China’s Imaginary Trade Data

The most important part of the IMF’s latest assessment of China is—alas—the appendix on China’s new methodology for calculating China’s trade balance.

It at least explains why China’s balance of payments trade surplus diverges from China’s customs trade surplus, and why the gap started to explode around 2022. As I have noted in the past, China’s data doesn't agree with itself. One measure of the goods deficit is a lot bigger than another measure of the goods surplus.

Appendix VII (way in the back of the staff report) also highlights why the IMF and other guardians of the global economy absolutely should stop using China’s reported current account surplus (falling like a rock, by the way, even as the goods surplus remains near its peak levels) in their assessments going forward.

You might think that a foreign firm producing in China for sale in China (“in China for China” is a thing) would not register in China’s trade data. After all, goods made in China and sold in China never cross a border, and thus should not show up in the customs data.

But in the new balance of payments data, China basically reports a trade deficit with itself because of foreign firms producing in China.

This is because of so-called “factoryless” manufacturing:

“BOP imports kept increasing relative to Customs imports in recent years and surpassed them for the first time in 2023. The upward trend in this ratio is supported by China’s growing domestic market and rising goods produced by factoryless manufacturing.”

Factoryless is, however, a misnomer in this case. The actual factories are in China (they just aren’t owned by a foreign company operating in China) and they are producing under contract for a foreign firm.

This seems to be what China is doing.

If a foreign firm contracts with a Chinese firm to manufacture that foreign firm’s goods in China, and then receives delivery of those goods in China, China counts this as an export. Even if the goods never leave China. The foreign firm is, well, foreign—so the transfer of the good from a Chinese producer to the foreign firm is a transfer to a non-resident (or so the logic goes), though this “trade” presumably settles in Chinese yuan.

That’s all fine and good if the goods are then sold outside of China. It would clearly be an export as it crosses the border when it is shipped abroad and enters into the customs data.

But the strange turn happens if the foreign firm turns around and sells the good that a contract manufacturer produced for it inside China. Such goods are now being counted as an import in the balance of payments data.

Thus, China exports goods to foreign firms operating in China, and then imports those goods back from the same firm even though the goods never leave China. If the goods are sold at a higher price than the contract manufacturer receives, it ends up being reported as a trade deficit in the balance of payments.

This is what the IMF reports:

“Exports and imports arising from factoryless manufacturing seem to have been reducing China’s overall goods trade surplus in BOP. When a Chinese contractor sells produced goods to the nonresident enterprise that outsourced the production, exports of goods are recorded in BOP even if the goods remain in China (e.g., in warehouses). If the nonresident enterprise subsequently sells the goods in China, imports of goods are recorded in BOP. Given that the Chinese contractors’ ex-factory price for the nonresident enterprise (China’s exports) is normally lower than the nonresident enterprises’ wholesale price for Chinese distributors (China’s imports), these transactions result in a deficit in the goods trade balance in BOP”

So Chinese production for the Chinese market by foreign firms is somehow generating a trade deficit in the balance of payments data. This, of course, makes no real economic sense.

It should be noted that if the foreign firm is really charging a huge markup on goods it buys from Chinese contract manufacturers, there should be a profit that already enters in the balance of payments (the profits earned by foreign firms in China are a debit in the investment income balance).

And if China is changing the methodology it uses to calculate its goods balance in the balance of payments so as to move that profit offshore, it should be netting out the profit this firm previously recorded in China. There would likely be a number of adjustments to the services balance as well, as the non-resident foreign firm contracts for storage and retail storefronts and the like.

In other words, there should be offsetting adjustments in other parts of the balance of payments—the aggregate balance should not change.

But that doesn’t appear to be what is happening.

There is a growing divergence between the balance of payments goods data and the customs goods data (and what the balance of payments goods data should look like based on China’s pre-2020 methodology, which moved the cost of insurance and freight out of the goods import bill and into services imports).

That divergence is not offset elsewhere in the current account, though. Rather, the gap between the reported current account surplus and the customs goods data (and the customs goods data adjusted for the tourism/travel deficit) is even bigger than the custom goods to balance of payments goods gap.

A detailed examination of the data shows that—consistent with the work of Adam Wolf—the balance of payments data is diverging from the customs data on both sides of the ledge. Balance of payments exports are way below customs exports, and balance of payments imports are now higher than customs imports when they should be lower (because of the cost insurance and freight adjustment).

These gaps are clear when the reported balance of payments data is compared to an estimate of what the balance of payments data for exports and imports should be based on pre-2020 data patterns.

Moreover, the logic behind the adjustment is difficult to understand. There is no good reason to think foreign firms want to report huge profits in China—and thus are charging huge markups in China. China is not a low jurisdiction, so firms generally do not want to show large Chinese profits. The whole transfer pricing game is usually to move the profit to a low tax jurisdiction with no capital controls, not keep the profit in the firms’ Chinese subsidiary.*

To take yet another step back, the original point of the balance of payments data was to tabulate inflows and outflows of foreign exchange (or gold) to help the central bank keep tabs on the forces driving its reserves.

Production in China for sale in China has no impact on the flow of foreign exchange and thus the “true” balance of payments. It doesn’t impact the central bank’s foreign exchange reserves.

What would impact the central bank’s balance is the repatriation of a profit that a foreign firm generates in China—but the foreign firms profits should be in the balance of payments (in the income balance).**

Bottom line: there is no good reason to think that this adjustment in captures anything important about how China’s economy interacts with the global economy. All this “fake” trade deficit does is reduce China’s reported current account surplus—as the goods surplus in the balance of payments is now about $300 billion (over 1.5 percentage points of China’s GDP) smaller than what it should be in the balance of payments data.*** I personally now estimate the current account surplus to be close to $700 billion even after the drop tied the resumption of tourism in 2023.

The data appendix also reveals why no one has been able to replicate the goods data in the balance of payments using the customs data; the data is all coming from an impossible to verify internal survey and not from an adjustment to the customs data. Paragraph seven of the data appendix notes:

“In order to capture exports and imports of goods that do not physically cross the border comprehensively, the Chinese BOP compilers switched the data sources from Customs data to direct reporting of financial records by large enterprises in 2022. According to the Chinese BOP compilers, more than 13,000 large enterprises directly report goods trade data, which account for about 70 percent of the total goods trade. The compilers use cross-border receipts and payments on goods trade obtained from the International Transactions Reporting System (reports from banks on international payments and receipts) for the rest of the enterprises.”

Given that the reported current account deficit was about $150 billion dollars below what it should have been in the second quarter, this obscure data question now is central to understanding China’s impact on the rest of the global economy.

And my strong view is that the customs data—which captures actual trade—gets the story right, not the marked to model (and internal survey) current account data.****

There is a second issue in the current account: China’s interest income on its foreign assets should have gone up with global interest rates, but it hasn’t.

As a result, China reports a large income deficit when, in my view, it should not—but that is for a future blog.

What matters for now is that a large number of analysts are using China’s current account data to assess China’s impact on the world without realizing that the fall in China’s surplus since 2021 is basically an artifact of difficult-to-justify changes in China’s balance of payments methodology. The real story is found in the old fashioned goods data.

* I would be curious to see, for example, how China accounted for Apple’s in-China sales before and after the change in SAFE’s methodology—across all the balance of payments line items.

** The income line in China’s balance of payments provides absolutely no detail, so there is no way of knowing what is and what isn’t in China’s data.

*** The BoP goods surplus should be larger than the customs trade surplus, as about 5 percent of the reported goods import bill is moved over to the services balance for insurance and freight. The Chinese adjustment literally used to be customs goods imports minus five percent.

**** Back in 2015 when the customs goods surplus was soaring because oil prices fell (China imports a lot of oil) China changed its methodology for calculating the spending of tourists abroad—adding something like $1000 per tourist to its estimated deficit. That adjustment subsequently adjusted and back-dated to 2014 (see Anna Wong’s paper for the Federal Reserve Board). That ancient sage just highlights how China has a history of changing the methodology it uses to calculate the components of the current account at opportune times. The adoption of a new methodology to calculate the goods balance in 2022 is the latest case in point.