Prices and cycles: Real prices and the long-run trend

4.1 A real price is a ratio, and the deflator decides the story

A nominal price is in the money of the day. A real price divides it by an index of other prices, answering a specific question: how much of something else must society give up for a tonne of this? Change the something else and the answer changes.

Radetzki and Wårell list GDP deflators, US producer and consumer price indexes, and the UN's index of dollar prices of manufactured exports, the MUV index. They prefer MUV because it compares commodities with the manufactures they trade against and avoids distortions when exchange rates move faster than national prices. Its weakness is that it omits manufactures from China and the trade in services. Erten and Ocampo also use it, and note that with the US CPI oil after the second oil shock looks dearer than around 2010, whereas with MUV the 2010 level is higher.

Tilton shows what rides on the choice. Potter and Christy, deflating by producer prices, found mineral prices fell over 40% between 1870 and 1957, almost all in the first decade. Nordhaus, deflating by wages, found falls of 97% for aluminium, 90% for petroleum, 87% for copper, lead and zinc between 1900 and 1970: an hour of work buys vastly more metal. A further bias runs the other way. Price indexes overstate inflation because they miss quality gains and new products; a US commission put the CPI bias at about 1.1% a year. Commodity grades do not improve that way, so real series understate true trends. Adjust oil's real price by 0.75-1.25% a year and a flat long-run line becomes a rising one.

• Radetzki & Wårell (2021), A Handbook of Primary Commodities · pdf pp. 112-113
• Erten & Ocampo (2012), Super-cycles of commodity prices, UN DESA WP 110 web ↗ · pdf p. 8
• Tilton (2003), On Borrowed Time? · pdf pp. 51-53, 60-63

4.2 The evidence: falling for crops and most metals, rising for oil, and never fixed

Grilli and Yang's classic study deflated 24 commodities by manufactures prices over 1900-1986. They found a decline of 0.59% a year, about 40% in total, steepest for metals at 0.84% a year, while tropical beverages rose. Cashin and McDermott, covering metals and agricultural raw materials from 1862 to 1999, found 1.3% a year and a fall of four fifths over the twentieth century. Jacks, using 30 series from 1850 to 2012 deflated by the US CPI, found real prices rising from 1950 onwards, with energy, precious metals and minerals rising steadily since 1900, while grains and softs fell. But his series ended in 2012, mid-boom; another end year could reverse the result.

Erten and Ocampo let the trend itself drift. Real non-oil prices edged up from 1865 to 1885, fell 47% over 109 years to 1994, then rose 8% to 2010. Metals fell 48% from 1881 to 1974 and then rose 44%. Tropical agriculture fell furthest, 67% between 1888 and 2002, which they link to surplus labour in tropical farming. Real oil, by contrast, rose 280% between 1962 and 2010.

Tilton's survey of the scarcity literature reaches two conclusions. First, despite consumption on a scale never seen before, depletion has not produced rising real prices that curtail use. Barnett and Morse found labour and capital per unit of mineral output fell more than 75% between 1870 and 1957, and Slade's U-shaped trends, suggesting scarcity setting in before 1973, were undercut by the price falls of the 1980s and 1990s; an update found the U-shape fitted only nickel. Second, the trends are not fixed. They shift with technology and growth, so extrapolating them is risky.

• Radetzki & Wårell (2021), A Handbook of Primary Commodities · pdf pp. 113-115
• Erten & Ocampo (2012), Super-cycles of commodity prices, UN DESA WP 110 web ↗ · pdf pp. 9-11, 25
• Tilton (2003), On Borrowed Time? · pdf pp. 49-57, 64-66
• Tilton & Guzmán (2016), Mineral Economics and Policy · pdf pp. 229-231
• Conway (2023), Material World · pdf pp. 212-213

4.3 Why the trend: Prebisch-Singer, technology against depletion, and Hotelling's rule

In 1950 Prebisch and Singer argued that productivity gains are shared unequally: in competitive commodity markets they pass to buyers as lower prices, while in manufacturing organised labour and firms keep them as higher wages and profits. Erten and Ocampo add low income elasticity of demand, and read their finding that each non-oil super-cycle averaged lower than the last as support. Radetzki and Wårell list less controversial causes: low income elasticity, falling freight costs that weigh more on bulky goods, and quality gains in manufactures that price indexes miss. Svedberg and Tilton found that correcting for quality turns copper's falling real trend into a rising one.

Tilton and Guzmán provide the supply-side frame. The long-run supply curve rises at first, then flattens because lower-grade deposits are far more plentiful. Over the relevant range demand growth barely moves the long-run price, though it changes output a great deal. The trend comes from shifts in the curve: depletion pushes it up, technology pushes it down, and for most of the past century technology won.

Hotelling offered the opposite prediction in 1931. A deposit is an asset. If its value in the ground grew slower than the interest rate, owners would mine faster, and vice versa, so in equilibrium the scarcity rent, or user cost, rises at the rate of interest and prices climb. That rests on strong assumptions: known, uniform deposits, no exploration and no technical change. Relax them and user costs can fall. Tilton reports studies finding user costs negligible for long-lived resources, oil included.

• Radetzki & Wårell (2021), A Handbook of Primary Commodities · pdf pp. 115-117
• Erten & Ocampo (2012), Super-cycles of commodity prices, UN DESA WP 110 web ↗ · pdf pp. 5-6, 25-26
• Tilton & Guzmán (2016), Mineral Economics and Policy · pdf pp. 115-117, 120, 223
• Tilton (2003), On Borrowed Time? · pdf pp. 24-27, 63-65
• Geman (2005), Commodities and Commodity Derivatives · pdf p. 193

4.4 How to read the price charts on a tile

Each tile with a public price series shows, in its cycles lens, the latest monthly price and its change against the same month one, five and ten years earlier, plus two charts: monthly nominal since 2000 and annual real since 1960, in 2010 US dollars deflated by the MUV index. Each answers a different question, and each can mislead.

Percentage changes depend entirely on the start month. Potassium chloride shows -50% over five years, but September 2021 was already deep into the run-up to the 2022 spike; against the 2020 lows the comparison would look very different. European gas shows only +11% over five years, from $22.84 to $25.42, yet in between it hit $70.04. Always look at the chart before trusting the stat.

Monthly averages smooth away extremes. Brent's April 2020 average is $23.3, yet for a few hours that month the US benchmark traded below zero.

Nominal records are not real records. Copper's $14,474 a tonne in September 2026 is its highest monthly price since 2000. In real terms the 2025 average of $8,568 is the highest annual figure since 1960, above 2011's $7,950 and the 1966 peak of $7,393.

Finally, check which price it is: Brent is a North Sea crude, the gas series is European, iron ore is 62% fines delivered to China. Other grades and places diverge: US Henry Hub gas, for instance, has long traded far below European gas. Many minor materials have no public series because they trade privately.

• Radetzki & Wårell (2021), A Handbook of Primary Commodities · pdf pp. 112-113, 124-125
• Humphreys (2015), The Remaking of the Mining Industry · pdf p. 105

See it in the table

Check your understanding

Nordhaus found far larger declines in real mineral prices between 1900 and 1970 than Potter and Christy did. Why?

According to Tilton and Guzmán, what mainly determines the long-run trend in real mineral prices?

A tile shows a commodity down 50% over five years. What is the first thing to check?