Critical Materials Atlas
Method · demand · apparent consumption

Who actually uses each metal?

Imports show who buys a metal; net trade shows who keeps it — but both erase the metal a country refines and consumes at home. Apparent consumption adds domestic production back in. It is the textbook measure, and every metal is graded A–D on a scorecard — only the ones that earn it are published as measured.

How it is built, and the validation scorecard

Apparent consumption = refined production + imports − exports, per bloc. Production restores the domestic use net trade erases. Production: IEA Critical Minerals Dataset 2024 (refining by country, CC BY), with EU-27 refined copper from Eurostat PRODCOM (ds-059358), and USGS MCS 2025 as a cross-check. Trade: CEPII BACI 2023, refined-form HS codes.

The scorecard (upgraded from a single pass/fail gate): each metal is graded on five checks — stage-match (production stage = trade stage), HS purity (does the code contain the refined form, or bundle intermediates?), contained-metal basis, global closure (world AC ≈ world refined production — but this ~cancels under balanced trade, so it mainly catches missing/units, not fine error), and a reproducibility check (does an independent build of production + net trade land near the study group's published usage figure?). Note the study group's usage is itself apparent consumption, so this reproduces our sum, it does not independently confirm it. A = clean stage + clean HS + current year + closure holds + a study-group figure to reproduce against; B = one compromised (stale year / bundled HS / no cross-check); C = stage mismatch or absurd arithmetic; D = HS bundles intermediates (bloc layer). → apparent.json.

Per-country — the real measurement

Refined production by country (USGS Minerals Yearbook) + refined-form trade (BACI), per ISO country — not bloc. This is the measured layer the whole demand arm was building toward, one metal at a time.

What no tier here certifies — the finding. None of these grades certifies against an independent measurement of consumption, because no free one exists. The consumption figures everyone cites for these metals — ICSG for copper, ILZSG for lead — publish “usage” that is itself apparent consumption (production + net trade), the same sum we do. So when an independent build lands within 1–2 points of the study group, that is the same sum done twice — a reproducibility check that our trade parsing and arithmetic are sound, not two sources agreeing. Nobody is measuring use bottom-up; production + net trade is the industry’s number. A tier here therefore means clean data plus a reproducibility check — not certified truth.

By bloc — the metals that earn a grade

Tier A or B, at bloc level. Each row: how apparent consumption is built, its scorecard, and the independent figure it matches.

Not publishable — and why that's shown, not hidden

Cobalt, nickel and rare earths are graded D: their trade codes don't mean what "refined production" means — so production + imports double-counts the intermediate feedstock, or the code misses most of the metal. The absurd shares are the evidence. The fix is real but not free: each needs national 8-digit customs lines (e.g. China 81052010 feedstock vs 81052020 refined cobalt) — a later phase of the registry.

The ceiling, stated plainly