The control that couldn't fire: Russian palladium and the dependency-too-deep signature

Trade-flow companion to the price wedge (R72), alternative-track. This corpus usually catches a control leaking — banned material re-entering a buyer through a laundered origin (antimony → Thailand), or a legally-relabelled one (Russian gold → Dubai recast). Palladium is the inverse and equally instructive case: the most single-origin-concentrated platinum-group metal, produced ~40% by one Russian firm, where the West declined to impose the hard control at all — and the one hard attempt (a US antidumping/countervailing case) was killed on a negative-injury finding. So there is no phantom-flag fingerprint to trace, because nothing was banned. The signal is the absence of the control and what that absence reveals: direct Western dependence on Russian palladium grew under a nominally hostile posture. This is a DUAL-SCORE / alternative-track reading — never folded into any Tier-1 exposure score. Research, not investment advice; every figure traces to a free public source; the ownership pipe is a disclosed subsidiary, not an alleged smuggling shell.

Live chart + method: [/situation-room → "Is the ban leaking?"](/situation-room).

Verdict

Russia is the world's largest palladium producer — Nornickel accounts for ~40% of global primary palladium (~2.8 Moz of ~7.0 Moz/yr; USGS puts 2024 Russian mine output ~75 t of ~190 t world), and Russia + South Africa are a ~80% duopoly of primary supply. Palladium is the irreplaceable autocatalyst and electronics metal — there is no near-term non-Russian, non-South-African substitute. That single fact governs everything the West did not do:

The gold ban stuck because Western use of Russian gold (jewellery, reserves) is substitutable. The palladium ban never came, and the trade-remedy substitute for it failed the injury test — because banning it would have injured the very domestic buyers it claimed to protect. "Too important to sanction" is not rhetoric here; it is the legal finding.

The inverse fingerprint: direct dependence grew

Every leak case in this corpus opens with the banned origin's line to the buyer collapsing (China Sb₂O₃ to US −85%). Palladium does the opposite:

Sources: USGS Mineral Commodity Summaries 2025 (Platinum-Group Metals, 2020–23 import-source shares); Jan–Jul 2025 tonnages via Discovery Alert citing USGS PGM monthly data. Production shares: World Platinum Investment Council; USGS MCS 2025/2026; Nornickel.

Russia's direct line to the US did not collapse — it rose to 43% of imports and became the single largest source, with Russian palladium exports to America up ~35% (2022→2024) even as the palladium price fell ~50%. There is no implausible-origin surge to detect because there was no ban to route around: the material walks in the front door, correctly labelled Russian.

Where re-routing did happen — two genuine legs, no phantom flag

The rerouting that occurred is buyer-diversion and genuine re-refining, not zero-capacity relabelling — which is exactly why it stays GATE 0:

1. China leg (re-routing to a non-sanctioning buyer). China emerged as Russia's largest palladium customer, its imports surging through 2023–24; Russia's precious-metals exports to China jumped ~80% to ~$1 bn in H1-2025 (Nornickel redirecting to Asian auto demand). This is the gold-case China leg repeated on the PGM axis — redirection to a willing buyer, not a wash back into the West. 2. Switzerland leg (genuine re-refining / mode-C potential). Switzerland is the main origin of EU palladium imports in semi-manufactured form. Swiss refiners (Metalor, PAMP, Valcambi, Argor-Heraeus) are genuine, high-capacity houses, so Russian metal recast there acquires a legally Swiss semi-fabricated origin — the transformation-washing mechanism that could dodge the UK 35% tariff and the LPPM Good Delivery block. But because the refiners are real (not ~0%-capacity), no free dataset isolates the Russian-origin fraction inside "Swiss" semi-manufactures — a labelled gap, the same structural blindness as Russian gold recast in Dubai.

The common-ownership pipe — disclosed, not a shell (tell = Y, qualified)

The antimony case's tell was a namesake shell (Guangxi Youngsun → Thai Unipet → Youngsun & Essen). Palladium's ownership pipe is the opposite in character — fully disclosed and legal:

> PJSC MMC Norilsk Nickel (Russia — producer) → Metal Trade Overseas AG (Zug, > Switzerland — wholly-owned Nornickel Group trading subsidiary) → Western industrial > buyers (e.g. BASF's autocatalyst business).

This channel is not evasion of a ban — it is Nornickel's ordinary European distribution arm, and it kept operating precisely because OFAC spared it: the 23 Aug 2024 US action listed several Nornickel service subsidiaries (and the Bystrinsky project) but deliberately excluded the production and sales companies (PJSC MMC Norilsk Nickel, its Polar Division, Kola MMC). The tell here is not "here is the shell the ban leaks through" but "here is the disclosed corporate pipe the West chose to leave open." Same object as the antimony pipe — a controlled producer's own offshore leg — inverted in legality.

Entity chain: S&P Global Market Intelligence (Metal Trade Overseas sourcing palladium for BASF); company registry (Metal Trade Overseas AG, Zug, Nornickel Group). OFAC scope: US Treasury 23 Aug 2024 action; Kelley Drye / Barents Observer analysis. Presented as the disclosed distribution channel — NOT a per-shipment accusation.

Why it matters for the buyer

1. Concentration without a control is still exposure. A risk team that screens for "sanctioned Russian material" finds palladium clean — no ban, no order, front-door imports. Yet the underlying dependence (Russia ~40% of world supply, ~43% of US imports) is higher than for several materials that are controlled. The absence of a sanction is the trap, not the all-clear. 2. The injury test is the real chokepoint governor. The US AD/CVD case proves the mechanism: a control on an irreplaceable input cannot survive the harm it does to domestic buyers. Any future hard palladium control faces the same wall — so the honest forward read is continued open dependence, not imminent severance. 3. The leak is pre-positioned. Should a hard ban ever land, the escape routes are already built and genuine: the China buyer-diversion channel and the Swiss re-refining origin-wash. A palladium ban would leak on day one through legal channels — the inverse-fingerprint case is the early map of where.

Method & honesty rails

• Control classification first. The instrument here is mostly absent (no import

ban) or failed (AD/CVD terminated on negative injury). Read the flow through that lens: rising direct imports are the expected result of a non-control, not a leak.

• No phantom-flag fingerprint exists — and saying so is the finding. The China and

Switzerland legs are genuine buyers/refiners, not ~0%-capacity surgers; the Russian-content fraction in Swiss semi-manufactures is unquantifiable in free data (labelled gap).

• Ownership tell = Y but disclosed: Nornickel → Metal Trade Overseas AG (Switzerland)

is a public, legal trading subsidiary, not a laundering shell. Framed as diversion/ distribution, never as smuggling.

• Alternative-track only: never touches buyerRelativeScore or base exposure.
• New signature (dependency-too-deep): the mirror of Russian gold. Gold — Western

use substitutable → the West could ban → material laundered through Dubai/China. Palladium — Western use not substitutable → the West could not ban → material flows openly and direct dependence grows. When a control's absence is itself the data, read the concentration and the injury finding, not a customs-origin table.