Manufacturing Costs

Copper, aluminum and iron ore disagreed in 102 of 199 months. They cannot identify one demand story by themselves.

Copper, aluminum and iron ore had mixed monthly directions in 102 of 199 observations from January 2010 to July 2026, and in 45 of 113 after requiring each absolute change to exceed 1%.

Editorial evidence cutoff: September 9, 2026. Published September 29, 2026. Observation periods are stated throughout; older figures are retrospective evidence.

Copper, aluminum and iron ore often appear together as a quick read on industrial demand. Their common uses make the grouping understandable. The historical record makes a single shared monthly interpretation much harder to defend.

From January 2010 through July 2026, the three benchmarks moved in mixed monthly directions in 102 of 199 observations. All three rose in 54 months and all fell in 43. Roughly half the time, the set did not even agree on the direction of the monthly price change. Agreement, when it occurred, still did not identify its cause.

For September 2026 planning, the retrospective tests how much evidence a combined metals narrative actually supplies before it is used to explain demand.

Count the disagreement before explaining the signal

The calculation is deliberately simple. For each month, compare every metal's price with its own previous-month value. Then classify whether all three changes are positive, all are negative or the signs differ. Dollar price levels are never averaged across the metals.

That distinction matters because a tonne of copper and a tonne of iron ore are different products with very different prices and uses. An unweighted average of their dollar levels would not create an economically meaningful industrial-cost index. The directional comparison asks a narrower question that the data can answer.

Its answer is that mixed signs were common. This does not mean the series are useless or unrelated. It means that a writer cannot routinely compress their behavior into one direction without omitting part of the observed evidence. The individual paths belong in the analysis before any common-demand explanation is proposed.

The longer archive contains a measurement warning

The saved histories reach back to January 1992, but the headline comparison begins in 2010. Earlier iron-ore observations contain long stretches with unchanged benchmark values. Counting those repetitions as if they were equivalent to actively moving monthly prices would distort a simple sign-consensus exercise.

The full archive contains 415 monthly levels. There are 199 monthly changes in the selected January 2010 to July 2026 period, and no unchanged observations in that three-series comparison. The start date is a stated boundary, not a claim that the earlier data is erroneous or should be discarded for every purpose.

Explaining exactly why the early values repeat requires the historical benchmark methodology. The observed pattern alone does not establish that explanation. Keeping it out of the headline analysis avoids treating a potentially different measurement regime as directly comparable while that source question remains open.

Remove tiny changes and disagreement persists

One objection to a sign count is that a negligible move can turn an apparently coherent month into a mixed one. To test that, we retained only months in which the absolute change in every metal exceeded 1%.

That filter leaves 113 observations. Forty-five still had mixed directions, a share of 39.82%. Forty-one had all three rising and 27 had all three falling. The disagreement becomes less common than in the full sample, as expected, but does not disappear.

The threshold is a sensitivity check rather than a claim that 1% is economically important for every buyer. It also changes which months are being discussed. The filtered result should therefore sit alongside the full 102-of-199 count, not replace it as if it were a correction to the underlying data. Both help distinguish meaningful variation from a claim driven entirely by rounding noise.

Annual agreement can hide a monthly reversal

In July 2026, all three prices were above their year-earlier levels. Copper was up 38.61%, aluminum 21.17% and iron ore 0.37%. That shared annual sign could easily become the basis of a broad strengthening-demand headline.

From June to July, all three fell instead: copper by 0.07%, aluminum by 8.16% and iron ore by 2.12%. The copper move was small, but the horizon reversal in the set remains a useful warning. Annual agreement does not establish that the latest month continued the same direction.

The June comparison was mixed again: copper rose 0.30%, while aluminum fell 6.00% and iron ore fell 7.02%. These observations do not tell us whether one horizon is universally better. They show that the measurement window belongs in the headline and chart whenever the grouped series are used to describe current conditions.

A common cause requires evidence beyond three prices

Industrial demand can matter to each of these markets. So can supply, inventories, currency movements, market expectations and the particular benchmark being measured. A common price movement does not identify the relative contribution of those influences.

Attributing agreement specifically to China would require appropriate evidence about Chinese production, consumption, inventories or end markets, together with consideration of competing explanations. This investigation contains no such control series. The fact that three globally traded materials move together cannot, by itself, locate the common cause in one country.

The same limit applies when they diverge. Different directions do not prove that one sector is strong while another is weak. Those are hypotheses worth testing against additional primary data. A compelling narrative about construction, electrification or manufacturing becomes an investigation only when the supporting evidence is actually brought into the comparison.

Keep three useful histories instead of forcing one verdict

The copper history, aluminum history and iron-ore history remain useful inputs for readers whose products or suppliers are exposed to those markets. Displaying them together can reveal the very divergence that a composite hides.

A better presentation preserves each source-defined product, states the time window and distinguishes observed co-movement from the proposed explanation. A manufacturer's actual exposure should then use its own quantities, grades, contract terms and delivered costs. Equal attention on a chart does not imply equal weight in a purchasing budget.

The investigation therefore stops short of a grand verdict about the world's factory floor. The result it can defend is specific: 102 of 199 monthly observations had mixed directions, and 45 of 113 still did after filtering out small moves. That is enough to question an automatic common-demand interpretation while leaving a clear research path for anyone who wants to establish one.

Sources and calculation

The analysis uses saved IMF global commodity-price series via FRED: copper, aluminum and iron ore. Prices are monthly USD-per-tonne benchmarks. Changes share exact calendar endpoints. The headline sample begins January 2010; earlier repeated iron-ore values and their methodology require separate treatment. No demand attribution, weighting model or predictive claim is made.

Sources and evidence

Evidence period: January 2010 to July 2026; longer source archive begins January 1992. The frozen evidence record lists the source files and verified hashes available September 9, 2026. Source revision: 5e4fb7726c3d40060c0151c086c903baae856cab. Later live-data updates do not alter the historical evidence in this article.

fred.stlouisfed.org/series/PCOPPUSDM

fred.stlouisfed.org/series/PALUMUSDM

fred.stlouisfed.org/series/PIORECRUSDM

Published 2026-09-29.