Is the duty leaking? Chinese aluminium extrusions re-labelled as "pallets" after the 2011 US AD/CVD orders

Alternative-track (R72) companion. Never folded into any Tier-1 score. Every other case in this series launders the origin. This one kept the origin (China) and laundered the product heading instead: extrusions were spot-welded into "pallets", a finished good outside the order's scope. The controlled line collapses and an adjacent uncontrolled line from the same origin swells. The origin stays honest while the tariff heading is false.

Anchor. Nothing in our register covers the 2011 US orders themselves (register coverage starts in 2020). The case is anchored to the nearest filed action on the same product, 2025-11-24-thailand-dft-ad-aluminium-extrusions-china (Thailand DFT AD1049). That filing is the same product meeting the same defence in a new buyer market, so this case's fingerprints are the ones to watch there. The controls are: US AD order A-570-967, 76 FR 30650 and CVD order C-570-968, 76 FR 30653, both published 26-May-2011, continued by 87 FR 66128 (2-Nov-2022) (Commerce, Federal Register).

Verdict

The ownership tell is proven in court. The volume fingerprint is only consistent with it, and is weak at HS-6.

1. Common ownership: Y. The importer *was* the exporter. A US federal jury (C.D. Cal., 23-Aug-2021) convicted six Liu-controlled US companies. DOJ says the 2.2 million "pallets" imported 2011–2014 had no customers and none were ever sold. They were stockpiled in Liu-owned warehouses while China Zhongwang's annual reports booked them as sales to independent third parties. The civil False Claims Act settlement: US$549.5m (DOJ, 12-May-2026). 2. Heading-hop: consistent with it, not diagnostic. The controlled heading collapsed in the order's first year. The adjacent basket heading from the same origin grew +69 % to its 2013 peak. It then fell back in 2014, the year DOJ says the scheme stopped (June 2014).

Table 1: US imports from China (UN Comtrade, reporter USA, annual, customs C00)

Source: UN Comtrade public API, reporterCode=842, partnerCode=156, flowCode=M, cmdCode 7604 / 761699, pulled 2026-09-28. Max-per-group rather than sum (partner2 = 0), to avoid dimension replication.

Why HS-6 is not diagnostic (read before quoting the +69 %).

• The unit value of HS 761699 did not fall. It was about US$6,085/t in 2010

and about US$6,215/t in 2013. Spot-welded extrusions entering at extrusion-grade value (HS 7604 ran about US$2,886/t in 2010) should have pulled the basket's unit value down if they dominated it.

• So the pallets were a minority of the basket, or they were declared at an

inflated value (plausible, because the sham sales existed to inflate reported revenue). HS-6 cannot tell which.

• DOJ gives a count (2.2 million pallets), not a tonnage, and does not state

the HTS-10 line used.

• The HS-10 split (Census intltrade API) did not respond from this VPS this

tick. Open: HTS 7616.99.51 by month, 2010–2015.

The corporate pipe (the common-ownership tell)

Every hop comes from DOJ C.D. Cal. press releases: the indictment of 31-Jul-2019 (returned under seal 7-May-2019, 24 counts) and the jury verdict of 23-Aug-2021. Entity names are verbatim from DOJ.

Note: the DOJ 2021 release spells it "Perfectus Aluminium"; the 2019 and 2026 releases spell it "Aluminum". Both are quoted as given.

The tell, stated as a rule: the buyer of record and the seller shared a controller, and the buyer had no downstream customers. That is the product-side version of the Youngsun → Thai Unipet pattern in the antimony case. In a heading relabel no transit country is needed, only a captive importer. So the registry to check is the importer's, not a transit exporter's.

Secondary lead: Vietnam (not a traced pipe, figures only)

The 2016 one-year spike of Chinese extrusions into Vietnam is the shape of a stockpile, not of demand. But no ownership hop into Vietnam is traced here. We found no primary registry, court or Commerce record linking the Vietnam volume to Zhongwang, and press accounts are not sources for this series. The Vietnam→US leg is also small in absolute terms. Open questions: a Vietnamese enterprise-registry pull for the 2016 importers of record; any Commerce circumvention inquiry on extrusions via Vietnam. A reported Mexico leg is also untraced. Mexico's Comtrade rows for HS 7604 carry zero net weights in 2014 and 2017 and are not usable.

Transmission chain

China Zhongwang (controlled origin, AD/CVD up to about 400 % per DOJ) → extrusions spot-welded into "pallets", declared as finished goods outside A-570-967/C-570-968 scope → Perfectus (Liu-controlled importer of record) → Southern California warehouses (no sale) → reported as third-party export revenue in HKEX annual reports.

What the Tier-1 score misses. The exposure score reads the order: Chinese extrusions face about 400 % duties, so US exposure to that origin is "closed". Official trade data agreed (HS 7604 from China −95.5 %). Both were wrong for three years. And the harm ran in both directions:

• Risk teams: US buyers competing against duty-free "pallet" metal had a

price benchmark the order said could not exist.

• Investors: Zhongwang's reported US export volume was the fraud vector. A

holder of the equity faced an exposure-identification problem, not a prediction problem. This is the product's lane: "sell exposure ID".

Watch-list use. In the new buyer-side AD markets (Thailand AD1049 and the other extrusion orders in our register), watch three things:

• a controlled HS-4 collapsing while a finished-goods basket from the same

origin rises;

• a basket unit value moving the wrong way;
• an importer of record with no downstream sales.

Method & honesty rails

• Inference framing. The HS-6 table is consistent with the adjudicated

scheme. It does not measure the pallets. The ownership pipe is adjudicated for the six US entities only. China Zhongwang, Liu, Chen and Shao are charged, not convicted, and have not appeared in a US court (per the DOJ 2021 release).

• Single-shipment innocence. HS 761699 carries real finished aluminium goods.

Most of its 2011–13 growth may be genuine.

• Data lags / gaps. Census HTS-10 and monthly data were not retrieved this

tick. Comtrade's USA figures are annual.

• Register-state. The 2011 US orders are not in our register (coverage starts

in 2020). Their absence there is not evidence that they lapsed. Commerce continued them in 2022.

• Sources: DOJ C.D. Cal. press releases of 31-Jul-2019 and 23-Aug-2021; DOJ OPA

press release of 12-May-2026 (US$549.5m FCA settlement); Federal Register 76 FR 30650, 76 FR 30653, 87 FR 66128; UN Comtrade public API.