Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Glenn Profile picture
@GlennLuk
Oct 29, 2024 • 45 tweets • 15 min read • Read on X
Contrary to a popular mainstream view, China's trade surplus has likely been overstated by hundreds of billions per year dating back more than a dozen years.

This 🧵 is a direct rebuttal against claims that the recent changes in BoP methodology are leading to systematic understatement of its trade and current account surplus.

Instead, I show how the change in methodology has addressed prior distortions.

@IMFNews @Brad_Setser
To understand why, we need to go back to the the 19th century during the first globalization boom.

This was a time when physical trade flows largely matched funds & value flow, as products produced entirely in one region were traded for those produced entirely in others. Image
Fast forward to present day: Trade has gotten significantly more complex.

Lower friction costs of trade made it possible to separate/outsource different segments of the mfg. value chain.

Int'l tax laws incentived firms to shift profits, which distorts customs data.Image
This is especially true for China, which became the world's factory floor — with the implication that China is the final destination for most of the world's manufactured consumer goods by global brands before they are shipped all around the world. Image
Rising mfg. complexity & profit-shifting increasingly untether measured physical trade flows from underlying value/fund flows.

The iPhone is one of the best illustrative studies.

This diagram shows how China Customs measures it using physical trade flows. Image
But if we examine actual fund flows — which is the ultimate objective in measuring trade — China switches from being counted as a large net exporter ($31.2B) of iPhones to a larger importer ($21.6B).

This is a $52.8B swing/distortion for a single product from one company.Image
Ultimately the direct funds flow approach is more accurate than the physical flows approach.

After all, China Customs' measurement of physical trade flows had merely been used as a proxy for the underlying fund flows & value transfers. Image
We can further break down the physical vs. fund flows distortions into two key categories.

First, is the overstatement of exports using physical vs. fund flows.

This is driven by the difference b/n customs valuation and actual payments to the contract manufacturer.Image
While there is some debate on the exact magnitude of the difference between customs value and payments to the contract manufacturer — as it will vary by industry and product — we know that it can only be distorted in one direction (i.e. overstatement).
Customs value cannot go below amounts paid to the contract manufacturer, but it can be much higher.

This was confirmed recently when analyzing differences in custom valuations between the U.S., Japan and Ireland for the iPhone.

Secondly, there is another key distortion on the measurement of imports between physical and funds flows approaches.

This time it is an understatement — which leads to further overstatement of the surplus.Image
As mentioned earlier, combined these distortions add up to an estimated $52.8B just for the iPhone.

And these distortions are not Apple-specific or limited to bonded zones.

They are features of most relationships between foreign brands and Chinese contract factories. Image
For example, we see the same phenomenon with global footwear brands like Nike and Adidas that manufacture shoes in China for customers both outside and within China.

Within branded footwear, I calculate a net overstatement of $27.5B.

Less than the iPhone, but still material. Image
Image
Indeed, we can generally apply this analysis across all sectors where foreign brands contract manufacturing to Chinese factories. Image
The scope of this net overstatement is large.

I estimate distortions equal to 3.2-4.7% of exports (as measured by China Customs) and 1.1-1.6% of imports, amounting to a combined $142 to 212 billion in '22 — and even this may be quite conservative.Image
Keeping all of this context in mind, the change in methodology in 2021-22 from using China Customs physical flows data to underlying funds flow data kept by SAFE (e.g. FX transactions and reported financials) was most likely to correct for these known distortions. Image
Correcting these known, observable distortions is a far more plausible explanation than speculative claims that the methodological changes were made as a way to understate or otherwise hide its trade surplus. Image
Indeed, as I discussed in my initial thread on the topics a few weeks ago, there are no signs of a the "hidden capital flight" that would be a necessary balancing implication of a current account surplus that is understated by half a trillion dollars.

Instead, this corrects longstanding physical vs. fund flow distortions that have been steadily rising from:

(i) ↗️ exports
(ii) ↗️ use of Chinese contract mfgs, and
(iii)↗️ sophistication of int'l tax optimization strategies via profit shifting to tax havens like Ireland
Indeed, if we apply general assumptions on distortion levels to BoP figures dating back to 2012, we can see how they adjust E&O in a way that is still consistent with known "hidden capital flight" surges like 2015-18 and 2021-22. Image
Further analysis will be helpful to further refine these assumptions but the general conclusion here is that it is far more likely that Chinese trade surpluses have been overstated rather than understated over the past dozen years.
Responses to Brad to follow:
1. U.S. customs price shows material diff b/n reported customs valuation ($400/iPhone) + iPhone BOM ($270/iPhone), creating the distortion.

The chart below actually confirms the idea raised in CF40 report that Chine’s reported customs exports match trading partners’ import data.
I go into detail in this prior thread to calculate the estimated difference between customs valuations & average iPhone BOM.

Brad keeps citing BOMs for high-end iPhones but the avg. iPhone is $885/retail, suggesting a significantly lower avg. BOM vs. high-end models.
Ireland’s much lower customs import price for iPhones strong indication of what is paid to contract manufacturers.

Ireland is a special case because IP is legally domiciled there for international sales.

(U.S. is also special because IP is developed in Cupertino)
2. I do not see the basis for the assertion that customs distortions only arise out of bonded zones.

China and IMF refer to “factoryless” manufacturing, which is much larger than trade our of bonded zones.
I specifically addressed this in my thread here using the example of branded footwear.

Thus Apple and bonded zones cannot represent an “upper bound” on possible export distortions.
3. I also addressed the import understatement issue in my thread.
The specific critique is with Brad’s proposed adjustments.

I agree it would be accounted for in the old methodology as a separate services, royalty or profit line item.
But in Brad’s adjustment, he compares the Customs surplus (which does not include this adjustment) with the new BoP surplus methodology based on fund flows (which does).

This is not apples to apples.
Thus, adding the full gap between Customs data and (post-adjustment) BoP Surplus double-counts this import understatement adjustment.

My position is that the new methodology properly accounts for the export overstatement and this import understatement.
4. If there are valid reasons, such as the rise of these identified distortions, then making the move to the new methodology that ends up reduces E&O should provide comfort, not arouse suspicion.
E&O could be a function of distorted official measures (such as the customs exports overstatement) or it could be a sign of hidden capital flight.

What doesn’t pass the sanity check are logical implications of Brad’s position that China’s CA surplus should be adjusted by $500B.
I discussed why here. There’s just no smoke / evidence that there is hidden capital flight anywhere close to the $500B that would be required to balance out the proposed adjustment.

I’m happy to change my position if evidence can be found of this, but none has been offered.
I have compiled customs data from OEC for 2022 for some of the larger European countries to fill in some of the remaining gaps.

Est. customs value for iPhone is significantly higher than U.S. (as expected) but also higher than Japan. Image
This raises the $/unit customs average to $457. I have used this data to update the previous slide: it now covers >90% of exported iPhones from China.

I've tweaked some of the assumptions (BOP as a % of retail, iPhone/Android price ratio) to be even more conservative. Image
Updated export overstatement of $27B representing ~32% of the customs export value.

Compares to 10-15% on foreign brand exports, which are ~1/3rd of total exports. This is appropriately conservative, as an iPhone has relatively high intangible content.

This '21 paper examines the "factoryless" mfg. and distortion effect on "trade vs. income flows".

Based on five US "factoryless" cos (Apple, Nike, Qualcomm, Cisco, AMD), researchers found $70B (~36%) of trade distortion on $190B in int'l revenue.

ideas.repec.org/p/ngi/dpaper/2…Image
Image
Image
This is consistent with the up-to-date analysis I have done on the Apple iPhone, providing a nice sanity check.

This distortion is generally applicable to all "factoryless manufacturing" where a foreign company works with a Chinese contract manufacturer.

This includes most of the $1.1T in exports that include a "foreign-funded" exporter (e.g. Foxconn, Pou Chen)

In other words, it is not just "bonded zones", which are a subset of "factoryless manufacturing" for products like the iPhone where there are large number of finished components that need to be handled logistically.

To calculate the range of potential distortions, I use 10-15% of this $1.1T in exports by foreign-funded entities.

This is actually quite conservative based on the Apple and Nike examples, where intangible asset (e.g. brand, tech) make up the majority of the value.
The value recorded by China Customs for exports is typically based on the Transaction Value method and determined by the importing firm, not by China Customs or the contract manufacturer.

bdo.com/insights/tax/t…
As noted here, intangible value like IP and royalty license should be included in this valuation.

The customs valuation is relevant for the importer because that is the value on which potential duties and VAT are calculated by the importing country. Image
Image
The Transaction Value (TxV) method is the dominant form to estimate customs valuation, used in “90-95%” of transactions.

It’s discrepancies between TxV — determined by the importer — recorded by China Customs and the price paid to the CM that create export overstatement.

• • •

Missing some Tweet in this thread? You can try to force a refresh
　

Keep Current with Glenn

Glenn Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @GlennLuk

Glenn Profile picture
Jun 21
Others have now raised this topic a few times, so allow me to share some thoughts on the BYD (and broader) supply chain financing story:

1⃣ BYD's high payables number actually reflects the strength of its underlying business model and market dominance for two key reasons (ability to extract favorable supplier terms; how that number is driven in part by rapid expansion in production capacity)

2⃣ Establishing industry norms that forces larger players like BYD to adhere to standard payment terms (voluntarily or involuntarily) is a positive step forward for the whole industry, leading to more efficient overall financing approach.

3⃣ BYD and other market leaders that also run large negative working capital balances are generally not a risk of insolvency by adhering to new industry norms as they are generally under-leveraged (with traditional debt financing) and will simply plug the financing hole with more traditional debt and equity financing. In BYD's case, I expect all or most of it to be to replaced with debt (long-term bonds).https://www.ft.com/content/e6ae000d-d506-4a21-898e-213002234ee2
1⃣ BYD's high payables figure reflects strength of its underlying business model and is in part a reflection of its rapid growth in production capacity

While the high payables figure has been portrayed as a potential weakness (with some even raising the idea that BYD is insolvent), actually it reflects the opposite.
BYD uses its scale to extract favorable terms from its suppliers. It trades volume for pricing as well as non-pricing advantages, like extended payment terms. It does this because that's what extremely competitive companies do: they try to exploit every advantage they have over the competition.

As BYD has only gotten bigger and more powerful, it has maintained its ability to sustain structural negative working capital state on its balance sheet.

Companies that can maintain negative working capital are often extremely competitive. This is a very desirable business model to run for rapidly growing companies because as revenue grows, working capital becomes a source of funding.

Amazon's marketplace business was an example of this. Amazon collects payment upfront and then pays out sellers later. This leads to a negative working capital balance, which is effectively a very low-cost form of growth financing for its marketplace business.

Ability to maintain negative working capital is even more rare in a capital-intensive businesses like the car sector. That reflects just how dominant BYD has become.

This doesn't mean it's a good thing for the industry overall (and I'll touch on this in the next point), but it does reflect on the increasing dominance of BYD individually.
Read 13 tweets
Glenn Profile picture
Jun 19
People have a tendency to compress complex, multi-decade stories into simple narratives that follow cause-and-effect storylines, often ones that tie into pre-existing narratives. This creates the risk of dangerous over-simplification.

In this case, the prevailing narrative goes something like this:
▪️ "China failed to build a competitive auto industry for decades."
▪️ "Then Tesla entered the market and became the magic fix that enabled China to develop a globally competitive car industry."
▪️ "Therefore, we should apply the same magic fix to our own industry."

In my view, this is a dangerous over-simplification. Reducing the story to a simple cause-and-effect narrative often leads to blissfully naive solution sets that fail to address the core issue: how do we re-industrialize America?

Believing that simply inviting Chinese car companies into the U.S. will serve as a "magic fix" — just as Tesla supposedly was for China — misses the mark, for two key reasons:

1. The "magic fix" narrative is a gross oversimplification of five decades of development in China's auto and broader industrial/manufacturing sectors.

2. The fundamental challenges China faced over those decades are very different from the ones the U.S. faces today.

None of this is to say that inviting Chinese automakers to invest FDI in the U.S. cannot be part of a LT solution**. But it must be done thoughtfully — and only in tandem with addressing core domestic issues — if the goal is truly to re-industrialize this country in a meaningful way.

** Of course, all of this assumes they even find the risk/reward decision to commit long-term capital to the U.S. in today’s geopolitical climate remotely attractive compared to FDI opportunities elsewhere.
1⃣ First, let me go through several points that were brought up in the excerpted sections of the interview as well as the post to show how reality was much more complex than presented**

** I full interview is not out and I haven't seen it, so perhaps there will be more nuance there; this is mainly a reaction to how the narrative on the rise of China's auto industry has been grossly oversimplified and in certain cases, simply wrong.
"The impact was brutal. When Tesla's Model 3 launched in 2020, it quickly became China's best-selling EV. BYD's total vehicle sales actually fell 7.7% that year to just 427,000 units."

This excerpt suggests that Tesla's market entry in China was the direct cause-and-effect reason why BYD's sales declined in 2020.

This is wrong. It may have played a minor role, but there were many other reasons why BYD's vehicle sales declined.
Read 35 tweets
Glenn Profile picture
May 31
This popped up on my timeline and was just a reminder of some of the sillier narrative framing of Chinese EVs just a year and a half ago.
Since this post in January 2024, Chinese NEV production has increased from a ~10-11 million run rate to ~>16 million as of mid-2025 and virtually ALL of the increase has been absorbed by the Chinese market …
… as EV exports have been more or less flat.

This means Chinese households are buying EVs as fast as they can be produced in the factories.
Read 8 tweets
Glenn Profile picture
May 31
Chinese chip designers finding workarounds for EDA software is roughly the same degree of difficulty / switching cost as moving away from CUDA/nVidia.

This is an effort measured in months, not years.

eetimes.com/u-s-restricts-…
By comparison the advanced lithography ban was an OoM more complex/difficult and the switching cost process measured in years, not months.

That was the largest source of leverage in the American tech/economic dominance toolkit and it was played early.
The other comp here is developing a homegrown OS and attracting a developer base.

Replicating EDA software and supporting libraries is the same order-of-magnitude task.

The development takes years, but this effort was also started years ago, with real efforts kicked off after the Meng arrest in 2018.
Read 11 tweets
Glenn Profile picture
May 29
Can we please stop with this fiction that there are "only 2-3 profitable Chinese EV companies"?

I count at least 8 profitable NEV operations + CATL/Huawei. And it is the only market that is close to profitable selling NEVs at the sectorwide level after accounting for subsidies.
In 2023, the first year after Beijing ended buyer rebates, China's car sector sold an ~9.5M NEVs generating revenue of ~$233B ($24.6k ASP).

Sectorwide gross margin was ~21% (~14% ex-subsidies) with operating profit of ~1.4B (negative ~$21B after subsidies).Image
Note that comparable figures in the U.S. in 2023 were:

▪️ $66B in sector-wide revenue
▪️ $2.6B (4%) in gross profit (negative $13B, or -19%, after subsidies)
▪️ -$27B in operating profit (negative $42B after subsidies)

There was only one profitable operation before subsidies: Tesla. And after subsidies, Tesla's U.S. car operation was not profitable.

Indeed, I suspect the only solidly profitable segment of Tesla's car operations today (ex-subsidies) are its exports out of Gigafactory Shanghai.
Read 27 tweets
Glenn Profile picture
May 14
Saying Apple "invests $55B in China every year" is financially illiterate nonsense.

We know exactly how much Apple has invested in China, as it discloses annually in its annual 10-K.

Apple has cumulative investment in "Greater China" (includes HK/TWN) of $4.8B as of 9/2024.Image
Included in this $4.8B balance sheet figure are leasehold improvements on Apple Stores, "inventoy prepayments" and owned "capital assets at its suppliers' faciliities" like molds and specialized equipment sitting in Foxconn's factories. Image
Image
Again, the $4.8B represents the cumulative aggregate of long-lived assets that Apple has in China, Hong Kong and Taiwan.

And notably, Apple has been liquidating (a.k.a. converting to cash and repatriating) this tangible asset base.

This number peaked at ~$13B in 2019.Image
Read 13 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

Send Email!

:(