides answering inside the same period the price steps once to the balance and never moves again. The swing was the lag, then, and not anything about commodities.

The equal lag result needs one honest qualification. Making the two lags equal is not by itself enough; making them equal at one period is. If both sides take three periods, nothing answers the step at all until period 4, and the price runs unchecked for three periods before both answers land together, so the swing is wider rather than narrower. Both sides being able to answer inside the period is what kills the overshoot, not the two speeds merely matching each other.

What decides how commodity prices affect corporate margins?

Everything above has treated crude as a price. For most of the businesses that touch it, crude is not a price at all: it is an input, and the price of an input is a cost line before it is anything else. The cost line is the whole bridge between a commodity and a company. When the Sankhya price moves from Rs 4,000/- to Rs 4,800/- a barrel in period 1, a 20.00 per cent move, nothing has yet happened to any margin. The next step depends on two things about the business, and neither of them is about crude. The effect of a commodity move on an operating margin is set by how much of the cost base that commodity is, and by how much of the rise the business can put into its own selling price.

Take two invented Sankhya businesses, built identically so that only those two things differ. Each sells Rs 100,00,00,000/- a year. Each runs a total operating cost baseEverything a business spends to run its operations in a period, added up before interest and tax. Wages, fuel, materials, power, rent and the rest. of Rs 80,00,00,000/-. Each therefore earns Rs 20,00,00,000/- of operating profit on a 20.00 per cent operating marginOperating profit divided by revenue, expressed as a percentage. An operating margin is worked out before interest and tax, so it describes the trading business rather than how it is funded.. Chandori Ceramics fires kilns, so crude is 60 per cent of that cost base, and it sells into a market where the buyer can walk to the next kiln, so none of a cost rise goes into the price. Talamba Instruments uses crude for 20 per cent of its cost base and sells specified parts on long contracts, so half of a cost rise can be recovered.

Chandori loses 9.60 points of margin. Talamba loses 1.89 points. The same 20.00 per cent commodity move, the same starting margin, and a difference of more than five times in what reaches the bottom of the trading account. Now split that difference into its two causes. The two are separable, and they are not the same size. If Talamba had passed nothing on, its margin would have fallen 3.20 points, exactly one third of Chandori's 9.60. One third is the share effect on its own, and one third is just the ratio of 20 per cent to 60 per cent. Passing throughMoving a cost rise into the price charged. How much of it a business can move, and how quickly, is worked through in the notes on inflation and on who ends up bearing a cost. half of the rise takes 3.20 points down to 1.89.

The last step hides something worth pausing on. Half of 3.20 points is 1.60 points, but the actual fall is 1.89 points. The extra 0.29 points is not an error. Passing a cost on raises revenue to Rs 101,60,00,000/-, so the margin is now measured against a bigger denominator, and a bigger denominator shrinks any given profit as a percentage. Recovering a cost in the price protects the profit in rupees more than it protects the margin in percentage points, and a reader who forecasts the second from the first will be short every time.

What makes one commodity's cycle longer than another's?

Two things, and the first is the one already on the table: how long it takes to build new supply. A commodity whose new capacity takes six periods to arrive spends six periods with only one side answering, so the price climbs for longer before anything meets it. In the Sankhya run, stretching the build from four periods to six lifts the highest price on the path from Rs 5,500/- to Rs 5,575/- a barrel and pushes the turn from period 4 out to period 6. The build time is not a market fact. The build time is a fact about rock, steel, permits and pipelines, and it differs by commodity for physical reasons.

The second is whether the commodity can be kept. If it can be stored cheaply, then some of yesterday's production is available to answer today's demand, and the supply side stops being purely a function of what is currently coming out of the ground. Storage has an exact equivalent in this arithmetic: barrels produced earlier can arrive now, so a store shortens the effective supply lag. Running the same step up with an effective supply lag of two periods instead of four narrows the whole swing. The highest price on the path falls from Rs 5,500/- to Rs 5,200/- a barrel, the lowest rises from Rs 4,293.75/- to Rs 4,650/-, and the distance between the two collapses from Rs 1,206.25/- to Rs 550/-, or 45.60 per cent of the wider swing. Storage shortens a cycle by letting production from an earlier period answer demand in this one, so a commodity nobody can hold has a sharper cycle for that reason alone and for no reason to do with the commodity itself.

The point holds without any arithmetic. Grain can sit in a shed, so a good harvest cushions a bad one. Electricity, at scale, largely cannot sit anywhere. A hot afternoon has to be answered by whatever is running at that moment and by nothing else. The same logic sorts commodities into sharper and gentler cycles before anyone looks at a single price. A national store, and what it actually buys at what cost, is a separate question worked through on its own.

What does knowing the cycle not let anyone do?

The cycle does not say when. A good mental model turns into a bad expectation at exactly this point. Everything above rests on two numbers, four periods and one period, and neither is knowable to the period in any real market. A build runs late. A consent lands early. Buyers cut faster in a downturn and slower in a boom, and demand also moves for reasons that have nothing to do with the price at all: weather, a new road, a change in what people do for a living.

Watch what happens to the date when only the build time changes. With a two period build the Sankhya price turns in period 2 at Rs 5,200/- a barrel. With four periods it turns in period 4 at Rs 5,500/-. With six it turns in period 6 at Rs 5,575/-. Three paths, one shape, three different dates for the turn, and the only thing that moved was a number nobody can observe directly. The turn sits at the end of a lag whose length is estimated rather than published, so knowing the shape of a cycle is not knowing the date of its turn.

The explanation reaches as far as why a commodity price moves the way it does, and how that movement reaches a cost line and a margin. Whether any price is high or low, what any price does next, and whether anything is worth holding, buying or avoiding sit outside it. A shape is an explanation. A shape is not a schedule, and treating one as the other is how a good idea becomes an expensive one.

What sits just outside the commodity cycle?