What the dataset can support
This page exposes the coded footprint of the database: how many events are actually anchored to a source and to a framing quote, how the source register breaks down by confidence, and where the research is thin. Every count below describes this coded sample — it is not a claim about how much policy activity any government has actually undertaken.
By actor
Events, framing anchors and linked sources coded per actor — a measure of how deeply this project has researched each government, not of how much each government has done.
| Actor | Events | Framing anchored | Linked sources | Primary sources |
|---|---|---|---|---|
| China | 16 | 16/16 | 30 | 18 |
| United States | 20 | 17/20 | 54 | 53 |
| European Union | 6 | 4/6 | 17 | 16 |
| Australia | 3 | 3/3 | 5 | 5 |
| Japan | 7 | 4/7 | 6 | 5 |
| Canada | 10 | 5/10 | 13 | 12 |
| United Kingdom | 3 | 3/3 | 3 | 3 |
| India | 1 | 1/1 | 1 | 1 |
By mechanism
Events carry more than one mechanism, so these sum to more than the event count.
- Export control15
- Dual-use license12
- Catalogue listing1
- Critical-mineral designation8
- Supply-chain security measure29
- Countermeasure1
- Customs enforcement1
- Funding or subsidy37
- Stockpiling5
- Offtake agreement5
- Investment screening1
- Trade action5
Source confidence
Every source in the register, graded by the four-tier confidence hierarchy.
- Primary source171
- Official translation4
- Government / state media2
- Secondary analysis12
Research boundaries
What limits these counts, stated plainly rather than smoothed over.
Coded date window
18 May 2018 – 9 Oct 2026
No event outside this window has been coded yet. A gap here is a research boundary, not a claim that nothing happened before or after it.
Unresolved coding fields
None — every event's original-language title has been transcribed against its primary.
Actor-coverage imbalance
United States accounts for 20 of 66 coded events (30%). An actor with few events here is one this project has researched less — not one that has acted less.
This imbalance is a property of the research, not the underlying policy record, and it is why the table above is published rather than summarised away.
Direct source citations across the registry
187 of 189 registered sources directly support at least one published event, material, project, organization, programme, financing instrument, control clause, designation, framing claim, actor profile or reviewed project milestone.
2 registered sources have no explicit published record citation. This is a cataloguing relationship, not a measure of independent corroboration or research quality.
Inspect uncited registered sources →No synthetic scores
Nothing on this page is weighted, normalised or rolled up into a single number. The project's first rule is that classifications stay categorical and source-grounded — there is no overall “coverage score” here, because a single number would invite exactly the comparison this coded sample cannot support.
Category and field definitions live on the methodology page. The full dataset behind these counts is downloadable from /data. Data version v0.6-capital-intelligence, last updated 2026-11-03.