August China Export Over 20%? Squeezing Water Out of CCP Exports! August China Export Increase Only 3% to 4%. Export Structure Clearly Visible, Understanding the Latest Trends of China's Only Growth Engine!

This episode provides a detailed analysis of China's export data released by customs in 2026. Through underlying economic logic such as population mobility, corporate working hours, and labor productivity, the host "squeezes the water" out of the 20% export growth reported on paper, pointing out that the actual growth in real goods exports is only between 3% and 4%. Most of the increase stems from rising chip prices, exchange rate conversions, bonded transit, and overseas inventory, rather than being driven by increased employment or domestic labor productivity.

Video link:

Full text:

Observe things as they are, and reject fragmented cognition. Last week, the General Administration of Customs of the CCP released the foreign trade data for August. Exports, calculated in US dollars, rose by 25% year-on-year, and the cumulative growth from January to August reached 19.3%. Then, they started bragging again, saying that "Made in China" is carrying the burden of exports. But just three weeks ago, the CCP's National Bureau of Statistics released the age-specific unemployment rate for July, and the youth unemployment rate (excluding students) jumped by three percentage points in a single month. Looking back further, the total employed population in the country in 2025 was 9.35 million less than in 2024—it is shrinking. Yet exports are skyrocketing. How so? Has China's labor productivity taken off like this in three years? It is just a common sense question. How many goods a country can produce, from a micro perspective, is what? Isn't it equal to how many people are working, multiplied by how long each person works, and then multiplied by how many goods each person can make in an hour? All three of these can be checked. After checking, we know one thing: how much of this 20% export growth on paper is actual goods, and how much is something else. This is not a very profound question. Exports are the only engine in the CCP's narrative this year that is still growing at a double-digit rate. Look, consumption has completely fizzled out; social retail sales only grew by one percentage point, and fixed asset investment and real estate investment are in negative growth. If this pillar is also hollow, then the Chinese economy has no pillars left. We will start from people to goods, and then from goods to the entire economy. Today, we will first open the ledger on people and take a thorough look.

First, let's talk about the statistical definitions. In the statistics of the Chinese Communist Party (CCP), there are two types of population: one is the 'resident population,' which includes anyone who has lived in a city for at least six months. The other is the 'non-local resident population,' which, to put it plainly, refers to migrant workers who do not have a local household registration (hukou) and are renting housing there. Guangzhou and Shenzhen no longer provide this figure directly. However, by subtracting the registered population from the resident population, you can calculate this number. Once you separate these two figures, the picture looks completely different. From the end of 2021 to the most recent announcement, the combined non-local resident population of Beijing, Shanghai, Guangzhou, and Shenzhen has decreased by 1.67 million. So why is the resident population still growing? It is mainly due to the granting of household registrations. The resident population is the headline figure in local statistical bulletins and serves as a basis for land quotas and transfer payments. Therefore, every major city is competing to attract people to settle down. What is the essence of this competition? It is to prop up the resident population figure. But who are the people being brought in? They are mainly young people with degrees. Who is leaving? They are the migrant workers without household registrations. So these two groups are moving in opposite directions. Let's look at Shanghai's numbers. In 2025, Shanghai's registered resident population increased by 141,400, while the non-local resident population decreased by 89,900. If you subtract them, 141,400 minus 89,900 equals 51,500. And the resident population growth written in Shanghai's statistical bulletin is exactly 51,500, not a single digit off. Shenzhen is the same. In the first half of 2026, the number of young people under 35 with college degrees or higher who came to Shenzhen for the first time and enrolled in social insurance was 304,000. However, the Shenzhen Human Resources and Social Security Bureau reported that the total new employment in the three major industries—integrated circuits, low-altitude economy, and artificial intelligence—was only 54,800 in the first half of this year. In other words, the people coming in are going through the channels of social insurance enrollment and household registration, but the new job positions are only a fraction of that. Therefore, from now on, I only recognize the non-local resident population figure; the resident population figure is no longer evidence to me.

So where are the migrant workers retreating to? The 2025 Migrant Worker Monitoring Report issued by the National Bureau of Statistics of the CCP provides this direction. Its data composition is as follows: the number of migrant workers moving across provinces decreased by 750,000 in one year, while those moving within provinces increased by 2.1 million. This means people are retreating within their provinces and toward provincial capitals; they are no longer heading to first-tier cities. For those who retreated, there used to be two main fallback jobs: one was driving for ride-hailing apps, and the other was food delivery. The characteristics of these two paths are that they don't require specific degrees or age limits, and you can start working on the same day. Therefore, those laid off in every round of downsizing eventually end up here. But starting this year, these two paths have also become extremely congested. The number of delivery riders nationwide is now approaching 20 million. However, based on the current national volume of 110 million delivery orders per day, only 4 million delivery riders are needed. What does this mean? It means the level of congestion in the industry is already five times the normal labor demand. As for ride-hailing, it is even more congested. Even the Shenzhen Transport Bureau has been issuing saturation warning reports for ride-hailing for more than two consecutive years. What is the situation for ride-hailing now? The situation is that a full-time driver works more than ten hours a day, and after deducting commissions, fuel or electricity costs, and depreciation, the net profit per hour is only around 10 yuan, and this market has no growth at all. What does that mean? It means that even if one more person comes in to drive, the total revenue of the system will not increase by a single cent; it just pushes the hourly wages of the existing people down another notch. So now, whether it's driving or delivering food, these jobs themselves have become another form of unemployment. Because you aren't making any money doing this, which is strange.

Hey, with fewer people, who does the work? What was the answer before 2025? Overtime. The CCP's statistics bureau has a figure often overlooked, called the average weekly working hours of enterprise employees. In 2018, the average weekly labor time was 46 hours, and it grew year by year, reaching 49 hours by 2023, the highest in the past twenty years, and still around 49 hours at the beginning of 2024. What was the pattern during that period? Jobs were decreasing, but the remaining workers had to do more and more. However, starting this year, the curve began to turn. In the first five months of 2026, the weekly average length stayed just over 48 hours, several months marking a new low compared to the same period four years ago. What was July's figure? 48.2 hours, about 1% less than the 48.6 hours in the same period of 2025. In other words, the nine-year continuous rise has ended. What explanation does the CCP give? Some nonsense about anti‑involution, referencing the EU's forced labor law—whatever, believe it if you want. Look at Chinese companies' operating logic, and you can refute the CCP's weak explanation, because it is completely opposite to corporate labor logic. The continuous change in working hours shows that reductions always start with cutting overtime, then halting hiring. When orders are insufficient, what does the boss stop first? Overtime, because overtime pay is the easiest cost to cut. Next comes stopping recruitment, letting natural attrition lower headcount, and finally layoffs, which require compensation and local government pressure—so this gets delayed. Hence the CCP's claim is fundamentally invalid.

By 2026, they have reached the stage of stopping recruitment. The Ministry of Human Resources and Social Security's urban new employment figures for the first half of the year were flat compared to the same period in 2025, but the 2026 college graduate cohort was 480,000 larger than the previous year. Yet the new jobs did not increase alongside the number of graduates. This halt in hiring is the most interesting part. Look at the data: in 2026, labor input—both headcount and hours—declined together, with the workforce down about 1% and hours down about 1%, meaning total labor input fell by roughly 2%. Hours are a harder metric than employment numbers because employment can be inflated by insurance adjustments and statistical tweaks, but how are hours calculated? Hours are gathered by sampling individuals and asking how many hours they worked per week—there's no way to fabricate this, and no department has an incentive to underreport, so I treat the decline in hours as the most reliable labor market reading.

Thus we see a paradox: headcount is falling, hours are falling, yet the volume of physical exports must rise, leaving only per‑capita output—labor productivity—as the potentially growing factor. Could labor productivity suddenly jump in 2026? First, look at the trend: from 2019 to 2024, the average annual growth of labor productivity across the whole workforce was about 5%, which is the CCP's bragging figure. How can productivity grow that fast? For comparison, the United States sees about a 2% annual increase, even in good years. Even Taiwan, whose economy is centered on chips and semiconductors and is currently in an AI capital‑expenditure cycle, only sees a 3‑4% labor productivity rise. What does a 5% rise mean? Is it that Chinese workers are taking drugs before work? Who can explain this? Given China's current economic structure, technological base, and environment, why would productivity jump 5%? Even if this is a clear exaggeration, we’ll still use it. Continuing, if we divide GDP by total employment for 2025, it rises by 6.4%, a figure inflated by a shrinking denominator.

What did China's export data say? Priced in RMB, it increased by 14.6%, right? But labor input decreased by 2%. If the growth data in this part is entirely true, then the export sector is abnormal. What would that imply? It would mean that the per capita output of the export sector increased by 16% annually, which is three times the average growth trend of China's abnormal 5%. Does such a thing exist in the world? Well, there is one example, though not strictly accurate, it's Taiwan's TSMC. From 2018 to 2024, TSMC's employee numbers increased by 70%, and its revenue nearly doubled in the same period, with per capita output increasing by 60%. But pay attention, when TSMC's per capita output increased, the number of employees, wages, electricity consumption, and capital expenditure all increased together. The average salary increased by more than 40% in over four years, and this single company consumed 8% of Taiwan's total electricity. So why would these necessarily increase together? Because real capacity expansion requires building new factories, which requires people, running machines requires electricity, buying equipment requires capital expenditure, and competing for engineers requires raising salaries. If any one of these four is missing, a new factory cannot operate. Therefore, if it were a true narrative of export explosion in high-tech industries, even with low employment elasticity, it would still leave records of electricity consumption, wages, employment, and capital expenditure simultaneously; not a single one could be hidden.

However, South Korea serves as a counterexample. Why do I say that? From January to August this year, South Korea's exports increased by 82.6%, with semiconductors alone increasing by 270%. But South Korea's manufacturing employment has been decreasing for more than two consecutive years, and employment has been decreasing for four consecutive years this year. In South Korea's input-output tables, every billion Korean won of output in semiconductors drives only 2.4 jobs, while the average for manufacturing is 5.1 jobs. So why haven't South Korea's employment numbers increased? Because once a wafer production line is built, its annual output is fixed. Even if the price doubles, the number of wafers produced remains the same. The additional output goes into the profit statement, not the payroll. Therefore, South Korea doesn't need to hire any extra people for its production lines. So what do people see? South Korean chip workers have been protesting low wages and unequal distribution, demanding that employers increase wages and share the dividends. So what is South Korea's pressure? This is clearly an export growth driven by the price increase of memory chips. Therefore, from an employment perspective, it is imbalanced. But South Korea is the leading country in memory chip exports, and a large part of its industrial structure relies on this. China's proportion is clearly not comparable to South Korea's. Therefore, this part of South Korea does not involve people, and the foreign-funded wafer plants within China, such as SK Hynix and Samsung, also do not involve people. So what are the records for China in 2026 regarding these four aspects? Employment can be said to be completely the opposite. Where can we see this? The employment sub-item of the manufacturing Purchasing Managers' Index is 48.7, still in the contraction zone. Wages are flat. The average monthly income of migrant workers in the second quarter was only 5,118 yuan. Let's look further, at Foxconn. The referral bonus at Foxconn Zhengzhou dropped from 9,600-9,800 RMB in August 2025 to 7,800 RMB in August this year. What is a referral bonus? It's the extra price added during the peak season when people are most needed. If this price drops, it means what? It means people are much easier to recruit than in 2025. The only record that has emerged is profit. From January to July 2026, the profit in the integrated circuit industry increased by 18.5 times, and profits also doubled along with export value. However, people and wages have remained completely unchanged. This is a characteristic of price increases, not productivity.

Some might ask, what about electric vehicles? Well, let's look. BYD's employee count at the end of 2025 was 84,000 fewer than in 2024. In 2025 BYD produced 4.6 million cars, then spent over 150 billion yuan on an automated production line, claiming a 20% increase in per‑capita vehicle output. This may be the only sector in 2026 Chinese exports where per‑capita output truly rose. But that rise came from reducing staff, not hiring. Taiwan's path is simultaneous growth in headcount and capital expenditure. BYD’s path, however, is capital spending replacing staff, and cars only account for about 4.4% of exports, with prices continuously falling. Even assuming the CCP’s fantastical 5% per‑capita productivity boost plus less than a one‑point gain from automation, there is no evidence that per‑capita output jumped that dramatically in a single year. When we calculate the three components and combine the formula, the formula does only one thing: it draws an upper bound on export output from the labor side, because a country's annual output equals the number of workers multiplied by each worker’s hours multiplied by the output per hour per worker. Converting all three to growth rates and adding them gives the output growth rate, right? So what do the three numbers read for 2026? Headcount fell by about one point, hours fell by about one point, and even if per‑capita output rose the CCP‑claimed five points, that’s –1 + –1 + 5 = 3. Adding the chip and EV automation boost (still under one point) brings it to roughly four points. Even with a generous estimate—headcount flat, per‑capita output at the CCP’s upper‑bound six percent—that’s at most five points. This upper bound means that even if China’s labor force worked at full capacity in 2026, the most it could increase output is about five points. Yet reported exports rose 14.6%, meaning the extra ten points cannot be from labor; they must come from price, exchange rate, transshipment, or inventory. When we break down the export structure, we see it.

Chips account for less than 10% of total exports, yet half of the extra export volume in 2026 comes from chips and AI hardware exceeding expectations, all of it right here. Export value from January to August 2026 was 20.17 trillion yuan, 2.57 trillion yuan more than the same period in 2025. Nine‑tenths of that 2.57 trillion yuan is electromechanical products, whose biggest component is integrated circuits. The increase in integrated circuits is about 850 billion yuan, roughly a third of the total increase. The electromechanical industry’s own figures are even clearer: in August, AI‑hardware computers and integrated circuits together contributed 55.8% of the month’s electromechanical export growth. However, integrated‑circuit export volume in the first half of the year rose only 7%, while unit price rose 83.3%. By July, export volume rose just 1.88%. Who is producing these chips? Mainly Samsung’s Xi’an plant and SK Hynix’s Wuxi plant—both foreign‑owned. A small share comes from Yangtze Memory and ChangXin, essentially the “bro” industry. The price surge is due to AI servers devouring global storage capacity. Contract prices for several mainstream memory chips have risen more than seven‑fold since their 2022‑23 lows, and Chinese factories ship entirely at global prices. So the goods are the same, only the book value has doubled. Looking at the CCP’s customs data, processing‑trade imports and exports rose 26.4%, while general trade rose only 10.5%; the fastest‑growing segment is precisely this kind of component processing.

Let's look at the CCP's narrative again. What did the CCP choose for its narrative? It chose electric vehicles. It said that from January to August, the export value of automobiles increased by 51%, the export volume of the China Association of Automobile Manufacturers increased by 66.7%, and the volume of new energy vehicles doubled. However, the value did not keep up with the volume; the average price per vehicle dropped by about 10%. This presents a contradictory logic: chips are valued by price increase, while electric vehicles trade price for volume. These two trends are exactly opposite. By August 2026, the total car sales during the Mid-Autumn Festival will be 5% lower than in August 2025. Let's look at the overall picture of this industry: exports are rising, but total sales are falling. Setting aside whether they are actually exporting, even if the exports are real, a considerable portion of the exported volume is what? It's goods that couldn't be sold domestically and were pushed abroad. Why was the narrative chosen to focus on electric vehicles? My judgment is that only the electric vehicle line has quantity, factories, independent brands, and can produce visual content. Look, rows of cars rolling onto ships at the docks. This scene can be used for bragging and propaganda. This can then create a contrast. What about the story of the increase in chips? It's about Samsung and Hynix factories in China earning money for South Korea. Wouldn't that be detrimental to the CCP's nationalist narrative? In fact, the CCP Customs' own report released in the first half of the year already placed the AI sector ahead of China's electric vehicles, but this statement cannot be included in the propaganda narrative. As for other labor-intensive industries, from January to August, they decreased by 0.6%. This means that the exports that truly require labor are shrinking. The decrease of 1.67 million people in the four major first-tier cities aligns perfectly with this export structure. In other words, in the export structure, none of the items require people to move to coastal areas.

After discussing the structure, let's verify it with real-world metrics, because exports need to be packed in containers, loaded onto ships, and require electricity. These things cannot be faked. The number of containers at the port is counted by the piece, and the electricity from the grid is priced by the kilowatt-hour. Customs values can increase, but the number of containers cannot. In 2021, when exports increased by more than 20%, road freight also increased by more than 10%. In 2024, exports increased by 7%, and containers also increased by 7%. The number of containers and goods seem to balance out. However, in 2026, exports increased by 20%, but the increase in containers only followed one-third of the export growth rate, and the freight data only followed one-fifth of the export growth rate. This is the year with the worst lag in containers and freight since 2019, except for 2022 when there were lockdowns. From January to July 2026, the throughput of containers at ports nationwide increased by 5.1%, while operational freight volume only increased by 2.8%. Electricity consumption in the secondary industry increased by 4.7%. Automobiles are loaded onto roll-on/roll-off ships and do not enter containers. If the export volume of automobiles increased by more than 60%, and this is weighted in, how much would it contribute? It would contribute more than two percentage points to exports. If we also factor in the more than 20% decrease in mobile phone exports in July, how much would it drag down? It would drag down the entire export growth by more than one percentage point. Using the export value in 2025 as a weight and summing up the quantities of each category, the overall increase in the physical volume of exports would be around 4%, which aligns with the figures calculated from the human side.

Is there another way to verify this? At the factory gates. Customs exports are goods at the point of exit, and they don't care where the goods come from. Even chips that enter and leave the bonded zone are counted as exports. The CCP's National Bureau of Statistics' export delivery value is self-reported by factories, counting only goods they produced themselves. When these two sets of accounts are compared, the transit portion is revealed. This is data from the CCP's National Bureau of Statistics itself. What is the export delivery value of industrial enterprises above designated size? From January to July 2026, it increased by 9.8%. However, the increase in integrated circuit exports, converted to RMB, accounted for more than 80% of the increase in export delivery value. If integrated circuits are removed, the export delivery value only increased by about 2%. Even if only half of the chips are included in the output of large industrial enterprises, after removal, it would only increase by 6%. This means that when the intersection of people, goods, and factories is calculated, the real export growth rate in 2026 emerges. This export growth rate is only between 2% and 5%, with a midpoint of about 4%. It is then clear what constitutes the reported increase of 14.6%. The increase in integrated circuit prices accounted for 4.5 percentage points, while the structural increase in unit prices of mobile phones, rare earths, server components, etc., accounted for 1 to 2 percentage points. The actual goods accounted for about 4%. The remaining 4 to 5 percentage points are due to bonded transit and inventory of automobiles in overseas ports. The 19.3% growth in dollar terms is increased by 4.7 percentage points due to exchange rate conversion. So, what is the conclusion? There simply aren't that many goods!

We will combine the conclusions again, and everyone can clearly see the trend of the floating population. The trend is that people are moving back to the interior of the province, while working hours are still decreasing, and per‑capita output still has to rely on the Chinese Communist Party’s fantastical 5% growth assumption. In 2026, the real increase in Chinese exports was very optimistic, estimated at around a four‑point increase, and three‑quarters of that export growth was driven by chip price hikes. In dollar terms, what is 4.7 points? Four point seven points come from exchange‑rate conversion, and the remaining part is transshipment and inventory. This judgment does not rely on any CCP narrative; it relies on labor hours, productivity, or the CCP’s fantastical metrics. For a worker who moves from the coast back to the interior, the accounting becomes very concrete. Exports rose by 20%, but coastal factories will not call people back to work, because none of that extra 20% comes from hiring more workers, so once the accounting is done, I just fill in the CCP’s bragging headline: Chinese manufacturing carries the export backbone, and that backbone is Samsung and Hynix factories in Xi’an and Wuxi, relying on memory prices that have risen seven‑fold. Compared to that, it’s a loss of 9.35 million jobs per year and half an hour less work per week for workers. The most impressive thing is still the CCP’s miracle of per‑capita productivity growth that appears out of nowhere, and using this dream‑like labor productivity, the calculated export growth is only about up to four percent, and this time everyone understands why, under such perverse capital controls, China’s export surplus repeatedly creates miracles but fails to accumulate foreign‑exchange reserves, right? Everyone gets that, right? Next time we’ll continue to look at the CCP’s fantastical GDP. That’s all for today’s chat. If you like my show, thank you for the little bell. If you want to support me, you can join the membership channel; the membership channel has a birthday program for each driver, welcome to join Hai Ren Inventory. See you in the next episode.