Is the gate closing? Chrome-ore front-running before South Africa's export-permit regime

Trade-flow companion to the price wedge (R72), and the corpus's first entry on the South-Africa chromium chokepoint. Every prior case reads a control that is already in force and asks whether the banned line collapsed or leaked. This one catches a control before the gate shuts, and finds a different fingerprint: the controlled origin's own direct export line surging to a record in the window between announcement and enforcement. This is a DUAL-SCORE / alternative-track signal — never folded into any Tier-1 exposure score. Research, not investment advice; the illicit-provenance leg is INFERRED from public enforcement reporting and analyst estimates, never asserted as smuggling on any single shipment.

Verdict

On 25 June 2025 South Africa's Cabinet approved placing chrome ore under export control via the International Trade Administration Commission (ITAC), plus work on a ~25% export tax and electricity-tariff relief for domestic ferrochrome smelters — the first hard quantitative restriction on a major SA mineral export since 1994. But the binding instrument is not yet in force: the DTIC notice was gazetted only on 3 Oct 2025 (Gazette 53477, GoN 6712), public comment closed 31 Oct 2025, and the final permit notice was still pending at end-2025. South Africa is >70% of world seaborne chrome-ore exports and ~half of world chromite mine output (21 Mt of ~44 Mt, 2024), and China — the dominant ferrochrome and stainless producer — is the buyer. So the announcement created a wide-open ~6-month-plus window with a known, dated tax/permit cost arriving at the end of it. The flow did exactly what that incentive predicts:

Source: Mysteel, "Chinese demand fuels 2025 surge in South Africa's chrome ore exports" (SA customs export data). Attribution given: Chinese ferrochrome capacity expansion + widespread production halts at SA's own ferrochrome smelters.

This is the inverse of the antimony collapse. There, a live ban zeroed the controlled origin's direct line and pushed volume into zero-capacity relabel hubs. Here, a pre-enforcement control accelerated the controlled origin's own direct line to an all-time high — buyers and shippers front-loading ore ahead of a 25% tax and a discretionary permit. The leak is temporal (before the gate closes), not geographic (around a gate already shut).

The domestic provenance-wash (the second leg)

The other circumvention channel is not a foreign phantom flag — it is inside South Africa. Because the ore is bulk and low-value-per-tonne, it cannot be re-consigned through a ~0%-capacity transit country the way antimony oxide was (the phosphate / bauxite / tin lesson). Instead the relabel happens below the country-of-origin line, as illegal ore washing into the legal export pool:

Note the ~4× spread between the two volume estimates — they use different bases (black-market-traded vs physically-stolen) and neither is a customs figure. Presented as a range, not a point estimate.

The mechanism, per the Global Initiative field report: operators pose as legitimate artisanal-permit holders (a regime meant for <5-hectare operations), extract with heavy machinery in daylight, aggregate stolen ore at 20+ "spiral plants," truck it to Mozambique or Johannesburg for warehousing, then containerise it from Durban / Richards Bay — where it "vanishes into the legal supply chain, leaving no trace of its illegitimate provenance," bound predominantly for China. The Mozambique/Maputo leg is genuine logistics, not origin-relabel — the ore stays SA-origin; what is laundered is its legality, not its flag.

That makes this structurally analogous to the Myanmar-KIO heavy-REE case: the relabel is intra-origin, invisible at country-of-origin resolution. A permit regime keyed on legal exporters of record is blind to ore that entered the legal pool one step upstream — the enforcement gap the delay is widening in real time.

The reverse-hilirisasi shape (why this case is new)

The action files chrome alongside Indonesia hilirisasi, the Zimbabwe lithium-concentrate ban and the DRC cobalt quota as critical-minerals-nationalism. But the vector is inverted. Indonesia's nickel/bauxite bans prevent the value chain from leaving — ban the ore, force smelters onshore, and Chinese capital rebuilds the chain inside Indonesia (mode-B relocation arriving). South Africa's ferrochrome value chain has already left: SA supplied only ~1.6 Mt of ~15.9 Mt of world ferrochrome in 2025 (~10%), hollowed out over a decade by Eskom power costs and smelter closures (Merafe, Samancor idling), while China built the dominant ferrochrome industry on imported SA ore. The chrome control is an attempt to claw a completed migration back — using ore access as leverage to re-industrialise domestically after the fact.

This is the first reverse-hilirisasi / value-chain-repatriation case in the corpus. And its success turns on the constraint the Minerals Council names: the binding limit on SA ferrochrome is electricity price, not ore availability, so an export tax that raises ore cost for everyone may not rebuild smelters it cannot power — a policy whose circumvention risk (front-running + illicit leakage) is visible today while its industrial payoff is not.

Why it matters for the buyer

1. A dated control is a front-running signal, not just a supply risk. The moment a tax/permit has an announced arrival but no live gate, expect the controlled origin's own line to surge, not fall — the opposite of the post-ban collapse pattern. A risk model that only watches for severed lines will read a record export year as "supply secure" when it is the sound of a gate about to close. 2. Chromium has no substitute in stainless steel, and one dominant supplier. SA ~half of mine output, >70% of seaborne trade; China the concentrated buyer. Any discretionary permit regime over that flow is a load-bearing lever over global stainless cost — the exposure sits with Chinese ferrochrome/stainless producers and their downstream customers. 3. The provenance-wash is a permanent blind spot the permit won't close. ~10% of output moving through illicit channels that launder legality (not origin) means a permit keyed on legal exporters is porous by construction — the same intra-origin blindness as Myanmar heavy REE, here inside an OECD-adjacent legal system.

Method & honesty rails

• Trade data: SA export totals and the China leg are SA customs figures via

Mysteel; the illicit-volume figures are enforcement/analyst estimates (Business Day, EWN, Global Initiative), not free-customs series — hence the explicit range and the GATE 0.

• Why GATE 0: the binding instrument is not yet in force, so no

post-enforcement flow table exists; the "leak" is an anticipatory surge plus an estimated illicit channel; there is no common-ownership tell (the illicit channel runs on artisanal-permit cover and syndicates, not a cross-border namesake shell). Every hard trade figure is traced; nothing is invented; the fingerprint is real but not customs-confirmed. Lift toward GATE 1 only when (a) the permit regime goes live and post-control customs data lands, and (b) a specific relabel or ownership pipe is traced.

• Alternative-track only: this never touches buyerRelativeScore or the base

exposure — it sits beside them, like the price wedge.

• Inference, not accusation: illicit-provenance laundering is inferred from

public enforcement reporting; no individual shipment is asserted illegal beyond what the cited record states. The Mozambique transit leg is genuine logistics, labelled as such.

• New signatures added to the detector: (1) the **anticipatory front-running

surge — read the calendar gap between announcement and enforcement, not just the post-control collapse; (2) domestic provenance-washing (legality, not flag) as an intra-origin relabel; (3) reverse-hilirisasi** — a control aimed at repatriating a value chain that already migrated, the inverse of Indonesia.