The method, and its honest scope
For each metal we take its ore/concentrate HS-2017 line and its refined / first-processed basket (refined metal plus the dominant ferroalloy or intermediate route, so the “refined” stage reflects the form actually traded — e.g. ferronickel for nickel, ferrochrome for chromium). For each stage we compute the exporter Herfindahl (HHI) on trade value, then the wedge = HHI(refined) − HHI(ore). Positive = processing more concentrated than extraction.
Source: CEPII BACI (HS-2017), 3-year mean 2022–2024, exporter value shares → refining.json. Value, not tonnage: BACI quantities carry unit artifacts (they put one EU smelter at 66% of world zinc by weight but 13% by value). Honest scope: this is concentration of tradeable processed supply. A refiner that consumes its output at home — China turning imported concentrate into battery cathode, or into magnets — exports little refined metal and is understated here; that capacity shows up in the production / GeoPolRisk layer instead. This page is specifically about who controls the metal that crosses borders in processed form.
Ore stage → refined stage, ranked by the wedge
Each bar shows the top exporters’ value share at that stage (darker = the single leader). A wide positive wedge means the refined bar is far more concentrated than the ore bar — a processing chokepoint. A negative wedge means refining is more spread out than mining (no furnace chokepoint beyond the ore itself).
| Material | ore exporters | refined exporters (form) | wedge | refining capacity reported (IEA/USGS) |
|---|
The refining you can’t see in trade — capacity vs exports, per metal
The last column above is the fix for this page’s central blind spot. Trade only sees refined metal that crosses a border; a country that refines and then uses the metal at home — China turning concentrate into battery cathode, magnets, chips — barely exports refined metal, so it vanishes from the wedge. Setting the reported refining-capacity share (IEA/USGS) beside the trade-export leader makes the gap explicit and per-metal:
The capacity wedge — the same question, on production instead of trade
Better than an overlay: for six minerals the IEA publishes mining and refining supply by country, so we can compute the wedge on physical capacity rather than infer it from exports — and it reaches the battery chain (lithium, graphite, magnet rare earths) whose refined stage has no clean trade code and is invisible to the wedge above. Concentration here is the top-3 share (the IEA’s own measure; it avoids a “rest of world” lump distorting a Herfindahl).
| Mineral | mined — leader (top-3) | refined — leader (top-3) | capacity wedge |
|---|
Where the two columns agree (niobium→Brazil, nickel→Indonesia, tungsten & tantalum→China) the refined form is genuinely traded, so the wedge sees the chokepoint. Where they diverge — manganese, copper, bauxite, cobalt, lead, zinc — the metal is refined and consumed at home, so trade goes quiet and only capacity data reveals it. The capacity column is computed for the minerals the IEA Critical Minerals Dataset (CC BY 4.0) covers — copper, cobalt, nickel here, plus lithium/graphite/magnets in the capacity wedge below — and curated (IEA GCMO 2025 + USGS MCS) for the rest, which the table labels. Why curated at all? Beyond those six there is still no clean open capacity series: the USGS production tables omit China at the refinery stage (their copper-refinery table lists 14 countries and not China), and facility-level capacity databases are proprietary. So the honest split is: computed where an open source exists, cited estimate where it doesn’t — never silently mixed.