Market transmission case study — US silicon-metal AD/CVD enforcement arc, 2020 and 2025–2026 (Ferroglobe petitions)

Clustered case for the per-event qualitative layer (2026-09-13). Covers four register entries that are one recurring petitioner playbook, not four independent events — Globe Specialty Metals / Ferroglobe PLC and Mississippi Silicon LLC have run the identical AD+CVD petition mechanism against successive waves of silicon-metal exporters twice in six years:

Only listed beneficiary in either cycle: Ferroglobe PLC (NYSE: GSM). Mississippi Silicon LLC (the co-petitioner in both cycles) is privately held — no public instrument exists for it. Research, not investment advice. All moves verified on Yahoo Finance daily closes (chart API).

Verdict

Both cycles: the reflex trade — long GSM into an affirmative AD/CVD determination — does not work, and the register's own text explains why. The petitions exclude China, which the 2025 CVD filing's own downstream section calls "the dominant global supplier" of silicon metal. Duties on Kazakhstan/Bosnia/Iceland (2020) or Angola/Laos/Australia/Norway/Thailand (2025–26) raise the cost of the marginal ex-China tonnage without touching the price-setter. GSM's reaction across four determination dates is small, inconsistent in sign, and fully swamped by unrelated moves within days:

The one genuinely large single-day move adjacent to this arc — 2026-02-23, +5.3% (4.91→5.17) on the underlying final LTFV (dumping-margin) determination for Angola/Laos, which is not itself a separate register entry — still reversed within the week (5.22→5.01 by 02-26) and sits inside a much larger context: GSM ran from a 2026-02-20 close of 4.91 up to a 2025-07-25 high of 5.56 and back down to a 2026-06-30/07-03 low of 3.42, a ~35% peak-to-trough round trip that has nothing to do with any single filing date. By 2026-09-13 GSM was back to 4.28–4.67 — roughly where it sat before the 2025-09-26 CVD determination, i.e. the entire enforcement arc net-nets to noise against GSM's own volatility band.

Channel walk

• Ch.0 direction gate — restrictive/protective (duties raise landed cost

of the named-origin tonnage), so this does not fail the direction gate the way the Brazil Gecex 799 or PIC cases did. It clears Ch.0 but fails on scope.

• Ch.-4 dominant-supplier-exclusion gate (new). When a trade remedy

protects a domestic petitioner from minor-origin dumping/subsidies but explicitly leaves the actual price-setting supplier untouched — here China, named in the action file's own text as outside the case's scope and still dominant — the protected name's stock reaction is capped to a small, fast-fading pop (or no pop at all) rather than a durable re-rating. The duties change the marginal supplier of the ~10-20% of US silicon-metal imports these origins represent; they do not change who sets the world price. Apply this gate before screening any AD/CVD action for a long domestic-incumbent trade: ask whether the actual dominant supplier is in-scope. If not, expect Ch.3 (already-priced) and Ch.7 (scale) to both bite even when the headline rate looks punitive (94.44%, 240%).

• Ch.1 ex-China pure-play — inverted here: there is no "alternative to

buy" because this isn't a supply-cut case, it's a domestic-incumbent protection case. The candidate long IS the incumbent (GSM), which is why this case belongs in the register as a distinct mechanism class from the Ga/Ge/Sb template.

• Ch.2 domicile trap — N/A, no restrictor-country asset exposure to short.
• Ch.3 already-priced — fails, both cycles. Each case ran 8-11 months

from petition/initiation to the determination the register files as the event (2025-05-21 initiation → 2025-09-26 CVD prelim = 4 months of visible process; 2025-09-26 prelim → 2026-04-16 final order = another 7 months). AD/CVD preliminary-to-final outcomes for a petitioner with a >70-year domestic-industry track record are close to a foregone conclusion once initiated (both cycles: 100% affirmative across every named origin) — the market prices the probability-weighted outcome well before the formal publication date, leaving only the AFA-severity surprise (was it 21% or 240%?) to trade, and even that surprise moved GSM under 2%.

• Ch.4 consumer short — SKIP per standing policy; also weak here, since

silicon metal is a small-percentage input for aluminum alloying/polysilicon/ silicones and the excluded-origin duties don't remove China-sourced supply those consumers can still buy.

• Ch.5 sequel — active: Australia and Norway's AD track (final LTFV

2026-06-30, no order yet as of this filing) will produce a fifth determination date. Per Ch.-4, expect the same muted pattern — flag it in 2026-07-05-sequel-watchlist-next-events-alternatives-map.md as a negative sequel entry (a name to skip on the news, not to buy).

• Ch.6 round-trip clock — the one real pop (2026-02-23, +5.3%) round-tripped

in under a week; the two register-filed dates round-tripped in 2-4 sessions.

• Ch.7 hard-catalyst filter — fails on materiality: none of the five

named origins (Kazakhstan, Bosnia, Iceland, Angola, Laos — plus Australia, Norway, Thailand pending) is large enough in US silicon-metal import share for a duty to move GSM's ~$400-500M market cap meaningfully once China remains the marginal price-setter globally.

• Ch.8 quality gate — N/A, no bench published; GSM itself is liquid and

legitimate (NYSE-listed, >$400M mcap), the gate that fails is materiality, not name quality.

Product implication

This is the register's first domestic-incumbent-protection mechanism class, distinct from the ex-China-alternative template (Ga/Ge/Sb) and from the liberalizing/immaterial negative controls (Gecex 799, PIC fund, JBIC loan). The new Ch.-4 gate generalizes: any US/EU trade remedy that protects a domestic petitioner should be screened for whether the dominant global supplier is named in target_countries before assuming a long-the-petitioner trade works. Two independent six-year-apart cycles against the same petitioner, five different target origins, rates from 21% to 240%, and not one produced a durable GSM re-rating — this is now a load-bearing pattern, not a single anecdote. No new bench added to the alternatives map; Australia/ Norway's pending final order is queued to the sequel watchlist as a skip-on-the-news case, not a buy.

Sources: Federal Register (2025-09-26 Laos CVD prelim, 2026-04-16 Angola/Laos AD order), trade.gov press releases (2020-12-03 Kazakhstan CVD prelim, 2020-12-11 Bosnia/Iceland AD prelim), Global Trade Alert state-acts 65301/65375/91772/91773; all price moves verified on Yahoo Finance daily closes (chart API) for GSM, period 2020-09-30 to 2026-09-13.