A Gap Is Not an Error

Two governments are keeping books on the same trade, and they disagree by nearly $100 billion. From January through July 2026, the U.S. Census Bureau recorded $158.3 billion in merchandise imports from China, including Hong Kong. Over the same seven months, China reported $258.1 billion in merchandise exports to the United States. The $99.8 billion difference is equivalent to 63 percent of what America says it imported, a share nearly five times its average during 2021 through 2024. If the average monthly gap persists, the cumulative gap for 2026 will exceed $170 billion, according to an analysis by Gregory Auclair for the Peterson Institute for International Economics, published on Friday 2 October.

In statistics, a gap like this is usually treated as noise, a footnote about methodology to be reconciled later. That instinct is wrong here. This gap is not an error in the measurement. It is the measurement of something real: the distance between a tariff as announced and a tariff as collected. Every tariff is two things at once, a rate printed in the Federal Register and a tax that must actually be extracted at the port. The $100 billion is, in large part, what falls between those two things.

The scoreboard grades its own exam

Consider what the gap does to the story Washington tells about its own policy. By July 2026, China reported its monthly exports to the United States running about 4 percent below their 2024 average. American data, meanwhile, showed imports from China 26 percent below that baseline. Read naively, the U.S. figures describe a rapid decoupling, exactly what the tariffs were supposed to produce. Auclair warns against that reading: the reported decline in U.S. imports from China, he writes, “should not be treated as a clean measure of either economic decoupling or the effectiveness of US tariffs.” The tariffs manufacture the incentive to evade them, and evasion degrades the data that would tell you whether the tariffs worked. The scoreboard is grading its own exam.

The evasion takes familiar forms. Goods are relabeled as originating in third countries to capture lower duty rates. Vietnam is the photographic negative of the China gap: from January through July, the United States recorded $149.6 billion in imports from Vietnam, while Vietnam reported $105.0 billion in exports to America, a $44.6 billion discrepancy running in the opposite direction. Adjusting both gaps for their 2021–24 averages, Auclair calculates that if Vietnam’s gap were entirely relabeled Chinese goods, it would account for 57 percent of the adjusted U.S.-China gap, though he cautions that aggregate data cannot establish how much relabeling actually occurred. Other shipments are simply undervalued. A New York Times analysis cited by Peterson found that the average value of merchandise per shipping container from China fell by around 40 percent between January 2025 and February 2026, and a Nikkei Asia investigation described allegations that logistics providers used shell companies as importers of record and manipulated invoices to cut the duties owed. Each of these is a rational business decision when the average effective tariff rate on goods from China and Hong Kong was 30 percent in 2025 and 24 percent this year through July, against 6 percent for imports from everywhere else.

An old pattern at a new scale

None of this is unprecedented, which is precisely the point. Federal Reserve researchers estimated that tariff evasion accounted for about $55 billion of the $88 billion shift in the reporting gap between 2018 and 2020, the last time Washington raised duties on China. The mechanism is the same. What has changed is the scale: during 2021 through 2024 the gap averaged roughly 13 percent of U.S. imports from China; it is now 63 percent. Nor can the gap be waved away as an artifact of small-parcel trade. The United States ended duty-free de minimis treatment for Chinese shipments in May 2025 and suspended it globally that August, a change that should have narrowed the discrepancy by pushing low-value goods into ordinary customs reporting. Instead the gap widened. Peterson also finds little sign that Chinese exporters inflating invoices to claim tax rebates explains the recent growth, since Beijing has been cutting those rebates. The timing points in one direction.

The crackdown and its limits

The White House has noticed. In a report released on Thursday 13 August, it estimated that the routing of exports through third countries costs $19 billion to $26 billion a year in lost tariff revenue, the Associated Press reported. The report cited estimates of between $34.2 billion and $303 billion of goods transshipped each year, and Peter Navarro, the White House trade adviser, said China was laundering its exports through more than 40 countries. The day before, Senator Bernie Moreno of Ohio wrote to the Homeland Security secretary, Markwayne Mullin, urging a crackdown and citing trade data showing a record $112 billion gap last year between what China reported exporting to the United States and what Customs and Border Protection recorded as arriving.

The tools are punitive. A July 2025 executive order set a 40 percent duty, in place of the ordinary country rate, on goods Customs finds to have been transshipped to evade duties, and barred any mitigation of the penalties. Navarro has said that imports found to have falsified origins can be tariffed retroactively for roughly a year, and that Customs has begun using artificial intelligence in a prototype program to catch transshipments.

Enforcement, however, is structurally downstream of the incentive. Customs officers see paperwork at the port and must decide, container by container, whether to believe it. Michael Kanko, chief executive of the trade-data firm ImportGenius, which found a gap of more than $100 billion for last year by converting shipment records into container units and comparing value per container, was careful about the cause. “Some of the incentive in this might simply be related to tariff and optimizing for tariff bills,” he told FreightWaves. A tariff is a tax, and a tax is only as good as its collection. The United States set an average effective rate on Chinese goods four to five times the rate on everything else and left collection to the same paperwork at the same ports. The wall leaks.

So what does the $100 billion actually measure? Not Chinese exports as such, and not American imports, which Washington’s data increasingly fail to capture. It measures, above all, the space between policy ambition and administrative reality, the size of the pretense that declaring a duty and collecting it are the same act. Until that space closes, every chart showing the triumph or failure of the trade war is drawn from numbers the trade war itself has distorted. A gap is not an error. It may be the most honest number in the whole debate.

Sources

The trade figures for January through July 2026, the 2021–24 comparison, the projection for the year, the Vietnam figures and the 57 percent calculation, the effective tariff rates, the Federal Reserve estimate, the de minimis history and the remarks on export rebates are from Gregory Auclair’s post for the Peterson Institute; the New York Times and Nikkei Asia findings are as that post describes them and were not read directly. The White House report’s revenue and transshipment estimates and Peter Navarro’s remarks are as the Associated Press reported them, in its story carried by PBS News; the report itself was not read directly. Senator Moreno’s letter was read in the press release on his Senate website. The transshipment penalty is from the text of the executive order of July 2025 on the White House website; whether that penalty has been affected by later court rulings on the president’s tariffs was not established. Michael Kanko’s remark and ImportGenius’s method are from FreightWaves’ summary of its interview with him.