Net trade, and why it still isn't full apparent consumption
For each material and bloc: net = imports − exports (value, ). A pure re-export hub imports and exports in equal measure, so its net → 0; a genuine consumer stays net-positive; a refiner-exporter goes net-negative. We overlay mine-production presence (USGS shares) so a bloc that both produces and net-imports reads as a true consumer.
Honest limit — and it is not the one this page used to claim. Net trade is not final consumption, but the reason is no longer “we lack production tonnes” (the production page added World Mining Data tonnages). Two real reasons remain. Apparent consumption breaks for by-products: gallium/germanium mine tonnage is host-country (bauxite, zinc), while the usable metal is recovered elsewhere — adding it to trade misattributes supply. And even correct apparent consumption misses demand embodied in finished goods (gallium in an imported chip), which for critical metals is the bigger channel; capturing it needs Raw Material Equivalents via a multi-region input-output model (EXIOBASE, Eurostat RME), and public MRIO resolves “electronics”, not gallium. So per-metal consumption is not observable in open data; netting is the honest ceiling. Inputs: flows × bloc_demand.json × data.json → net_demand.json.
Net trade position by bloc — demand right, supply left
Net-demand ranking — who pulls most once re-exports are removed
| Bloc | net import pull, key metals |
|---|
Where the demand arm lands
Netting out re-exports is the correction the whole demand arm was building toward: it separates the countries that use a metal from the ones that merely move it. The result sharpens the strategic picture — the West and the East-Asian manufacturers are the net pullers of the squeezed metals, while the dominant processor is, in net terms, their supplier. What it is not, stated plainly: this is a re-export correction, not a consumption model. Two things sit beyond it, and neither is closed by more trade data. Full apparent consumption would need refined-form production by country — mine tonnage misattributes by-products like gallium, whose usable metal is recovered far from where its host is dug. And final consumption proper would need to trace the metal embodied in finished goods (cobalt in an imported battery). We tried. A full material-footprint model (Raw Material Equivalents via a multi-region input-output table) captures every tier but is sector-coarse — it sees “electronics”, not gallium. And a bottom-up estimate (product trade × metal intensity), which we prototyped for cobalt on battery trade, turned out to answer a different question: it cleanly shows a trade-stage fact — intermediate cobalt is China-bound, finished cells are net-imported by the US and EU — but calling that “cobalt demand by country” is a category error, because net-export clamping erases China’s own large use, product-weight intermediates don’t convert to contained metal, and cobalt also rides in imported cars and electronics outside the battery code. So even cobalt resists a per-country final-demand figure. This page reads trade pull as a stage fact and names what sits beyond it, rather than dressing net trade as consumption.