What makes something a commodity: The economics the families share

3.1 Storage, perishability and seasons decide how far stocks can smooth prices

Radetzki and Wårell note that refrigeration and preservation have made every commodity storable to some degree, at a cost. Metals, grains, rice and jute keep for years; meat, milk and fruit do not. Storability matters twice over. Drawing down stocks lets supply vary by more than production can, raising the short-run elasticity of supply. And storage invites speculation, since holding a good is a bet on its future price. It also ties spot to futures prices, which is why exchanges need it; Geman notes that electricity, barely storable, loses the usual link between the physical good and its forward contracts.

Seasonal production adds a calendar. Geman describes the US wheat harvest moving from Texas in June to Minnesota in September. Storage holds roughly half the crop, so the rest is sold at once and prices sag every summer, then firm through the autumn until southern hemisphere wheat arrives around December.

For storable farm goods one ratio sums up the balance: ending stocks divided by the year's use. Geman puts the typical range at 20 to 40 per cent. In a tight market any supply shock sends prices very high, and the carry-over of stock from one crop year to the next is what links futures prices across years.

Perishability shapes trade too. Refrigerated ships from the 1880s first let meat and fruit cross oceans, after steam had opened the grain trade: in Britain the share of bread grain grown at home fell from two thirds in the 1850s to 20 per cent by the 1880s. Perishable dairy now trades largely through the Global Dairy Trade auction platform.

• Radetzki & Wårell (2021), A Handbook of Primary Commodities in the Global Economy · pdf pp. 24-25, 42, 118, 147
• Geman (2005), Commodities and Commodity Derivatives · pdf pp. 49, 154-155, 162
• USDA (2026), World Agricultural Supply and Demand Estimates, October 2026 web ↗ · World Wheat Supply and Use, 2025/26 Est.

3.2 How fast supply and demand can respond

Supply lags differ by orders of magnitude. Wheat, sugar and bananas can expand by the next season if farmers plant more land. Geman says new cocoa trees need about seven years to bear and coffee bushes three to five. Radetzki and Wårell put the lag from exploration to mine production at five to ten years or more, and Erten and Ocampo cite up to twenty years from first investment to revenue. Iron ore demand took off around 2003 with China, yet supply did not rise substantially until 2014.

Radetzki and Wårell argue that wheat and copper may have similar long-run supply elasticities, but copper's short-run elasticity is far lower. Supply cuts therefore pay better for mineral and tree-crop producers, whose rivals cannot respond quickly, than for growers of annual crops.

Secondary supply loosens the constraint. Precious metals, steel, copper, lead, rubber and wool gain supply from recycling, which primary producers do not control, so it raises overall supply elasticity and weakens cartels. Tilton and Guzmán report that old scrap alone has met roughly 35, 10 and 70 per cent of US aluminium, copper and lead consumption. New scrap from manufacturing is cheap to recycle and almost all of it already is, so it adds little at the margin.

On the demand side the question is substitutes. Bananas and lamb have close ones, so demand shifts quickly when their prices rise. Platinum, chromium and rare earths have few in key uses, so demand barely responds; coffee is inelastic through habit. Long-run demand is always more elastic than short-run, because users redesign around a material that has become too dear.

• Geman (2005), Commodities and Commodity Derivatives · pdf pp. 164-165
• Radetzki & Wårell (2021), A Handbook of Primary Commodities in the Global Economy · pdf pp. 42-44, 203
• Erten & Ocampo (2012), Super-cycles of commodity prices, UN DESA WP 110 web ↗ · pdf p. 3
• Tilton & Guzmán (2016), Mineral Economics and Policy · pdf pp. 78-79

3.3 By-products: supply that ignores its own price

Tilton and Guzmán define the terms by what drives the mine. For a main product, its price alone decides how much ore comes out of the ground. Co-products share that role, as lead and zinc often do. A by-product matters so little to the mine that its price leaves ore output unchanged.

Two consequences follow. First, by-product supply has a ceiling set by the host. Molybdenum recovered from copper ore cannot exceed the molybdenum in the ore actually mined, so near that limit the supply curve turns vertical, even in the long run. A higher molybdenum price does not raise copper output; if it did, molybdenum would not be a by-product. The ceiling moves only when the host market moves. Second, only costs specific to recovering the by-product count. Joint costs of mining and milling are carried by the host, so by-product supply is often cheap until it hits the ceiling, and is recovered only where the price covers the specific costs.

The result is a supply curve that is flat and then a wall. Nassar, Graedel and Harper found that 38 of 62 metals obtain more than half their primary output as companions; that percentage is their measure, companionality. Indium sits in zinc ore at a few parts per million, far too little to mine for itself, and about 71 per cent of silver comes from ores of zinc, lead, copper and gold.

Host booms can depress by-product prices. Radetzki and Wårell note that silver tends to be oversupplied when base metals boom, and that palladium depends on platinum mining in South Africa and nickel mining in Russia. Status can shift with price: Nassar and colleagues found rhodium almost self-supporting in 2008, when its price was very high.

• Tilton & Guzmán (2016), Mineral Economics and Policy · pdf pp. 66-67, 75-77
• Nassar, Graedel & Harper (2015), Science Advances web ↗ · pdf pp. 1-6
• Radetzki & Wårell (2021), A Handbook of Primary Commodities in the Global Economy · pdf pp. 105-106

3.4 Value per kilogram decides whether a price is global, regional or local

Radetzki and Wårell give rough values per kilogram at 2018 prices: iron ore 0.07 dollars, steam coal 0.11, wheat 0.21, crude oil 0.50, standard steel 0.59 and copper 7.00, against 15 for a car and 1,000 for a laptop. Freight acts like a tariff. Where it is a large share of the delivered price little trade happens; as it falls, trade grows and regional prices converge.

Transport costs fell in two spurts. Steam cut the cost of shipping cotton and wheat from New York to Liverpool from 55.1 dollars a ton in 1825 to 3.5 dollars in 1910, at 1910-14 prices. The second spurt followed the Suez crisis of the mid-1950s, when shippers turned to huge bulk carriers. Freight for Brazilian iron ore to Europe fell from 24 dollars a ton in 1960 to 7 in the early 1990s, and the freight advantage of Swedish ore shipped through Narvik shrank from 16 dollars to 3. Long-distance seaborne iron ore rose from 23 per cent of world output in 1960 to 36 per cent in 1990; by 2017 two thirds was exported.

Gas is the exception among fuels. Pipelines reach only so far and liquefaction is costly, so three regional markets persisted; until the mid-1990s East Asian prices were twice US prices, and after about 2008 US shale gas pulled the regions apart again.

At the bottom of the value scale markets are local, like the many crushed stone markets of the United States. Only 7 per cent of world timber and 10 per cent of rice output was exported in 2018, against 72 per cent of cocoa, 78 per cent of coffee and 88 per cent of natural rubber. Rice is a reminder that taste, not only freight, can keep a crop at home.

• Radetzki & Wårell (2021), A Handbook of Primary Commodities in the Global Economy · pdf pp. 21, 24-27, 42, 48-49
• Tilton & Guzmán (2016), Mineral Economics and Policy · pdf p. 31

See it in the table

Check your understanding

A by-product metal's price doubles. What usually happens to its supply in the long run?

Why does storability matter for a futures market?

Which commodity in Radetzki and Wårell's 2018 figures had the smallest share of world output exported?