We Don't Know What Impact China Tariffs Are Having

Searching for $86 billion in missing imports.

Amid Trump’s chaotic style of economic policymaking, there’s been one reliable constant: tariffs on China. Starting in 2018, Trump implemented a string of import taxes on the country, which were largely maintained (and even increased) under the Biden administration. In his second term, Trump has raised the stakes, increasing tariffs on Chinese goods by an additional 20% to date.

Seven years on from the initial tariffs, what impact is the China trade war having? As it turns out, this is a surprisingly difficult question to answer — and not just due to political or economic disagreements.

Owing to substantial discrepancies between reported data from America and China, there’s about $86 billion in ‘missing’ annual trade between the countries. In fact, it’s not even clear whether US imports from China have increased or decreased since the trade war began. Suffice to say, if we can’t tell whether trade with China has gone up or down, then it will be nearly impossible to determine what impact tariffs are truly having on the US economy.

$86 Billion in Missing Trade

All economic data is inherently imperfect, subject to flaws in the underlying collection process. International trade data, however, offers a unique chance to double-check those imperfections. Because there are two sides to every transaction, imports and exports are reported to separate statistical authorities, capturing two independent measurements of the same phenomenon.

When Chinese exporters ship goods to America, their value is initially reported to Chinese authorities. When those goods are accepted by importers in America, their value is independently reported to US authorities. In theory, the value of exports should be exactly equivalent to the value of imports — but that’s not what we find in practice.¹

According to American data, the country’s imports from China have declined by about $44 billion between 2015 and 2024. But Chinese data tells a different story, showing exports to America increasing by $115 billion over the same time frame.

To underscore this discrepancy, we can look at America’s imports from China as a percentage of total imports. Going by American data, tariffs have had a sizeable impact, dropping China’s import share by 8.1 percentage points since 2015. But Chinese data shows a far smaller drop of just 2.1 points.

In 2024, about $86 billion worth of goods went ‘missing’ from the time the exports were shipped from China to when the imports were received in America. Although we should expect some discrepancy due to mistakes, lost shipments, or differences in data collection, this type of gap is too large to be explained by such factors alone.

What’s Driving the Trade Gap?

There’s no single explanation for China’s missing trade, but at least three factors appear to drive much of the difference.² While fraud is part of the story, some of this discrepancy is also rooted in legitimate efforts to optimize trade practices under the current rules.

#1: De minimis exemptions

One of the biggest factors driving the data gap is the de minimis tariff exemption. This rule offers small-value packages a blanket exemption from tariffs and duties; the logic is that collecting tiny amounts of revenue on these shipments isn’t worth the administrative headache.³ In 2016, the de minimis limit was increased to $800 per package.

Although de minimis trade is still reported to Chinese export authorities, the value of these goods isn’t reflected by official US import data. In 2021, however, US authorities estimated that China was responsible for 58% of total de minimis imports by volume. If we assume that China accounted for an equivalent level of de minimis imports by value that year, then this exemption could explain about 40% of the data gap in 2021.

In recent years, usage of the de minimis exemption has exploded. In 2018, the total number of de minimis imports to the US was just 410 million. By 2024, the rise of Chinese e-commerce retailers like Shein and Temu more than tripled that figure to 1.36 billion.⁴

Some of this growth reflects legitimate efforts to leverage the de minimis carve-out. Shein, for instance, limits US customers to purchasing just $800 worth of goods in a single day. There is also evidence, however, that American importers are systematically undervaluing goods to skate under the de minimis limit.

#2: American importers undervaluing goods

Although import undervaluation is a crime, there is ample indication that it occurs. An implausibly high 16% of de minimis goods are valued at $1 or less. In addition, far more imports from China are valued at just under $800 than just over, resulting in a suspicious discontinuity at the de minimis limit.

Of course, American importers have an incentive to fraudulently undervalue all tariffed goods. This is trickier to pull off for large-value imports, however, due to stricter paperwork requirements. Still, there is empirical evidence that broad undervaluation does occur, as the data gap tends to be larger for import categories with higher tariff rates.

On its own, import undervaluation may not necessarily lead to a data gap as long as both sides of the transaction report the same (inaccurate) value. However, just as American importers have an incentive to undervalue goods, Chinese exporters have an incentive to overvalue them.

#3: Chinese exporters overvaluing goods

Chinese overvaluation is motivated by the country’s export tax rebate system. While the details of the system are a bit complicated, the basic idea is that Chinese firms earn a rebate for domestic taxes paid when they export goods.⁵ As a result, exporters can maximize this rebate by overstating the value of their goods.

If Chinese exporters are systematically overvaluing goods, then we should see a positive relationship between the rebate rate and the data gap on specific products. In fact, this is exactly what one study by Chinese researchers found, concluding that overvaluation occurs through both outright misreporting and the misclassification of goods to high-rebate categories.

Moreover, exporters may also overvalue goods as a way of smuggling dollars into China. Although China has strict cross-border capital controls, high export values can provide a ‘legitimate’ reason to bring foreign currency into the country. A 2014 investigation by China’s foreign exchange regulator uncovered at least $10 billion in fake trade relating to this practice.

Conclusion: Mind the Gap

These last two factors — American undervaluation and Chinese overvaluation — likely account for a significant chunk of the data gap. By the very nature of these activities, however, their true scale is hard to quantify. One estimate from Federal Reserve researchers indicated that these factors accounted for about three-quarters of the increase in the data gap between 2017 and 2020.

But this isn’t the end of the story. While the data gap is the most stark evidence that we’re missing the true impact of tariffs, other obfuscating practices don’t appear in official data. One such practice is ‘entrepôt trading,’ which involves shipping goods to an intermediary country before re-exporting those goods to their final destination.

One popular entrepôt trading route is China → Vietnam → America. Technically, under America’s ‘country of origin’ policy, these imports should be labeled as Chinese unless the product undergoes substantial transformation in Vietnam. In practice, what counts as substantial can be fuzzy, and China tariffs have coincided with a noticeable increase in US imports from Vietnam.

Here’s the bottom line: Due to a mix of fraud, loopholes, and grey area practices, we don’t know what impact US tariffs are truly having. Depending on the data you look at, US imports from China could be up, down, or sideways. Thanks to haphazard implementation and poor oversight, America’s tariff policy may be nothing more than an irritating set of rules with little meaningful impact.

1. ‘Exactly equivalent’ is actually a bit of a stretch. Due to factors like the time lag between different years, differences in statistical territory definitions, and the inclusion of transaction costs in import valuations, the two sides will always have some slight discrepancies.

2. There is another relatively well-understood factor driving trade data discrepancies between the US and China: Chinese re-exportation of goods to the US through Hong Kong. China counts these as exports to Hong Kong, but the US counts them as imports from China (under the country of origin policy).

Interestingly, this implies that the data gap is actually bigger in reality since the Hong Kong effect means that China’s reported exports should be even larger. Hong Kong’s re-exports to the US totaled around $37 billion in 2024. If we assume that three-quarters of those goods are from mainland China, that means the data gap should actually be in the range of $114 billion.

3. On February 1st, Trump announced the repeal of de minimis exemptions as part of his broader tariff changes. Just six days later, that repeal was reversed after millions of small packages began piling up in US ports. While the de minimis exemption remains for now, it could well be eliminated in the coming months.

4. The small dollar figure of these packages means that de minimis shipments still make up less than 2% of America’s total import value. For reasons explained in the following section, however, systemic underreporting of de minimis package values could mean that this 2% figure is a significant understatement.

5. A key part of the added complexity here is that the rebate is only a partial refund of domestic taxes paid. Therefore, overstating export values is only part of the fraud — firms also need to overstate the amount of taxes paid. This has fostered a thriving black market for fake invoices showing higher tax payments than actually occurred. Humorously, the paper I cite contains a screenshot of a researcher conversing with a fake invoice dealer and a photo of one such invoice.