Governance-weighted criticality

The same supply concentration is more dangerous when the producers are badly governed. Here I re-weight every material’s concentration by the governance of its producers — and watch the risk list reorder.

How it’s built

Each producer gets a governance-risk weight g = (2.5 − WGI)/5 from the World Bank Worldwide Governance Indicators, which feeds a governance-weighted concentration score Σ shareᵢ²·gᵢ. From that I build an EU/SCRREEN-shaped supply-risk proxy (bottleneck-stage governance-weighted concentration, discounted by recycling, scaled by substitutability) and a Graedel-style supply-risk axis — both set beside my own index.

These are proxies on public approximations, not the official EU or Yale scores (which use CRM-specific, stage-granular, expert-weighted inputs) — read it as a governance lens. Depletion-time and environmental axes are omitted (no tonnage/LCA data), and production concentration uses top-N mine shares. Computed by build_criticality.py from public data.

The rankings sit in the same family — Spearman ρ +0.76 (EU-shaped vs our index), +0.60 (Graedel-shaped vs ours), +0.81 (the two proxies) — but this is partly mechanical: all three share inputs (concentration, recycling, substitutability), so read it as "same family", not independent validation. What the governance weighting changes: it pushes up materials with a high-governance-risk producer in the mix — Cobalt oxides & hydroxides (🇨🇩Congo [DRC] g0.82), Aluminium ores / bauxite (🇬🇳Guinea g0.71), Tantalum, unwrought (🇨🇩Congo [DRC] g0.82), Natural borates (🇹🇷Turkey g0.63), Antimony, unwrought (🇹🇯Tajikistan g0.69) — and down those mined in well-governed countries — Lithium carbonate (🇦🇺Australia g0.19), Helium (🇺🇸United States g0.32), Coking coal (🇦🇺Australia g0.19), Strontium carbonate (🇮🇷Iran g0.76), Hafnium, unwrought (🇦🇺Australia g0.19).

#	Material	EU-SR	Graedel	ours	producer risk	gov. effect
1	Gallium ⛓	100	88	57	🇨🇳 0.55	·
2	Beryllium, unwrought	81	71	74	🇺🇸 0.42	·
3	Ferro-niobium	66	79	64	🇧🇷 0.53	·
4	Cobalt oxides & hydroxides	64	76	43	🇨🇩 0.76	▲ 7
5	Natural graphite	58	78	51	🇨🇳 0.56	·
6	Silicon, < 99.99%	53	76	47	🇨🇳 0.52	·
7	Magnesium, unwrought	53	75	60	🇨🇳 0.55	·
8	Rare-earth permanent magnets	48	75	52	🇨🇳 0.54	·
9	Aluminium ores / bauxite	44	58	45	🇦🇺 0.50	▲ 6
10	Germanium ⛓	43	72	44	🇨🇳 0.55	·
11	Fluorspar, >97% CaF2	42	75	51	🇨🇳 0.55	·
12	Ferro-vanadium	38	72	45	🇨🇳 0.57	·
13	Platinum, unwrought	36	72	39	🇿🇦 0.55	·
14	Tungsten, unwrought	35	70	46	🇨🇳 0.55	·
15	Lithium carbonate	34	66	64	🇦🇺 0.33	▼ 11
16	Phosphorus	32	68	57	🇨🇳 0.53	·
17	Natural borates	31	62	42	🇹🇷 0.51	▲ 3
18	Tantalum, unwrought	27	71	33	🇨🇩 0.66	▲ 5
19	Arsenic	26	68	31	🇵🇪 0.55	▲ 2
20	Titanium, unwrought	26	65	43	🇨🇳 0.49	·
21	Manganese ore	25	66	46	🇿🇦 0.51	·
22	Strontium carbonate	20	67	57	🇮🇷 0.56	▼ 4
23	Palladium, unwrought	20	64	25	🇷🇺 0.58	·
24	Helium	18	69	22	🇺🇸 0.38	▼ 8
25	Antimony, unwrought	13	61	35	🇨🇳 0.61	▲ 3
26	Nickel, unwrought	12	59	32	🇮🇩 0.51	·
27	Phosphate rock	11	64	31	🇨🇳 0.53	▲ 2
28	Baryte	8	63	15	🇮🇳 0.57	▲ 2
29	Feldspar	8	55	25	🇹🇷 0.55	▲ 2
30	Coking coal	5	62	24	🇦🇺 0.40	▼ 5
31	Hafnium, unwrought	4	64	32	🇦🇺 0.42	▼ 4
32	Refined copper cathodes	0	54	23	🇨🇱 0.50	·

⛓ gallium and germanium share one HS6 code (811292; hafnium is separate). WGI: World Bank (mean of 6 estimates). Computed from data.json + wgi.json → criticality.json.

Concentration in risky places