What transshipment actually is. Why a White House report named Japan

Hello, this is Ryuta Hamamoto from TIMEWELL.

On 13 August 2026 the White House Office of Trade and Manufacturing Policy published a 25-page paper, The Great Transshipment Scam¹. By the next morning the numbers that travelled were $75 billion a year, 450,000 American jobs, and a list of more than 40 countries that included Japan. Those figures are in the report. They are not what the report treats as settled fact.

What the paper actually does is estimate the scale of tariff evasion through third countries, then describe the enforcement the United States intends to run. The sentence that stayed with me is the one that puts Japan in Tier 1, next to Canada, the EU, India, Israel, Mexico, South Korea, and Taiwan. The sentence that should travel with it is this. Japan does not appear in the table of methods, and it does not appear in the table that pairs foreign corridors with American factory towns. Read those two facts together or you will misread the whole document.

If you already suspect a shipment's paperwork and its physical life are drifting apart, start with the three-minute export-control check. The rest of this column is easier to use once you know where your own holes are.

The numbers left the building first

Reuters led with $19 billion to $26 billion a year in lost tariff revenue, then $75 billion of transshipped goods and 450,000 jobs². Japanese papers put "Japan among 40 countries" in the headline³. Every one of those figures can be found in the report. None of them is presented as a single official count.

The report itself says it reviewed five estimates, two from government and three from the private sector. They run from about $40 billion to $303 billion a year. They cannot be added. They cannot be compared directly¹. The datasets differ. The methods differ. Even the definition of illegal transshipment differs. The $75 billion that became the headline is not the median of the five. It is the midpoint the report adopts as its central case from one private analysis. The "12 trillion yen" that circulated in Japan is not in the English text at all. That is a newspaper conversion.

I am not here to argue that the problem is small. Hidden trade will always come out as a range. The problem is a single point estimate leaving the page without its method. Only people who have seen how the number was built can argue with it.

Transshipment is not "it went through a third country"

Illegal transshipment, as the report defines it, is not the fact of a third-country stop. It is relabeling, repacking, re-invoicing, light processing, and a false country-of-origin claim used to obtain a tariff that would not apply if the goods' true economic origin were declared¹. In plainer words, making the paperwork look like a different country made the thing.

Confuse that with ordinary supply-chain design and you start suspecting work that the rules allow. Shipping a finished machine, shipping parts for assembly at destination, or doing real processing in a third country so that origin actually changes. Manufacturers have done this for as long as tariff schedules have existed. Design that meets the origin rule is not a crime. The line is whether the facts meet the rule, and whether the entry matches the facts. I walked that line in finished goods versus parts.

U.S. customs still turns on substantial transformation. Did the processing create an article with a new name, character, and use. Simple assembly, packing, and a new label generally do not. The report keeps returning to the same test. Can that third-country plant actually do the work it claims. A stamp is not an answer.

In Japanese companies the split is still a language problem. Customs asks how much duty is owed. Export control asks whether the goods may leave. Different desks, different files, same container. The enforcement this report describes does not care that your org chart is split.

Japan is in Tier 1. Japan is not in the method tables

The report groups more than 40 jurisdictions into three tiers. Tier 1, the diversified scale leaders, is Canada, the EU, India, Israel, Japan, Mexico, South Korea, and Taiwan. Tier 2 is the large economies more tightly tied into China-linked production. Tier 3 is the smaller, opportunistic waypoints¹. The prose also says China's biggest "enablers" run from Mexico and Canada on the land border to the EU, India, Japan, and South Korea.

That word is the report's. It is not a finding that Japanese firms cheated. A list in an analytical paper is a regulatory or analytical bucket, not a verdict on a company. The report's own reason for putting Japan in Tier 1 is structural. China-linked volumes are large. Japan is a major platform for U.S.-bound exports. Risk, the paper says, is embedded in otherwise legitimate trade flows.

Look at the other two tables and the point sharpens. The functional-architecture table names Cambodian microhubs, a Polish and Czech processing belt, Jebel Ali, Port Klang. The "ugly sister city" table pairs Guanajuato–Querétaro with Detroit on electric motors, and Gyeonggi with Phoenix and Austin on integrated circuits. No Japanese city appears in either list.

For Japanese readers that is bad news and usable news at the same time. Scale keeps Japan on the list. There is no logic that removes a large, diversified exporter. It is also not, yet, a paper that has pinned a method to a Japanese corridor. There is still room to explain. South Korea appearing in the city table and Japan not appearing are two facts in the same document. Neither one is a moral judgment on every firm in those countries.

$75 billion and 450,000 jobs are the output of the central case

The five estimates sit like this.

Source	Estimate	What it measures
Goldman Sachs	$40 billion	Top-down product-flow model (2023)
CEA	$60 billion	Rounded midpoint of $34.2–89.6 billion
Private supply-chain screen	$75 billion	Midpoint of $51.1 billion directly screened and a ~$100 billion ceiling
Commerce OTEA	$109 billion	Trade-transfer benchmark, 459 HS6 codes
Private facility-level map	$303 billion	Broad upper bound

The report says do not add these rows¹. The $75 billion in the headlines is the midpoint between $51.1 billion of shipments a private model flagged and a wider systemic ceiling. The number I find hardest to dismiss is OTEA's stricter cut. Same quarter, same region, the same HS8 product imported from China and re-exported to the United States, pulled from transaction-level customs data. Even limited to Mexico, India, and Vietnam, that cut still produces $67 billion in 2025 and $28 billion in lost tariff revenue. That is more than 60 percent of the broader $109 billion benchmark.

The 450,000 jobs are simpler arithmetic. The report uses a rule of thumb. Six thousand U.S. jobs displaced per $1 billion of trade deficit. The sources are EPI papers on NAFTA (about 5,300) and on U.S.–China trade (about 11,000). The authors took 6,000 on purpose, they say, as a conservative choice rather than a midpoint. Seventy-five times six thousand is 450,000. That is the whole construction. The narrow $40 billion case becomes 240,000. The $303 billion case becomes 1.82 million. The report prints all three.

It also says, in plain English, that these are model-based estimates, not observed job counts. Apply the same method to tariff revenue and you get the $19 billion to $26 billion that Reuters ran. That is the central case times illustrative 25 percent and 35 percent duty gaps. Anyone who quotes 450,000 as "jobs China took through Japan" is not reading the paper. They are reading the output of an assumption.

The body of the report is enforcement

After the tables the paper becomes an enforcement memo. That is the part a compliance team should keep.

Executive Order 14411, signed 3 June 2026, is the legal spine⁴. It tightens who may serve as importer of record, raises bonds and domestic-asset floors, demands ownership and affiliation disclosures, and makes "good standing" a condition of importing. The report names the targets. Shell importers. Foreign importers beyond easy U.S. reach. Under-bonded entries. Opaque ownership. Repeat violators. For a Japanese group the sting is the shift of weight onto the U.S. importer. If a U.S. subsidiary or a sales agent is the importer of record, the extra bond, the ownership file, and the penalty land inside the group.

Origin rules are the piece most readers skip. The report says today's origin test sits in customs case law, not in a clear statute, so application is uneven and entries can meet the letter of the law while missing its spirit. It asks Congress to amend and codify the standard. If that happens, substantial transformation itself gets rewritten. A third-country plant that passes today may not pass next year.

Some of the enforcement is already visible. CBP compared two 526-day windows around the inauguration. Shipments flagged with post-release discrepancies rose from 93,744 to 323,677, up 245 percent. Related revenue assessments rose from $9.6 billion to $25.8 billion, up 169 percent¹. Those are operational counts, not models.

The next layer is what the report calls an AI "Detective Border." Declared origin, routing history, and bill of materials against an expected pattern. Computer vision on container marks, packing, and X-ray images. The aim, in the report's phrasing, is to make a product's digital identity match its physical body. The 90-day dwell threshold that one private method uses belongs to the same family of tests. Goods that pass through in under 90 days look like transit, not manufacturing. Fast, legitimate inventory will trip the same wire. The rebuttal file has to exist before the hold, not after.

The most honest paragraph is in the conclusion. It is too early to know the net effect of the Administration's tariffs and anti-transshipment measures. Trade and customs data arrive late. Several provisions of the June order are still being built. The tariff gap can raise the incentive to cheat. The new tools can raise the cost of cheating. Which force wins cannot be measured yet. The paper, it says, is a framework for that later assessment¹. For a policy document, that is unusually plain. The version that will travel the world will still be $75 billion, 450,000 jobs, and "Japan among 40 countries." The reason to open the PDF is the gap between those three phrases and the method underneath.

Tariff evasion and technology diversion run on the same pipes

This last section is my view, not the report's.

The shadow network the paper describes, free zones, bonded warehouses, re-export desks, swapped invoices, swapped names, is infrastructure for tariff arbitrage. It is also, physically, infrastructure for export-control and sanctions evasion. A third country that blurs origin, end user, and beneficial owner can carry a dutiable toaster or a controlled tool. If the United States pours data into catching the toaster, the resolution on the tool goes up with it.

In a Japanese company that means three jobs that used to live on three spreadsheets become one question. Customs and origin. Classification. Beneficial owners and the 50 percent rule. Washington is already writing them as one enforcement story. On 10 November 2026 the BIS Affiliates Rule is scheduled to return⁵. Tariff diversion and technology diversion will arrive as the same sentence. Does this cargo match its papers.

TRAFEED is the export-control agent we built so classification and counterparty checks can run in one place. In a joint study with Okayama University, on about 30,000 past review records, we measured AI classification accuracy above 95 percent (our own study). The system is covered by Japanese patent 7862062. The final classification still belongs to your export-control officer. What the report is asking for is not a perfect model. It is a file that lets a person explain the gap between paper and metal.

If I had to pick Monday morning's list, it would be three items. Match third-country capacity to the origin story you tell. Know who the U.S. importer of record is, and whether that entity can survive a bond and ownership request. Stop keeping origin, classification, and ownership in separate Excel books. The companies that already hold the explanation are the ones that can keep a polite distance from a $75 billion headline.

If you want to walk a specific plant or a specific bill of lading, talk to the TRAFEED team. Do not start with the headline. Start with whether your papers and your cargo still tell the same story.

Sources

Footnotes

1. The White House, Office of Trade and Manufacturing Policy, The Great Transshipment Scam, August 2026. https://www.whitehouse.gov/wp-content/uploads/2026/08/The-Great-Transshipment-Scam.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷

2. Reuters, "Trump White House says it's losing $19B-$26 billion a year in revenue as countries dodge tariffs," 13 August 2026. https://www.pbs.org/newshour/politics/trump-white-house-says-its-losing-19b-26-billion-a-year-in-revenue-as-countries-dodge-tariffs ↩

3. Chunichi Shimbun, "関税逃れの迂回先40カ国 米報告書、日本含む," 13 August 2026. https://www.chunichi.co.jp/article/1296079 ↩

4. The White House, Executive Order 14411, Strengthening Customs Enforcement, 3 June 2026. https://www.whitehouse.gov/presidential-actions/2026/06/strengthening-customs-enforcement/ ↩

5. Bureau of Industry and Security, "One-Year Suspension of Expansion of End-User Controls for Affiliates of Certain Listed Entities," Federal Register, 12 November 2025. https://www.federalregister.gov/documents/2025/11/12/2025-19846/one-year-suspension-of-expansion-of-end-user-controls-for-affiliates-of-certain-listed-entities ↩