Secondary-Sanctions Triage Report (50% Rule Proxy)

Date: 2026-04-20 Method: Graph-based adjacency analysis. Organizations with strong litigation or ownership linkage to already-sanctioned entities are flagged as candidates for OFAC 50% Rule investigation (or equivalent EU/UK/UA rules). This does not apply the Rule itself — it provides the analyst triage set.

Scope universe

• Primary sanctioned: 5,655 Organizations flagged sanctioned=true (US OFAC SDN)
• Multi-list sanctioned: 4,363 Organizations linked to EU/UK/UA SanctionedEntity nodes via ON_SANCTIONS_LIST edges

Tiered triage results

Tier A gap (important)

Our SHIFT_EVENT UBO layer only covers sanctioned organizations — UBOPersons are linked to SDN orgs but not to non-SDN ones. This blocks the cleanest 50%-Rule proxy (shared beneficial owner detection).

Fix: EGRUL bulk enrichment for non-SDN Russian entities would unlock Tier A. Without it, we can't surface orgs owned by the same person as an SDN but not currently flagged. Roughly 4-5M RU legal entities have EGRUL records; full ingestion would take ~1-2 weeks.

Tier B — highest-confidence candidates

Three entities meet the strict criterion:

Tier B2 — high-value triage (top 20)

Policy interpretation

1. Federal administrative entities dominate the top tier — Moscow Property, Bailiff Service, Pension Fund, FAS regional branches. These are not 50%-Rule candidates in the traditional sense (they're sovereign organs), but their heavy SDN litigation footprint is policy-relevant evidence for CAATSA §231 (secondary-sanctions on institutions facilitating SDN transactions).

2. Industrial/utility primary targets confirmed: ChMK (steel), Мосводоканал, Россети Сибирь, Мосэнергосбыт, Т Плюс, Сухой Aviation — these are individual-designation candidates, not just 50%-Rule scope. Several already surfaced in our prior briefs.

3. "Rifey" (7204189710) and "Спецметалл" (7456021049) are mid-tier commercial LLCs — these are the true 50%-Rule candidates (small commercial actors heavily entangled with SDN, not sovereign administrative bodies).

How to query

Each flagged Organization has:

• secondary_sanction_candidate = true
• secondary_tier ∈ {'B', 'B2'} (tier assignment)
• secondary_via_inns — array of up to 20 sanctioned-org INNs linked
• Outgoing :SECONDARY_SCOPE edges with reason ∈ {'heavy_litigation', 'moderate_litigation'}

Example:

MATCH (o:Organization {secondary_sanction_candidate: true})-[r:SECONDARY_SCOPE]->(s:Organization)
WHERE o.secondary_tier = 'B'
RETURN o.inn, o.name, collect(s.name) AS via

Limitations

• No true 50%-Rule arithmetic. We don't have share-percentage ownership at scale. Tiers B/B2 are adjacency proxies, not share-weighted ownership computations.
• Tier A blocked by data gap. Without non-SDN UBO data, cross-boundary shared-controller analysis isn't possible.
• Tier C blocked. The direct procurement edges (CONTRACTED_WITH) don't cross the SDN boundary in our current graph — sanctioned orgs don't have CONTRACTED_WITH edges to non-sanctioned orgs in the data we have. A different query over AWARDED_CONTRACT → Contract → Contract.customer_inn/supplier_inn might surface supply-chain-level candidates. Worth a follow-on tile.
• Recomputed on every pipeline run — analysts working with tier flags should pin their findings to a specific timestamp.

Artifacts

• Module: pipeline/neo4j_ingest/propagate_sanctions.py (wired into run_all.py MODULE_ORDER, runs after ubo_shifts)
• Graph properties: o.secondary_sanction_candidate, o.secondary_tier, o.secondary_via_inns
• Graph edges: Organization-[:SECONDARY_SCOPE {reason}]->Organization
• Related: bridge_entities_discovery.md, federal_ministries_sdn_entanglement.md, okhrana_rosgvardiya_memo.md

Next steps to improve coverage

1. EGRUL bulk UBO ingestion — biggest unlock, fills the Tier A data gap
2. Rebuild procurement graph to include customer_inn / supplier_inn INNs as direct Organization-to-Organization edges (currently via Contract intermediary which breaks the query pattern)
3. Share-percentage data from rusprofile/EGRUL for orgs with SDN ownership — enables true 50%-Rule arithmetic on a subset of targets
4. Periodic re-run via cron — captures new OFAC/EU designations automatically as they cascade into graph properties