Data Protection Atlas

Sub-national patterns

method structure federalism

Federated countries do not vary in one shape. Four distinct patterns, what each one means for where the law actually lives, and why "solve the US, solve them all" is wrong.

“Does this country have sub-national data protection law?” is the wrong question, because a yes covers at least four structurally different situations. Every jurisdiction record declares which pattern it follows, and the pattern decides what a sub-jurisdiction page even contains.

Pattern 1 — Patchwork, no floor

Example: the United States.

No general federal private-sector law. Each state’s comprehensive act is independently substantive — different thresholds, different rights, different enforcement — and a federal sectoral layer (HIPAA, GLBA, COPPA, FCRA) sits alongside, unrelated to the state layer. There is nothing to inherit from. This is the hardest shape to model: each sub-unit is a full instrument in its own right.

Pattern 2 — Floor with substitution

Example: Canada.

A federal default applies everywhere unless a province has passed its own “substantially similar” law, in which case the provincial law displaces the federal one entirely for intra-provincial private-sector activity. Structurally closer to Pattern 1 than to the EU — the provincial laws are independently substantive, not derogationsDerogationA permitted departure from a rule. The GDPR's opening clauses let member states legislate nationally in defined areas, so national acts are derogations from a shared text rather than independent statutes. from a shared text — but there is still a named default that applies where a sub-unit has not replaced it.

Pattern 3 — Uniform private law, devolved public sector or regulator

Examples: Germany, Switzerland, Mexico, Australia.

The consumer-facing law is one national statute, full stop. Sub-national variation is confined to which regulator has jurisdiction, to public-sector bodies only, or to one narrow sector such as health data. The substantive private-sector law never forks. Sub-units here get a regulator lookup table, not a duplicated country profile — which is what stops the model generating sixteen near-empty pages with nothing in them.

Pattern 4 — Fully unified

Examples: the United Kingdom, and most of the world.

No sub-national axis at all. Note that this can be true even where other law is devolved: in the UK, data protection is reserved while freedom of information is devolved in Scotland — the two diverge in exactly opposite directions, and assuming one from the other is a reliable way to get it wrong.

Why the distinction is load-bearing

It is tempting to build the hardest case (Pattern 1) and declare federated jurisdictions solved. They are not. Pattern 1 work validates Pattern 1 and, partially, the independent-substantive-sub-unit mechanics that Pattern 2 also needs. It does not validate Pattern 3, which is a different problem — jurisdiction-of-regulator and public/private split, not competing substantive rights — and needs its own check when a Pattern 3 country is profiled in depth.

The build order follows from that: prove the hard pattern on two deliberately different sub-units, then bank easy wins elsewhere, alternating so neither half of the scope races ahead of the other.

Last verified 25 July 2026