The Physical Limits of China's Economy

The Thermodynamics of Growth

In January 2025, the National Bureau of Statistics released a set of data unseen since records began in 1996: fixed asset investment fell by 3.8% year-on-year for the full year, with a single-month decline of over 11% in October, while private investment dropped by 6.4%. The investment engine of the Chinese economy reversed for the first time in forty years.

In the same year, the National Bureau of Statistics reported a real GDP growth rate of 5.0%, while the nominal growth rate was only 4.0%. Independent estimates from the Rhodium Group were more pessimistic, placing real growth between 1.5% and 3%. Capital Economics stated bluntly that official data was overestimated by at least 1.5 percentage points. The results derived from these three measurement methods differ vastly, as if the efficiency of a heat engine were approaching its Carnot limit.

The Metaphor of the Heat Engine

An economic system can be viewed as an energy conversion device where low-entropy inputs (such as cheap labor, undeveloped land, and low-cost capital) pass through an institutional "combustion chamber" to be transformed into high-entropy outputs (such as goods, services, and infrastructure). According to the Second Law of Thermodynamics, the maximum theoretical efficiency of a Carnot engine is determined by the temperature difference between the high-temperature heat source and the low-temperature heat sink.

For the past forty years, the Chinese economy enjoyed a massive "temperature difference": extremely low factor prices (the high-temperature source) and extremely high global market demand (the low-temperature sink). This potential energy drove the fastest industrialization process in human history. Data from CF40 Research shows that in 2025, China's investment as a share of GDP still exceeded 40%, far higher than the global average of 23-27%.

However, when labor is no longer cheap, land is no longer abundant, and environmental carrying capacity reaches its limit, the high-temperature source begins to cool. When global markets reach saturation, the low-temperature sink warms up, the temperature difference narrows, and thermal efficiency naturally declines. The ICOR (Incremental Capital-Output Ratio) has climbed from 2.5 in the early stages to 6-7 today. This is not because today's investment decision-makers are more incompetent than those twenty years ago, but because the low-hanging fruit has long been picked—even the trees have been cut down for firewood.

In January 2023, Zunyi Road and Bridge Construction Group, a typical Local Government Financing Vehicle (LGFV) in Guizhou, announced a restructuring of 15.594 billion yuan in bank loans. The term was extended to 20 years, with only interest payments and no principal repayment for the first 10 years, and the interest rate was reduced from over 7% to 3-4.5%. This was the first publicly disclosed large-scale restructuring case in China's LGFV debt crisis. On Zunyi Road and Bridge's balance sheet, the book value of roads, bridges, and pipe networks far exceeds their market value: this infrastructure can never generate enough cash flow to repay the construction loans. The balance of LGFV debt in Guizhou Province has exceeded 60% of its GDP, ranking among the highest in the country.

Soviet economists faced a similar dilemma in the 1970s. From 1960 to 1980, the Soviet capital stock grew fivefold, but output only doubled. The solution during the Brezhnev era was to increase inputs, which only accelerated the "heat death." The 15.594 billion yuan loan of Zunyi Road and Bridge, the entropy increase solidified in reinforced concrete, is a microcosm of declining thermal efficiency.

Diminishing Pulses

The policy history of the past fifteen years resembles a medical record of gradually developing drug resistance. The 4 trillion yuan stimulus package in 2008 was like a shot of adrenaline, quickly pulling the economy into a V-shaped recovery. The monetization of shantytown redevelopment in 2015 served as a second stimulant; despite the side effect of skyrocketing housing prices, it maintained the illusion of prosperity for several years. However, by 2025, Reuters data showed that new loans for the year fell to 16.27 trillion yuan, a seven-year low. The transmission mechanism of the credit pulse has broken down.

The proportion of new loans in Total Social Financing (TSF) fell below 50%. Even when the central bank releases liquidity, funds struggle to flow into the real economy. Where the water flows is more noteworthy. Analysis by CREA (Center for Research on Energy and Clean Air) for Carbon Brief shows that over 90% of China's net new investment in 2025 flowed into the clean energy sector. The total scale of the clean energy industry reached 15.4 trillion yuan (approximately $2.1 trillion), approaching the economic size of Brazil or Canada, nearly doubling within three years (it was 8.4 trillion yuan in 2022). If clean energy investment were excluded, GDP growth would drop from 5% to approximately 3.5%, below the "around 5%" target set by Beijing.

The growth of the entire heat engine relies almost entirely on one cylinder; the other cylinders are not only slowing down but even reversing. Real estate investment fell by 17.2% for the year, private investment fell by 6.4%, and retail sales in December grew by only 0.9% year-on-year, the lowest in three years.

In biology, this is called "downregulation," where the balance sheets of enterprises and local governments develop a tolerance to debt stimulus. Increasingly larger doses are required to produce increasingly weaker effects, accompanied by increasingly severe side effects (such as bank bad debts and local debt risks).

In an economy that refuses to borrow because the return on investment is too low, lowering the cost of capital is like trying to push a string.

After the stimulus fails, what will the economy look like? Three rulers provide starkly different answers.

Three Rulers

The first ruler is the official real GDP growth rate of 5.0%, which measures the quantity of physical output—for example, how many tons of steel were produced, how many square meters of housing were built, and how many kilowatt-hours of electricity were generated. The second ruler is the official nominal GDP growth rate of 4.0%, which measures the monetary value of these outputs in the market. When the nominal growth rate is lower than the real growth rate, it indicates that society as a whole is selling more goods at lower prices. Economists call this a negative GDP deflator; physicists call it "doing negative work."

The PPI (Producer Price Index) has been negative for more than three consecutive years since October 2022, with a year-on-year decline of 2.6% for the full year of 2025. This is the longest industrial deflation cycle since China became a market economy. Factories are working overtime to produce things that are becoming increasingly worthless.

The third ruler comes from the Rhodium Group, whose independent estimates suggest real growth may be between 1.5% and 3%. An analysis by the Atlantic Council is even more direct: official data is overestimated by at least 1.5 percentage points. The root of the discrepancy lies in the disagreement over the definition of "growth": is it the intensity of statistical production activity, or the increment of value creation?

Suppose a factory produces 1 million photovoltaic modules that no one buys. In 2025, the solar industry suffered total losses of about $60 billion (Morningstar 2026). The first ruler of the Bureau of Statistics would record GDP growth brought by increased inventory; the second ruler of market prices would record the shrinkage in value caused by price-cutting promotions; the third ruler attempts to strip away the ineffective work of producing for the sake of production. When the three show such drastic divergence, growth itself has become a semantic trap.

Consumption data provides corroborating evidence, with final consumption expenditure accounting for about 39.9% of GDP, far below the 60% average of developed economies. The bustling activity on the output side contrasts sharply with the desolation on the consumption side: the pistons are moving violently, but the vehicle is almost stationary.

The three rulers measure different facets of the same heat engine. Regardless of which reading is chosen, one fact is unavoidable: energy that cannot be converted into useful work is accumulating in the form of waste heat.

The Destination of Waste Heat

The laws of thermodynamics state that energy does not disappear; energy that cannot be converted into useful work must be transformed into waste heat. In an economic system, the "waste heat" generated by ineffective investment manifests as deflation, overcapacity, and a decline in fiscal extraction capacity.

In 2025, the CPI remained flat year-on-year (0%), while the PPI fell by 2.6%. This is not just a price fluctuation, but the physical squeezing of supply-side overcapacity seeking an outlet.

The most intuitive manifestation of waste heat is the sunk investment solidified in reinforced concrete. The Kunming Changshui Airport Economic Demonstration Zone is a typical sample. This industrial new city, with a planned area of 135 square kilometers, was benchmarked against the Shanghai Hongqiao Business District when construction began in 2012, with plans to introduce aviation logistics, cross-border e-commerce, and high-end manufacturing. Ten years later, the vacancy rate of completed standard factory buildings in the park exceeds 40%, and the number of resident enterprises is less than one-third of the planned target. In 2024, data from the Kunming Municipal Finance Bureau showed that land transfer revenue had fallen by more than 60% from its 2020 peak. A city built for growth has become a monument to the failure of growth.

When traditional assets cannot be liquidated, local governments begin to capitalize everything that can be priced. In November 2024, Pingyin County in Shandong auctioned the operating rights for the "low-altitude economy" (drone logistics and airspace usage rights) for 30 years, with a transaction price of 924 million yuan. The winner was a state-owned enterprise wholly owned by the Pingyin County Finance Bureau (Jamestown Foundation 2026). Money moved from the left pocket to the right pocket, and GDP statistics gained an "investment." In May of the same year, Nanjing Public Transport Group used 70 billion pieces of bus route data as collateral to obtain a 10 million yuan credit line from a bank. Rongjiang County in Guizhou auctioned the franchise rights for its funeral parlor for 127 million yuan. Yi'an County in Heilongjiang auctioned the rights to dispose of reservoir silt, and the only bidder was a local state-owned enterprise established only 11 days prior.

The roads and bridges built with loans by Zunyi Road and Bridge, the vacant factories in Kunming Changshui, and the airspace Pingyin County bought from itself are not failed investments, but physical evidence of entropy increase.

More direct evidence comes from the fiscal side, where tax revenue as a share of GDP has slid from 18.4% in 2012 to approximately 12.5% in 2025. If 5% GDP growth is primarily composed of loss-making clean energy projects and low-margin manufacturing expansion, the tax sources the government can extract will naturally dry up.

The contraction of fiscal revenue and the weakness of nominal GDP are mutually causal. Local governments face a brutal "scissor gap": the input costs required to maintain growth (interest expenses, stability maintenance costs) are rigid, while the fiscal returns brought by growth (land transfer fees, corporate income tax) are elastic. Personal income tax accounts for only 1.1% of GDP, compared to an average of over 8% in OECD countries. The shrinking of the tax base is not the result of tax cuts, but a symptom of the deterioration of growth quality. This heat engine is consuming its own structural components to keep running.

Prognosis

The following nine chapters will explore the specific physical ceilings facing this heat engine: population, debt, infrastructure saturation, water resources, energy, productivity, exports, and pensions. Each is an independent constraint equation, which together form a system of simultaneous equations with no positive real solution.

If the growth rate of Total Factor Productivity (TFP) does not return to above 2% before 2027, the middle-income trap will transition from an academic hypothesis to a clinical diagnosis.

The criteria for judgment are very specific: observe the composition of labor productivity growth. Currently, the impressive figures for China's labor productivity are mainly contributed by "capital deepening"—equipping each worker with more machines. If this proportion exceeds 70% for a long period, and the contribution of technological progress and resource allocation efficiency cannot be improved, then this growth is unsustainable.

Because machines depreciate and debts mature, only improvements in efficiency can resist the erosion of the Second Law of Thermodynamics. The 20-year extended loan of Zunyi Road and Bridge merely erased the due date from the calendar and wrote a new one. In 2025, 87% of the principal of maturing local government bonds was repaid by issuing new bonds (Yicai Global 2026), and the average bond maturity was extended from 14.4 years to 15.4 years. The debt did not disappear; it just received a more distant deadline.

The signal is now very clear: this massive input-driven heat engine has touched the Carnot limit. The problem is not how to restart the engine, but how to redesign all the seats on a decelerating train.

The Arithmetic of Population

In 2025, the number of newborns in China was 7.92 million. The last time the birth population fell to such a low level was in 1738, the third year of the Qianlong Emperor's reign, when the total national population was only 150 million. Now, 287 years later, a population of 1.4 billion has produced the same number of infants. Meanwhile, according to data from CSIS ChinaPower, China newly installed 295,000 industrial robots in 2024, a figure exceeding the total of all other countries in the world combined. A phenomenon has emerged where machines increase while the population decreases. The answer to the arithmetic problem depends on the question being asked.

The Precipice

China is experiencing the fastest population contraction in human history, at a speed that makes even South Korea appear slow.

According to data from CS Monitor and the SuperSally Substack, China's Total Fertility Rate (TFR) dropped from 1.3 in 2020 to 1.01 in 2023. This decline took only three years, whereas it took South Korea 17 years to complete the same drop. Such a vertical plunge has not only shattered official population projection models but has also instantly invalidated all economic planning based on "gradual aging."

Behind this lies the force of cultural inertia, not just policy failure. The thirty-five-year one-child policy affected more than just population numbers; it reshaped the very concept of "upbringing." Several generations have grown accustomed to a resource-concentration model of "six wallets supporting one child," pushing the cost of child-rearing to irreversibly high levels. In February 2026, Financial Analyst Net estimated that the average cost of raising a child to age 18 in China is approximately 538,000 yuan, more than 6.3 times the GDP per capita—a ratio far higher than in the United States, Japan, or Europe.

Policies can be abolished overnight, but household balance sheets cannot be adjusted instantly. The government's fertility subsidies (such as 3,600 yuan per year in some regions) account for only 0.07% of GDP, while France's spending on family benefits exceeds 3% of its GDP. The scale of subsidies is negligible.

Marriage registration data provided an earlier signal. Annual data from Sinification/Scour shows that the number of marriage registrations dropped from 13.469 million couples in 2013 to 6.106 million in 2024, a 50% decrease in a decade. Marriage is the prerequisite for childbirth; when the number of certificates issued is halved, a decrease in births follows logically.

The Cruelty of Exponentials

The most dangerous misconception in demography is thinking it is just about "a few fewer children being born." In reality, negative compounding has already begun.

The current precipice in birth rates is essentially a second-order effect of the previous round of declining birth rates. An analysis by TippInsights points out that the number of women aged 20-34 (the primary group contributing 85% of births) will drop from 105 million in 2025 to 58 million by 2050. Even if Chinese families miraculously regain the desire to have children and double the fertility rate in the future, the total number of births can only barely be maintained because the number of women of childbearing age has already been halved.

This positive feedback loop means that fewer births lead to fewer women of childbearing age in the future, which in turn leads to even fewer births. Based on the current fertility rate (approximately 1.0), generational replacement presents a cold geometric progression: 100 adults produce only 50 children, those 50 children produce 25 grandchildren, and those 25 grandchildren produce 12.5 great-grandchildren. After three generations, the family size shrinks by 87.5%. This is the direct result of arithmetic.

The United Nations (UN) medium-variant projection suggests that China's population will drop to about 650 million by 2100. Independent models by scholars such as Bert Hofman show that, considering the current accelerating downward trend, the point at which the population falls below 1 billion may be moved forward to before 2070. Four consecutive years of population decline (a decrease of 850,000 in 2022, 2.08 million in 2023, 1.72 million in 2024, and 3.39 million in 2025, totaling approximately 8 million over four years) is only the beginning of this long downward process. The 2025 birth rate of 5.63 per thousand is the lowest since 1949, and the TFR has dropped to approximately 0.98, lower than the world record once set by South Korea. A nation of 1.4 billion people is shrinking at an unprecedented speed.

One-Third of the Robot

As the labor force gradually disappears, the mainstream narrative offers "automation" as the solution. China is indeed advancing the largest machine-for-human replacement plan in human history. With 470 industrial robots per 10,000 workers, it ranks third globally. Robotic arms in factories are tireless, and algorithms require no rest; automation effectively solves the problem of "who will produce."

However, "robots replacing labor" only solves part of the problem, because the economic cycle also requires consumers and taxpayers. An industrial robot can efficiently weld a car chassis, but it will not buy a car, will not pay purchase taxes for a car, will not take its family on a weekend trip, and certainly will not contribute pension premiums to the social security fund.

Household disposable income accounts for only 43% of China's GDP, which is low among major economies. Automation further depresses the share of labor income in national income, and the contraction of consumption capacity will be faster than the contraction of the labor force. Automation alleviates supply-side pressure but exacerbates demand-side and fiscal dilemmas. A production line can run 24 hours a day, but it will not go to a restaurant after work. The number of great-grandchildren for 100 Chinese adults is 12; for 100 robots, the number of great-grandchildren is always zero.

The Arithmetic of Pensions

All demographic problems eventually transform into fiscal problems. Data from the National Bureau of Statistics shows that the proportion of the working-age population has dropped from a peak of over 70% to 60.6%. Meanwhile, the population over age 60 is expected to reach 400 million by 2035. Within ten years, China's elderly population will exceed the total population of the United States.

The current pension system uses a pay-as-you-go model, which presupposes that there are enough young people to pay for the elderly. When the dependency ratio drops from 5:1 to 2:1 or even lower, no amount of parameter adjustment will be effective.

The delayed retirement policy announced in 2024 (gradually extending the retirement age for men from 60 to 63) is mathematically only a minor mitigation. The earlier warning from the Chinese Academy of Social Sciences (that the pension surplus will be exhausted by 2035) is gradually becoming a real-time countdown on financial statements.

If China's Total Fertility Rate (TFR) does not return to above 1.3 before 2030, the pension system will face a deficit around 2035; this is an inevitable conclusion of arithmetic deduction. The criterion is very direct: whether the ratio of active contributors to retired recipients falls below 2:1. Once this point is crossed, policy options are left with only three difficult scenarios: cutting pension payments, increasing the tax burden on active workers, or diluting debt through large-scale currency issuance.

The Northeast Preview

To understand what China will look like in 2035, one does not need a time machine, only a plane ticket to Heilongjiang. The three provinces of Northeast China are the frontier of China's population crisis; this is where the birth rate is lowest, pension accounts were the first to sound the alarm, and economic stagnation is most evident. In 2021, Heilongjiang saw 100,000 births and 460,000 deaths; its population decreased by 17% over a decade.

Longmay Mining Group is a microcosm of the crisis. At its peak, this largest state-owned enterprise in Heilongjiang had 248,000 workers spread across the four coal cities of Hegang, Jixi, Shuangyashan, and Qitaihe. In March 2016, miners in Shuangyashan took to the streets with banners saying "We want to survive, we want to eat." The cause was Heilongjiang Governor Lu Hao's claim at the National People's Congress that all Longmay workers' wages had been paid in full, while a miner named Chen told AFP that more than 60% of wages for 2014 and 2015 remained unpaid. Lu Hao had to admit he was wrong. Subsequently, more than 30 miners who led the protests were detained, and Longmay announced layoffs of 100,000 people, accounting for 40% of its total workforce.

The Northeast demonstrates the full picture of a negative population growth cycle: the real estate market is frozen because there are no buyers (Hegang-ification), infrastructure is aging due to fewer taxpayers, and public services are downgraded due to fiscal constraints. In 2020, the population of Hegang was approximately 689,000, a 35% decrease from 1.06 million in 2010. Housing prices dropped to 1,878 yuan per square meter (Anjuke data), and the topic "buying a house in Hegang for 30,000 yuan" on social media has tens of millions of views. "Hegang-ification" has spread to the south: a 132-square-meter house in Jieyang, Guangdong, was listed for 238,000 yuan, or 1,831 yuan per square meter. Housing prices in Hegang are low not because the quality of the houses is poor, but because no one needs to live there. The value of a city ultimately depends on how many people are willing to stay.

This is far more than just a regional economic issue; it is a microcosm of the entire country's future in 10-15 years. The Northeast did not make any special mistakes; it simply reached this point 20 years earlier than other regions. When the national demographic structure converges with that of today's Northeast, the fiscal transfer payment capabilities of the Yangtze River Delta and the Pearl River Delta will encounter the same limits.

Population is the only truly unidirectional variable among all economic factors. You can print money, build bridges, and change laws, but you cannot print 20-year-old youths.

The Gravity of Debt

In 2008, for every 1 yuan of GDP growth created, China needed an input of approximately 1.5 yuan in new debt; by 2023, this figure exceeded 8.4 yuan. In the same year, total non-financial sector debt reached 285% of GDP, a ratio that doubled in 15 years. According to calculations based on Ministry of Finance data, 87% of local government bonds maturing in 2023 relied on "rolling over" (issuing new debt to pay old debt) for repayment. In physics, when the mass of a celestial body exceeds a critical value, even light cannot escape; in economics, this critical value is often vaguely referred to as "risk," but Chinese economic data is beginning to provide a more precise definition: a gravitational singularity.

The Gravitational Field

Debt is like mass. During periods of low debt ratios, the gravitational effect is negligible, acting merely as a lever to pry open growth. When the debt-to-GDP ratio climbed from 135% in 2008 to 285% in 2023 (Statista, 2023), while nominal GDP growth slowed from double digits to around 4%, the rules of physics changed.

Newton's law of universal gravitation states that gravity is proportional to mass and inversely proportional to the square of the distance. In macroeconomics, "mass" is the debt stock, and "distance" is the escape velocity provided by economic growth. The current situation is that mass has doubled and escape velocity has halved; the intensity of the "debt gravitational field" is not growing linearly but strengthening with nearly 10 times the intensity.

In a high-intensity gravitational field, every economic decision is distorted: companies no longer invest based on market opportunities but sell assets based on debt repayment pressure; local governments no longer plan based on public interest but build cities based on financing capacity. Light bends near massive celestial bodies, and capital has also undergone path deflection in China's current debt environment: funds refuse to enter the real economy and instead circulate idly within the financial system because the real rate of return has fallen below the cost of debt.

In the first quarter of 2024, China's banking industry added record credit injections, yet M1 (narrow money) growth rarely fell into negative territory. In fluid mechanics, this is equivalent to injecting water into a pipe while the water pressure drops because the pipe itself is leaking. Funds are used to pay interest on old debts rather than create new transactions. This is a physical characteristic of a black hole's accretion disk: matter accelerates and spins before falling into the event horizon, emitting a dazzling light (massive social financing data), and then vanishes without a trace.

If total factor productivity (TFP) growth fails to return to above 2% before 2027, or if nominal GDP growth cannot stay above the average debt interest rate for long, gravity will evolve from a "drag" into a "collapse."

The Mathematics of Rolling Over Debt

In 2023, Chinese local government bonds paid approximately 1.48 trillion yuan in interest, a year-on-year increase of 9.6%, and this is just the interest. The repayment of principal is even more stark: 87% of maturing bonds were repaid by issuing new bonds.

Finance textbooks define the behavior of relying primarily on new financing to pay interest or principal on old debt as "Ponzi Finance"; in the context of China's Ministry of Finance, this is called "debt management."

In November 2024, the Standing Committee of the National People's Congress approved a 10 trillion yuan hidden debt swap plan. According to the announcement, this includes increasing the local government debt limit by 6 trillion yuan over three years and arranging 4 trillion yuan from new local government special bonds. Although the market cheered it as an "epic bailout," analysis still needs to see the essence of the balance sheet.

The plan swaps high-interest, short-term hidden debt (usually held by Local Government Financing Vehicles, LGFVs, with interest rates reaching 7%-10%) for low-interest, long-term explicit government bonds (with interest rates around 2.3%). In a mathematical sense, it indeed saves about 600 billion yuan in interest expenses (Ministry of Finance estimate), much like a person using a platinum credit card with a 5% APR to pay off an online loan with a 15% APR.

The total amount of debt has not decreased by a single cent.

Such swaps not only fail to reduce leverage but instead solidify what was originally vague, theoretically defaultable commercial debt into sovereign credit debt that must be rigidly repaid. Risk is transferred from the capillaries of the financial system (LGFVs) to the main artery (provincial finances), buying about 3 to 5 years of time at the cost of overdrawing future fiscal space and locking in future policy options.

When an economy needs to issue more than 10 trillion yuan in new debt annually (2024 data) just to keep the existing debt scale from collapsing, the initiative of fiscal policy has effectively been lost. Every round of "rolling over debt" is an unconsented tax levied on the next generation.

The Paradox of Deleveraging

China is attempting an operation rare in human economic history: deflationary deleveraging.

When dealing with debt crises, the United States and Japan usually choose "inflationary deleveraging," pushing up nominal GDP (the denominator) by printing money to dilute the real value of debt (the numerator). While this plunders savers, it saves the debtor. China has taken the opposite path: attempting to reduce the leverage ratio by cutting the absolute value of debt in an environment where inflation is controlled or even deflation is tolerated.

This corresponds to a mathematical death loop: Leverage Ratio = Total Debt / Nominal GDP.

When the government tightens the money supply, cracks down on real estate, and rectifies local debt, the numerator (debt growth rate) indeed slows down. However, these measures simultaneously lead to falling asset prices, declining corporate income, and stagnant wages, which in turn causes the growth rate of the denominator (nominal GDP) to fall even faster.

The data demonstrates this cruel paradox: China's real GDP growth in 2023 was claimed to be 5.2%, but nominal GDP growth was only 4.6% (or even lower, depending on the calculation of the deflator). The PPI (Producer Price Index) has fallen year-on-year for 18 consecutive months. When the nominal growth rate is lower than the real growth rate, it indicates that the economy is in a state of widespread price decline.

In a deflationary environment, the real value of debt rises. Companies earn shrinking cash flows but repay rigid debt principal and interest. The result is that the harder they work to repay debt, the harder the remaining debt becomes to pay; the more they cut spending, the colder the macroeconomy becomes, the smaller the denominator gets, and the higher the leverage ratio actually climbs.

Data from the Bank for International Settlements (BIS) confirms this: despite three years of strict regulation and deleveraging efforts, China's non-financial sector credit-to-GDP gap remains high. Attempting to deleverage in deflation results not in a healthier balance sheet, but in a balance sheet recession.

Contagion of the Balance Sheet

The "balance sheet recession" theory proposed by Richard Koo accurately describes the current micro-level pain. When asset prices plummet (real estate) while liabilities (mortgages) remain unchanged, the primary goal of households and enterprises shifts from "profit maximization" to "debt minimization."

This shift is vividly reflected in the data of China's household sector: in 2021, the average propensity to spend (consumption expenditure / disposable income) of Chinese residents was about 101.4% (China Banking News, 2021), meaning people were borrowing to consume and overdrawing the future; by early 2025, this ratio plummeted to around 80%. In just four years, the shift from "spending more than one earns" to "defensive saving" formed a 20-percentage-point psychological avalanche.

The spillover effect of the real estate crisis is devastating. According to estimates by Bloomberg Economics, the real estate downturn has destroyed approximately $7 trillion to $18 trillion in paper wealth for Chinese households, and the negative wealth effect has triggered a contraction in consumption.

A more dangerous contagion is occurring, forming a perfect negative feedback loop: a wave of early mortgage repayments in the household sector leads to a plunge in consumer demand; the corporate sector is forced to lay off workers, cut wages, and stop investing due to declining income; ultimately, this leads to the deterioration of bank assets. In 2024, the number of individual loan defaults reached 25 to 34 million (aggregated financial sources), double that of 2019, with the number of people with overdue payments even as high as 83 million. Compared with Japan in 1990, the challenges China faces are more severe. When Japan entered a balance sheet recession, its per capita GDP was about $30,000 (in 2023 dollars), making it a wealthy aging society; China's current per capita GDP is about $12,700, making it a society that has grown old before getting rich.

Japan had 30 years to digest its bubble because Japanese household savings were sufficient to support a long-term low-interest environment. Although Chinese household savings are high (savings rate of 32%), they face an incomplete social security system and an extremely high dependency ratio, making the cushion much thinner.

The 10-year government bond yield fell below 2.3% in 2024 and further dropped to 1.793% in early 2025, approaching Japan's trajectory back then. Bond market traders are voting with real money: they do not believe there will be a strong recovery in the short term and are betting on long-term low growth and low inflation.

Predatory Fiscal Policy

As physical limits approach, the system seeks the path of least resistance to release pressure. For local governments that have lost land finance, this path points toward microeconomic entities.

Land transfer revenue once accounted for more than 40% of local fiscal revenue, but this source has plummeted by more than 50% from its 2020 peak, falling to 4.15 trillion yuan in 2025, the fourth consecutive year of decline (Yicai Global 2026). Against the backdrop of meager central government transfer payments, while massive debt interest and rigid expenditures (civil servant salaries, operating costs) remain unchanged, the result is a mutation in the method of fiscal extraction.

In 2024, the Henan Provincial Public Security Department mobilized more than 1,600 police officers to Guangzhou to conduct a surprise raid on Archealth, a private health technology company in Guangdong, freezing company accounts and forcing it to terminate its ongoing IPO plans (Jamestown Foundation 2025). Archealth was not a drug company or a fraud syndicate, but a compliant company preparing for listing. The travel, accommodation, and enforcement costs for these 1,600 police officers were ultimately reflected in Henan Province's fiscal reports. In the same year, a valve manufacturer in Shanghai had 19.3 million yuan in assets frozen by a Shanxi court "without any business dealings, solely to ensure cash flow"; Zhijiang Liquor received a tax bill of 85 million yuan dating back to 1994, facing a 30-year retroactive audit.

The above phenomenon has a colloquial name: "Deep-sea Fishing." Previously, local governments "raised chickens to get eggs" by attracting investment and providing tax incentives; now, facing the gravity of debt, the logic has changed to cross-provincial "fishing." Confiscation and fine income in places like Lianyungang and Quzhou increased by more than 100% year-on-year, not because public security suddenly deteriorated, but because of extreme fiscal hunger.

The scale of the problem was so large that the Supreme People's Procuratorate had to intervene. A special rectification campaign launched in March 2025 had, by the end of the year, handled more than 19,000 cases of improper cross-regional law enforcement. More than 3,000 criminal cases were withdrawn, more than 3,500 people received non-prosecution decisions, and more than 26 billion yuan in illegally seized, detained, or frozen assets were ordered to be returned (China Daily 2026). This scale directly confirms the profound nature of fiscal plunder.

The ratio of tax revenue to GDP has fallen to a historic low of 12.5%, but this does not mean the burden on enterprises has lightened. National fiscal revenue in 2025 fell by 1.7% year-on-year, the first contraction since 2020 (Reuters 2026). The disorderly expansion of non-tax revenue is destroying the predictability of the business environment. For a company, a clear 30% tax rate is an acceptable cost, while a surprise raid by 1,600 cross-provincial police officers is an unhedgeable existential risk.

The aforementioned predatory fiscal policy is the ultimate projection of debt gravity at the social level: to maintain the operation of the massive debt machine, the system begins to devour the economic cells it relies on for survival. The conclusion of physics is simple: there is only one way to counter gravity, and that is to maintain sufficient speed; when speed disappears, gravity will dominate everything.

The Saturation Point of Cement

Dushan County in Guizhou has a population of 350,000, roughly equivalent to the entire population of Iceland. This county has accumulated 40 billion RMB in debt to build the "World's Number One Water Bureau Building," a university town, and a highway. Most of these facilities were idle upon completion. Dushan County is by no means an isolated case; it is a miniature specimen of China's infrastructure investment model and a living dissection of marginal returns hitting zero. Dividing 40 billion RMB by 350,000 people results in a per capita debt of 114,000 RMB, which exceeds 20 years of disposable income for rural residents in Guizhou.

Saturated Solution

In chemistry, the saturation concentration of a solution depends on two variables: the nature of the solute and the temperature of the solvent. In economics, investment is the solute, and the economic growth rate is the temperature. Over the past twenty years, the Chinese economy has maintained high-temperature operation, allowing massive infrastructure investment to be "dissolved" and digested by the rapidly growing economic volume.

Now, the temperature has dropped. When the economic growth rate (temperature) slows down, the solution's saturation capacity decreases accordingly. Continuing to add salt to an already saturated solution will not result in a more concentrated solution; it will only produce precipitate. Chemists call this "supersaturation," economists call it "overcapacity," and local officials call it "major projects."

The forms of precipitation are diverse: highways with no vehicles, idle regional airports, and unfinished industrial parks. The Autonomous Rail Rapid Transit (ART) Demonstration Line 1 in Shaanxi's Xixian New Area cost approximately 700 million RMB. It opened in March 2023 and ceased operations in January 2026, with a lifespan of less than three years. Several key stations never opened due to land compensation disputes, and fare revenue could not cover operating costs. Journalist Liu Yu called it a "vanity project": "The focus was on being the 'first' and 'filling a gap,' rather than whether people actually needed it." Nationwide, cases similar to Dushan County are scattered across every province. Whether it is Ordos in Inner Mongolia or the Chenggong New District in Yunnan, "ghost cities" are not anomalies; they are the inevitable precipitation of investment supersaturation.

Price signals have confirmed this precipitation process. The PPI (Producer Price Index) has been negative for a long time since 2022, and for the full year of 2025, it stood at -2.6% (CF40 Research, 2025). The root cause of the long-term price decline is by no means an improvement in production efficiency, but rather supply far exceeding demand. This is the projection of physical laws onto economic statements: when the solute cannot be dissolved, it can only pile up at the bottom in the form of inventory and idle capacity.

The Dushan Specimen

Regarding Chinese infrastructure, there is a common defense that China's per capita infrastructure stock is still lower than that of developed countries, and therefore there is huge room for growth. However, this argument confuses "stock" with "marginal returns."

The case of Dushan County proves that the core issue is not whether infrastructure is "enough," but "how much economic return a new unit of infrastructure can generate." For a county with a declining population, the marginal utility of a fourth highway is close to zero, or even negative (because maintenance costs exceed toll revenue).

The Incremental Capital-Output Ratio (ICOR) is a key indicator for measuring such efficiency. Although official precise data is no longer published, according to independent estimates, China's ICOR has climbed from around 2.5 a decade ago to the 6-7 range. In other words, whereas investing 2.5 RMB used to generate 1 RMB of GDP growth, it now requires an investment of nearly 7 RMB. Efficiency has dropped nearly threefold, while the total volume of investment continues to increase. For comparison, India's current ICOR is approximately 4, and Vietnam's is approximately 3.5, both at the level China was at ten years ago. The root of the gap is not management capability, but the stage of development: the low-hanging fruit has already been picked.

Dushan County's 40 billion RMB debt (Chinese financial media reports, 2024) was not a simple decision-making error, but an extreme expression of the law of diminishing marginal returns. When every project with economic value has been built, capital can only flow into those without value. The Water Bureau Building is not a tourist attraction; it is the solidified form of capital in the absence of effective targets.

Misinterpreted Environmental Achievements

In 2025, carbon dioxide emissions from China's cement industry fell by 7% (industry reports, 2025), which was interpreted by some media as a victory for the green transition.

In reality, this is a dangerous misinterpretation. The driver of the emission decline was not a breakthrough in carbon capture technology or a revolutionary improvement in fuel efficiency, but a physical contraction in cement production. Several provinces implemented strict capacity caps, which was merely an acknowledgment of the collapse in demand. When an industry's "environmental achievement" comes from the shrinkage of its core business, environmental departments write the success into reports, while economists should write the same set of numbers into a medical record.

This decline is structural. With the plunge in new real estate construction area, cement demand has lost its largest support point. If emissions are reduced because factories have stopped working, the blue sky is a byproduct of economic contraction.

What is seen is not an improvement in energy efficiency, but a reduction in the total volume of physical activity. This is substantially different from the carbon peaking achieved by Western countries through technological upgrades: one is metabolic optimization, the other is dieting and starvation. China's per capita cement consumption was once six times that of the United States. Now this multiple is falling, not because alternative materials have been found, but because there are no more new buildings requiring cement.

Withdrawal Symptoms

If investment is a drug, the Chinese economy has developed a deep drug dependency: when investment contributes 40% of GDP, it has long since transcended the role of a single growth engine and evolved into a life-support system for the entire social structure. In 2025, fixed asset investment fell by 3.8% year-on-year, with a single-month decline of over 11% in October (Bloomberg via RWATimes, 2025).

The meaning of this figure goes far beyond the reports of the Bureau of Statistics. Investment contraction means the unemployment of construction workers, the disappearance of orders for construction machinery, the shutdown of steel mills, and the drying up of land sale revenues for local governments. Every link in an industrial chain is decelerating simultaneously.

Data aggregated by TechSoda shows that China currently has approximately 80 million vacant or unsold housing units. This number is enough to house the entire population of Germany; Germans do not need to live in the empty cities of Guizhou. This inventory is not merely a misallocation of resources, but a massive source of deflationary pressure. For every additional vacant house, there is more downward pressure on surrounding housing prices; for every bit of downward pressure, developers have less incentive to acquire land; with less incentive to acquire land, local finances lose a source of income.

This creates a core physical dilemma: continuing to invest is like adding salt to a saturated solution, creating more vacancies and debt; stopping investment is like cutting off the life-support system, triggering severe withdrawal symptoms.

Policymakers are not facing a choice between "growth" and "recession," but a gamble between "chronic poisoning" and "acute shock." The 4 trillion RMB stimulus of 2008 proved the cost of the former, while the investment cliff of 2025 demonstrates the pain of the latter. Both paths lead to the same wall; the only difference is the speed of the collision.

The Next 4 Trillion

To alleviate the withdrawal symptoms, a new dose has been injected: the "15th Five-Year Plan" proposes power grid investment as high as 4 trillion RMB, a 40% increase over the "14th Five-Year Plan" period (TechSoda, 2025). The investment direction has shifted from real estate and traditional infrastructure to ultra-high voltage (UHV), new energy, and digital infrastructure.

This is tactically correct; the marginal return on power grids is theoretically higher than that of cement roads. But strategically, while the direction has changed, the model has not. It still continues the old logic of "investment-driven growth."

Housing prices have fallen for four consecutive years, with a cumulative decline of about 12% from their peak (TechSoda, 2025). Real estate's function as a reservoir has failed. The question now is whether 4 trillion RMB in power grid investment can replace the massive vacuum left by real estate.

The answer from physics is that energy conversion always involves loss. The conversion from cement to copper cables cannot fully offset the loss of momentum caused by the contraction of the real estate industrial chain (which involves more than 50 upstream and downstream industries). More critically, if the growth in electricity demand cannot keep pace with the speed of grid construction, the result will simply be turning "vacant buildings" into "idling electricity meters."

A falsifiable prediction: if fixed asset investment growth does not return to positive territory by 2028, and the actual utilization rate of power grid investment is lower than 60% of design capacity, then this 4 trillion RMB investment will ultimately suffer the same fate as the Dushan County Water Bureau Building—expensive assets with meager returns. Even if the solute changes from cement to copper cables, the solution remains saturated.

The Water Ledger

In 1997, the Yellow River ceased to flow for as long as 226 days. Tracing upstream from the Shandong estuary, nearly 700 kilometers of riverbed lay exposed, like a parched scar stretching across the North China Plain. That same year, China's GDP growth rate reached 9.2%. Economists saw a soaring curve in their charts, while hydrologists read signals of crisis on the riverbed.

Twenty-eight years later, the Yellow River no longer runs dry—not because of increased rainfall, but because the Middle Route of the South-to-North Water Diversion Project transports approximately 9.5 billion cubic meters of water to the north annually. However, this has not filled the deficit on the water resource ledger. China is currently constructing 73 new wafer fabs in an attempt to build a semiconductor industry fortress, yet over 40% of these sites are located in river basins expected to face high water stress by 2030. The problem of the Yellow River has never been a simple engineering challenge; it is a ledger problem of resource allocation.

The 35% Starting Point

The core of China's water resource crisis lies in an unalterable physical constant: per capita water resources are only 35% of the global average (Alwihdainfo; Frontiers in Sustainable Food Systems, 2026). This is not just a symbol of scarcity, but the root of an allocation dilemma.

This 35% of water resources is distributed extremely unevenly. The region south of the Yangtze River possesses 80% of the nation's water resources but cultivates only 36% of its arable land. In contrast, the region north of the Yangtze, despite having 64% of the arable land and nearly half the population, is allocated only 20% of the water. This imbalance has led to a deep tension in China's economic geography: both food security and energy security (as major coal-producing areas are concentrated in the north) are built upon a water resource deficit.

Even more severe is that this limited 35% is being further eroded by pollution and overuse. According to a joint report by the World Bank and the State Environmental Protection Administration (SEPA), as early as 2003, economic losses caused by water pollution had reached 362 billion yuan, accounting for 2.68% of the GDP at the time; if calculated using the Value of a Statistical Life (VSL) method, this proportion soared to 5.78%. Although governance efforts have intensified in recent years, the intergenerational transmission effect of legacy pollution persists. Approximately 54% of the sections of the seven major river systems have water quality unsuitable for drinking. On the already scarce water resource ledger, nearly half are "bad assets."

The fading of the population dividend is widely discussed, yet the overdrawing of the water resource dividend is rarely mentioned. Population numbers can fluctuate through policy adjustments, but the physical limits of water resources are an insurmountable barrier. When per capita water resources drop below the internationally recognized "water stress line" of 1,700 cubic meters, economic growth is no longer a simple matter of input and output, but a squeeze on living space.

Water Debt

In economics, debt is a tool for bringing future income into the present. As long as interest can be paid, debt can be continuously rolled over. However, the North China Plain is accumulating a special kind of debt: "water debt."

Zhi Huazi has been digging wells around Shijiazhuang for 20 years. In the jargon of this veteran well-driller, wells shallower than 200 meters are considered "shallow wells," usable only for irrigation or industry; drinking water wells must be drilled to 400 to 500 meters, with a recent well even exceeding 600 meters (Eco-Business/Chinadialogue). Village elders recall that 50 years ago, water could be found by digging a few shovelfuls of soil, but today, moving from a few shovelfuls to a 600-meter drill bit is not a sign of technical progress, but a symbol of the exhaustion of geological resources. Data from the China Geological Survey shows that between 1999 and 2013, the groundwater level in one monitoring well dropped by 17 meters, an average of 1 meter per year. Shijiazhuang's per capita water consumption is only 229 cubic meters, less than half of the United Nations' "absolute water scarcity" standard (500 cubic meters); Taocheng District in Hengshui City is even more extreme, with per capita water resources of only 120 cubic meters, deep groundwater levels dropping by 2 meters annually, and an over-extraction cone area exceeding 6,300 square kilometers (GWP).

The agricultural miracle of the North China Plain is built upon the predatory extraction of groundwater. According to World Bank data, the region extracts approximately 25 billion cubic meters of non-renewable deep groundwater annually, 90% of which is used for agricultural irrigation. Deep groundwater is fossil water formed over tens of thousands of years of geological processes; once extracted, the geological structure collapses like a sponge losing its support.

This collapse is already happening. Satellite InSAR data for Beijing's Chaoyang District shows a land subsidence rate of 110 millimeters per year, equivalent to CBD office buildings sinking 11 centimeters annually (Remote Sensing 2016). Subsidence rates in Changping, Shunyi, and Tongzhou are 6 to 8 times those of the central area. A 50-kilometer section of the Beijing-Tianjin Intercity High-Speed Railway passes through the subsidence zone. Before the South-to-North Water Diversion project began supplying water, sections with annual subsidence exceeding 30 millimeters accounted for 26.5% of the total length of the Beijing section, with extreme annual subsidence reaching 105 millimeters (Journal of Wuhan University). Researchers from Beijing Jiaotong University warn that subsidence has a "strong impact on train operation safety," and more than 50 Chinese cities face similar issues.

Financial debt can be mitigated through money printing, inflation, or debt restructuring, but water debt has no such options. No central bank can "issue" water resources, and no quantitative easing can replenish aquifers formed tens of thousands of years ago. When wells hit bedrock and land cracks due to subsidence, the only result is a physical end: a permanent decline in agricultural output.

Trading ten thousand years of water for one year of grain.

The Water Footprint of Chips

To break free from geopolitical constraints, China is undergoing the largest expansion of semiconductor capacity in human history. Since 2019, China has invested over $63 billion, planning and constructing 73 new wafer fabs (3D Printing Progress; IDTechEx). This is viewed as a national security strategy aimed at reducing dependence on external supply chains, especially Taiwan.

However, this strategy is facing a severe challenge from water resources. Semiconductor manufacturing is an extremely water-intensive industry, accounting for approximately 27% of China's total manufacturing water consumption. Taking TSMC as an example, its water consumption reached 101 million cubic meters in 2023, while Samsung Electronics' daily water consumption is as high as 344,000 tons.

Ironically, while China attempts to replicate Taiwan's chip capacity, it ignores Taiwan's predicament regarding water resources. From 2020 to 2021, Taiwan experienced its worst drought in 56 years, with the Baoshan Reservoir's storage rate once dropping to 14.35%. TSMC allocated 800 million TWD (approximately $28.6 million) for water trucks, extracting 20,000 tons of groundwater daily to transport to the Fab 15 plant in the Central Taiwan Science Park (Fortune/WCCFTech 2021). A typical wafer fab uses about 20,000 tons of water per day, equivalent to the water consumption of a city of 58,000 people. Rice farmers in Tainan County were banned from planting for three consecutive years, with the government paying subsidies in exchange for fallowing, because water resources were prioritized for semiconductor factories. Tomato farmer Yang Kuan-wei told NPR: "Our water was already barely enough, and now you want to give more of it to others." The choice between chips and grain is a preview Taiwan has already performed for China.

In August 2022, Sichuan Province experienced another form of preview. The worst heatwave in 60 years caused reservoir levels at hydroelectric dams to drop by half. The Sichuan provincial government ordered all factories to halt production for six days, with officials calling it the "most severe and extreme moment for power supply." Factories of Intel, Texas Instruments, and Onsemi in Sichuan all suspended production, and Foxconn's iPad production lines and CATL's battery factories were also affected (The Verge 2022). Drought led to insufficient hydropower, which in turn triggered power rationing and chip plant shutdowns. A water resource crisis does not need to wait for the day the tap runs dry; it only needs the reservoir level to be too low for generators to operate.

The site selection for China's new wafer fabs also exposes this contradiction. According to semiconductor industry water consumption reports, over 40% of wafer fabs built after 2021 are located in river basins expected to face "high" or "extremely high" water stress by 2030. Beijing, Xi'an, and Inner Mongolia—these water-scarce regions are becoming new chip hubs.

A dangerous substitution: China is attempting to escape technological dependence on Taiwan through semiconductor self-sufficiency, but it may fall into a dependence on the fragile water resources of the north, shifting from a geopolitical vulnerability to a hydro-geographical vulnerability. Tainan's rice farmers were forced to fallow to protect TSMC; North China's wheat farmers will sooner or later face a similar choice.

Invisible Water

If semiconductors are the obvious water-consuming giants, then Artificial Intelligence (AI) is a water resource black hole disguised as "clean technology."

Public and media attention often focuses on the cooling towers of data centers. Indeed, data centers require water for heat dissipation, and 40% of existing data centers are located in high water stress areas (Chemical Engineering). However, according to research reports from Xylem and GWI, data centers account for only 4% of the total water consumption increment in the AI value chain. The true water footprint is hidden further upstream: power production accounts for 54%, and chip manufacturing accounts for 42%.

The training and inference of large AI models require massive computing power, and behind that computing power is electricity consumption. The production of electricity (whether coal-fired, nuclear, or hydro) inherently relies on water resources. Projections show that by 2050, the water demand of the AI value chain will grow by 129%, adding approximately 30 trillion liters of annual water consumption (Chemical Engineering, 2026).

When discussing "AI's energy consumption," we are actually discussing "AI's water consumption." Every conversation with a large model, every generated image, is consuming the reserves of some distant reservoir. For China, the explosive growth of the AI industry will, within the same time window, squeeze the same already overdrawn account alongside semiconductor expansion, urbanization processes, and precipitation fluctuations caused by climate change.

Most of this new demand for 30 trillion liters will fall on the already overburdened northern power grids and wafer fabs. A dimension completely missing from public discussion is that while people stare at the PUE (Power Usage Effectiveness) of data centers, they ignore that WUE (Water Usage Effectiveness) is the critical issue.

The Failure of Pricing

Faced with severe water resource challenges, China's response strategy has long favored supply-side engineering miracles over demand-side price mechanisms. The South-to-North Water Diversion Project is a feat in human engineering history that has indeed relieved the north's urgent needs, but it has masked the true root cause.

Current signals are complex. On one hand, during the "14th Five-Year Plan" period, China achieved "zero growth" in total water consumption, and water consumption per 10,000 yuan of GDP dropped by 20% (Alwihdainfo, 2025). The market size of the water-saving industry has exceeded 760 billion yuan, which seems to be a reflection of efficiency improvements.

On the other hand, water prices remain low. In most northern cities, industrial and residential water prices are far from covering the full cost, including resource scarcity, environmental governance, and water diversion projects. Low water prices send the wrong signals to the market, instead stimulating the expansion of high-water-consuming industries like semiconductors and AI in water-scarce regions.

This has led to a classic Jevons Paradox: the improvements in water use efficiency brought by technological progress are offset by the growth in total demand brought by scale expansion. The water delivered by the South-to-North Water Diversion is quickly consumed by new wafer fabs and data centers.

If industrial water prices in northern China fail to rise to full-cost recovery levels (including externality costs) before 2030, the over-extraction of groundwater in the North China Plain will cause irreparable geological damage.

The criterion for judging whether this prediction comes true is simple: see if the groundwater level in the North China Plain stops falling. But current trends are not optimistic. Relying on engineering water diversion to solve water scarcity is like trying to treat arterial bleeding with continuous blood transfusions while refusing to suture the wound. As long as price signals fail, any supply-side effort will eventually be swallowed by infinitely expanding demand. Water is not just a natural resource; it is an economic lever. Only when the price of water puts pressure on wafer fabs will true water-saving technologies move from paper to practice.

The Chains of Energy

In 2023, Tongwei Co. (Tongwei Co.) achieved a net profit of 13.58 billion RMB, becoming one of the most profitable companies in China's photovoltaic industry. However, by 2024, the company lost 7.04 billion RMB, and in 2025, it is expected to lose between 9 and 10 billion RMB (PV Magazine 2026), transforming from a money-printing machine into a money-shredding machine within three years.

Tongwei's experience is not a result of management failure, but a microcosm of the core contradiction in China's energy transition: clean energy has become a key driver of GDP growth, but this growth line itself is bleeding.

Also in 2025, China set two records: 78GW of newly commissioned coal power and 315GW of new solar installations. It is unprecedented in global energy history for an economy to expand fossil fuels and clean energy at maximum speed in the same year. The 78GW of coal power has a design life of 40 years and will operate until 2065, by which time China has pledged to have been carbon neutral for five years. The 315GW of solar has a capacity factor of approximately 15%, meaning its actual power generation is roughly equal to 100GW of coal power. Two tracks moving forward together, two chains, imprisoning the same captive.

The Illusion of Installed Capacity

When wind and solar installed capacity first exceeded thermal power, the media called it a historic turning point. This was a victory in statistical terms, not a replacement in physical terms.

The key lies in the capacity factor. Photovoltaic panels cannot generate electricity at night, and their efficiency drops sharply on rainy days, with an average annual capacity factor of only 15-20%; whereas coal power plants can operate around the clock, with capacity factors usually between 50-60%. Therefore, the actual power generation capacity of 1.2 billion kilowatts of solar installations is only equivalent to about 400 million kilowatts of coal power. Measuring the progress of energy transition by installed capacity is like measuring an app's daily active users by its number of registered users—the numbers are bright, but they do not reflect the true intensity of use.

Physical reality is cold and ruthless. In 2025, renewable energy installations surged, but coal power generation only decreased by 1.9%. Meanwhile, total social electricity consumption grew by 6.1%, reaching a record 10.37 trillion kilowatt-hours (China Daily 2026). The vast majority of new electricity demand is still filled by coal and natural gas because the incremental growth of renewable energy cannot even cover the margin of demand growth.

Shanxi is a vivid illustration of this misalignment. In 2025, Shanxi's clean energy installations reached 90.48GW, accounting for 55.1% of the province's total capacity, exceeding coal power for the first time (Xinhua News Agency, February 2026). This sounds like a victory, but in reality, only about one-third of the electricity consumed comes from clean energy. The gap between installed capacity and actual power generation is the physical cost of the capacity factor. More notably, between 2021 and 2025, while Shanxi retired 4GW of old coal power units, it commissioned six new million-kilowatt-class coal power units. A national advanced technology photovoltaic demonstration base was built on the subsidence area of the Datong coal mine, with solar panels covering the scars left by the coal mine. Shanxi supplies power to 24 provincial-level administrative regions; in the same year that clean energy installations exceeded coal power, the absolute installed capacity of coal power also grew.

Injecting both coal power and solar energy into an already near-saturated grid brings not cleaner electricity, but more discarded power and a larger peak-shaving gap. Stepping on the accelerator and the brake at the same time keeps the car moving, but the lifespan of the gearbox is being shortened.

The Engine of Loss

Analysis by CREA (Center for Research on Energy and Clean Air) for Carbon Brief shows that if clean energy investment were excluded, GDP growth in 2025 would drop from 5% to approximately 3.5% (Caixin Global 2026). The total scale of the clean energy industry reached 15.4 trillion RMB, doubling within three years, with more than 90% of new net investment flowing into this field. Clean energy is no longer just a tool for carbon reduction, but a key support for macroeconomic growth.

This engine runs by burning capital, and Tongwei's experience is a microcosm of the industry. In 2021, polysilicon prices soared to over 300 RMB per kilogram, and Tongwei made a fortune with its cost advantage. Subsequently, the entire industry expanded production on a large scale. By 2025, polysilicon capacity exceeded demand by more than twofold, and prices fell below 40 RMB per kilogram, lower than the cost line for most companies. Tongwei's gross profit margin plummeted from 36% in 2022 to negative figures. The entire solar industry lost approximately 60 billion USD in 2025 (Morningstar 2026), and the stock price of LONGi Green Energy (LONGi) has fallen more than 80% from its 2021 peak.

Overcapacity brings price collapses, and price collapses lead to massive corporate losses, yet investment cannot stop in order to maintain local GDP and employment. In 2024, the five leading companies—LONGi, Trina, Jinko, JA Solar, and Tongwei—laid off a total of approximately 87,000 people, with an average layoff rate of 31% (EconoTimes 2026), and more than 40 photovoltaic companies went bankrupt, delisted, or were acquired. Local governments need the tax revenue and employment data from photovoltaic industrial parks, banks need growth in loan volume, and statistics bureaus need the support of investment data. Thus, even if Tongwei loses 10 billion, photovoltaic industrial parks in Guizhou are still attracting investment, and silicon material factories in Inner Mongolia are starting operations as usual. Business schools call this the "sunk cost fallacy," but in the context of China's political economy, it is called "maintaining growth."

The causal chain is clear: overcapacity pushes component prices below the cost line, corporate losses erode cash flow, banks tighten credit exposure to the photovoltaic industry, new investment slows down, local governments lose the tax and employment brought by new energy industrial parks, GDP growth comes under pressure, fiscal revenue decreases, and subsidy capacity is further weakened. When the clean energy industry is forced to contract investment due to losses, it is not just carbon reduction goals that are affected, but also GDP growth. Once the treasury can no longer afford to pay for the losses, the engine of growth will stall and energy transition will stagnate simultaneously.

Two Out of Three

In energy transition, there exists an "impossible trinity": rapid coal phase-out, guaranteed power supply, and controlled transition costs. Any country can only achieve two of these.

Germany chose "rapid coal phase-out + guaranteed supply," and the price paid was extremely high costs. The Energiewende plan cost nearly 700 billion Euros (The European), and due to the shutdown of nuclear power, it had to rely on expensive natural gas, causing carbon emissions to actually rise. Japan, after the Fukushima nuclear accident, chose "guaranteed supply + controlled costs," resulting in a stagnant coal phase-out, with only 14 of its 54 reactors restarted and the proportion of fossil fuels remaining high.

China chose "guaranteed supply + controlled costs." In the face of 10.37 trillion kilowatt-hours of electricity consumption (more than twice that of the United States), any supply gap is unbearable, so the coal phase-out has been long delayed. In 2025, in addition to the 78GW newly commissioned, another 161GW of new coal power projects were submitted or had their applications restarted (Eco-Business 2026), equivalent to 13% of existing coal power capacity.

This is not a "temporary measure for the transition period," but a deep-seated choice. Policymakers are well aware that until energy storage technology achieves a breakthrough, cheap and stable coal power is the only cornerstone for maintaining the "World's Factory." Germany's experience shows that even one of the world's wealthiest economies finds it difficult to transcend physical constraints in the speed of coal phase-out. Japan's experience shows that once nuclear power is abandoned, the lock-in effect of fossil fuels will last for decades. China has learned from both countries and chosen a third path, the cost of which is that the carbon neutrality commitment becomes a long-term promissory note.

Chokepoints and Substitutes

The Malacca Strait carries about 80% of oil imports. In 2025, China's crude oil imports reached a record 11.6 million barrels per day (Columbia CGEP 2026), which means about 23 supertankers pass through the Malacca Strait toward Chinese ports every day, each one a mobile strategic vulnerability. In February 2026, the Malaysian Maritime Enforcement Agency intercepted two crude oil tankers 24 nautical miles west of Muka Head, Penang, discovering that the two ships were conducting a ship-to-ship transfer, and seized crude oil worth 512 million Ringgit (about 130 million USD) and 53 crew members; one ship came from the direction of Iran, and the other was heading toward China. Malaysia subsequently quietly released the two ships with fines of 100,000 to 200,000 Ringgit (gCaptain February 2026). The hypothetical scenario in textbooks has just become real news.

Although electric vehicle penetration is increasing, the replacement of internal combustion engine vehicles still takes time. Physical inertia dictates that in the window between 2025 and 2040, oil import dependency will remain above 70%. If the Malacca Strait were blocked for more than 30 days due to geopolitical conflict, strategic oil reserves (covering about 80 to 90 days) would be exhausted within 3 months, at which point industrial output would be forced to cut by 20-30%.

The capacity of alternative routes is far from enough to fill the gap. The Sino-Myanmar Pipeline (Sino-Myanmar Pipeline) has a designed annual oil transmission capacity of 22 million tons, or about 440,000 barrels per day, but its actual utilization rate has been below 50% for a long time since it began operations in 2017 (Reuters 2023). The Gwadar Port (Gwadar Port) of the China-Pakistan Economic Corridor has a designed throughput of 400 million tons per year, but its actual throughput in 2024 was less than 1 million tons, with a utilization rate below 0.3% (Dawn 2025). The combined actual capacity of these two alternative routes is less than 5% of total imports. 440,000 barrels per day plus the almost negligible oil transmission from Gwadar Port, compared to the total import of 11.6 million barrels per day, makes the alternative solutions almost purely symbolic.

To hedge against this risk, China is betting on critical minerals. 58% of global lithium processing, 65% of cobalt processing, 91% of rare earth refining, and 92% of permanent magnet manufacturing are concentrated in China (IEA/USGS 2025). China Molybdenum (CMOC) owns two mines in the Democratic Republic of the Congo, Tenke Fungurume and Kisanfu; in 2024, their cobalt production reached 113,000 tons, accounting for about 45% of total global production (CMOC 2024 Annual Report), which is seen as a strategic trump card in the era of energy transition.

The dependency chain has simply shifted from the "Strait of Hormuz" to the "mines of the DRC" and "lithium mines in Australia." Although China Molybdenum's cobalt production is high, the ore is still buried in the soil of the DRC, and transportation relies on Indian Ocean shipping routes. In February 2022, the Lubumbashi Commercial Court suspended China Molybdenum's management of Tenke Fungurume on the grounds of "concealing copper reserves," and the state-owned mining company Gecamines implemented an export blockade. China Molybdenum eventually paid an 800 million USD settlement fee, plus a commitment of at least 1.2 billion USD in dividends during the mine's operation period (Mining Technology 2023). In February 2025, to raise collapsed cobalt prices, the DRC directly banned cobalt exports, and China Molybdenum and Glencore simultaneously declared force majeure. The ban lasted until October before being changed to a quota system, with an annual limit of 96,600 tons, less than half of 2024 global production. The US IRA Act and the EU's "Critical Raw Materials Act" are pushing for supply chain localization. The window of absolute dominance in the clean energy supply chain may only last 10 to 15 years, after which it will face a "peak demand" problem similar to that of current oil-exporting countries.

Falsifiability conditions: If oil import dependency falls below 60% before 2030, or if the actual capacity of the Sino-Myanmar pipeline plus Gwadar Port exceeds 3 million barrels per day, the Malacca risk will be substantially reduced. However, current data does not support such an optimistic scenario.

The Carbon Debt of 2065

The 78GW of coal power newly commissioned in 2025, combined with projects under construction and application, brings a massive time mismatch: these power plants have a design life of 40 years, meaning they will operate until 2065.

To fulfill the 2060 carbon neutrality commitment, these power plants built in 2025 must be forced to shut down early after 35 years of operation, or be equipped with carbon capture and storage (CCS) equipment that is costly and commercially unviable. Trillions of RMB in assets will become stranded assets (Stranded Assets) in the future.

If these power plants are allowed to run until the end of their design life to preserve asset value, the 2060 carbon neutrality goal will be physically impossible to achieve. The International Energy Agency (IEA) estimates that if existing global fossil fuel infrastructure operates to its design life, its locked-in carbon emissions already exceed the carbon budget for the 1.5°C temperature control goal. China's coal power installations account for more than half of the global total; continuing to add 78GW plus 161GW of applications on this base makes exceeding the carbon budget not a matter of probability, but a matter of arithmetic.

Tongwei's polysilicon is being sold at a loss, and the chimneys of coal power plants will continue to smoke until 2065. The cement and steel of 2025 are casting the default notices of 2065.

The Productivity Wall

In 2024, China's total social Research and Development (R&D) expenditure reached $785.9 billion (PPP), surpassing the United States' $781.8 billion for the first time in total volume. In the same year, the growth rate of Total Factor Productivity (TFP) fell below 1% (Oxford Economics 2023). Prior to 2007, this figure maintained a level of around 4%. However, as an economy's R&D investment reaches the global top, the efficiency growth per unit of input is shrinking. This seems contradictory but is actually precise: the world's largest R&D budget supports a massive "technological digestion system" where basic research accounts for only 6-7%. In 2025, 12.22 million university graduates flooded the job market, and the youth unemployment rate for those aged 16-24 was recorded at 16.5%, while employer demand for vocational school graduates is rising. The most capital is invested, the most highly educated talents are produced, yet the harvest is the lowest efficiency growth and the most severe mismatch. This wall is not built of capital; it is built of institutions.

Digestion System and Innovation Engine

The $785.9 billion R&D expenditure is a highly misleading victory. While the media cheers for "surpassing the United States," what is ignored is the internal structure of the spending. Of the $781.8 billion in the United States, approximately 15-17% flows into basic research—those "useless studies" that show no commercial value at the moment but can redefine industrial boundaries a decade later.

Translated into specific amounts, the United States invests about $120 billion annually in "0 to 1" breakthroughs, while China invests about $47 billion.

The vast majority of the remaining 93% flows into experimental development, which is a form of "technological digestion"—converting existing principles into products or fine-tuning existing products. It is like a restaurant whose ingredient procurement volume is the highest in the city, but 80% of the budget buys MSG rather than core ingredients. This investment structure determines that China is extremely good at driving the cost of existing technologies to the extreme, but often appears powerless when opening up new tracks.

In 2025, data from the China National Intellectual Property Administration (CNIPA) showed that the total number of patent grants for the year was 3.0995 million, a year-on-year decrease of 16.18%, with utility model patents decreasing by 27.07% (National Law Review 2026). This is not an innovation recession, but the result of regulators actively squeezing out bubbles. Over the past decade, a large number of low-quality patents intended to defraud subsidies and tax incentives flooded the system; the current decline is a return to common sense. Even after squeezing out the water, the density of high-value invention patents (such as triadic patent families) remains far lower than that of the US and Japan. The catch-up in quantity has ended, but the catch-up in quality is stuck on the shortcoming of basic research.

To increase the proportion of basic research to 15%, what is needed is not more money, but an institutional environment that tolerates failure and a long-cycle evaluation system. In a system that demands annual or even quarterly returns, few are willing to pay for possibilities ten years down the line.

The Cost of Mismatch

Economists Chang-Tai Hsieh and Peter Klenow once proposed a classic framework: if developing economies could reallocate resources to the most efficient firms, actual output could double. In China, mismatch is not an accidental market failure but a direct result of policy preferences.

Data from the Peterson Institute for International Economics (PIIE) shows that the Return on Assets (ROA) of State-Owned Enterprises (SOEs) has long been maintained at 2-3%, while private enterprises, even during the difficult financing years of 2024 to 2025, maintained an ROA in the 8-10% range. Yet, credit resources have long tilted toward inefficient sectors. It is like a football team where the coach insists on keeping the fastest striker on the bench while letting the goalkeeper play center-forward, on the grounds that the goalkeeper is "more obedient" or "more stable." When TFP drops from 4% to 1%, it is not that the players' (enterprises') abilities have declined, but that the logic of the lineup has shifted from "winning the game" to "control."

Another hidden mismatch comes from "sanctions-driven innovation": in 2025, Naura rose from 8th to 5th in the global ranking of chip equipment manufacturers, and DeepSeek demonstrated astonishing model efficiency under computing power constraints (China Daily 2026). Such cases are often cited as evidence that "blockades are failing," but from the perspective of physical limits, this may be an expensive "local optimum."

When enterprises are forced to use domestic 28nm lithography machines to produce 7nm chips through multiple exposures, they indeed achieve "autonomous control" technically, but economically, it is a regression in efficiency. DeepSeek V3 trained a 671-billion-parameter model using 2,048 castrated NVIDIA H800s (the export-compliant version of the H100), claiming a cost of only $5.576 million. SemiAnalysis estimates the actual total investment to be about $1.6 billion, including hardware procurement and operating expenses. Huawei's Ascend 910C was designed with 50 cores, but after mass production, only 24 could be activated, with performance only 1.2 times higher than the first-generation 910, while industry leaders' performance tripled during the same period. These cases demonstrate a special form of innovation: squeezing extreme efficiency under constraints. The problem is that while this type of defensive innovation is strategically necessary, it brings huge opportunity costs economically. Enterprises are investing capital that should have been used to explore next-generation technologies into the process of "reinventing the wheel."

Industries may become locked into sub-optimal technological paths, with the Japanese memory chip industry of the 1980s being the most precise cautionary tale. In 1976, the Ministry of International Trade and Industry (MITI) coordinated five companies—NEC, Toshiba, Hitachi, Fujitsu, and Mitsubishi—to establish the VLSI Joint Research Project. By the mid-1980s, Japanese companies controlled about 80% of the global DRAM market. The 1986 U.S.-Japan Semiconductor Trade Agreement set floor prices for exports, weakening Japan's price competitiveness. Samsung adopted a counter-cyclical investment strategy: expanding aggressively when Japanese companies cut investment due to the recession. In 1992, Samsung surpassed NEC to become the world's largest DRAM producer. Japanese companies were locked in vertically integrated keiretsu structures, unable to match Samsung's focus and aggression. In 1999, the DRAM divisions of NEC and Hitachi merged into Elpida, Japan's last independent DRAM manufacturer. In February 2012, Elpida filed for bankruptcy and was acquired by Micron for $2 billion in 2013. The fall from 80% to 0% took 25 years.

The Mismatch of 12.22 Million People

In the summer of 2025, 12.22 million university graduates set a new historical high, while the unemployment rate for those aged 16-24 (excluding students) remained at a high of 16.5% (EconoTimes 2026).

This is not just a cyclical economic downturn, but a rupture between educational output and industrial demand. The Ministry of Education produces 12.22 million precision screwdrivers every year, while the manufacturing and service industries currently desperately need wrenches. According to recruitment platform data, the proportion of employer demand for vocational technical school graduates rose from 8.5% in 2023 to 11% in 2025, while the demand for master's degrees fell from 20.3% to 17.4%.

The mismatch is triggering a chain reaction. High youth unemployment suppresses household consumption expectations; meanwhile, the waste of human capital is startling. When a Master of Computer Science is forced to deliver takeout, the Bureau of Statistics views it as "flexible employment," while physics considers it "energy dissipation." High-skilled labor engaged in low-skilled work drags down the labor productivity of the entire society. A more profound impact lies in the subsequent damage to long-term TFP. When society witnesses that high education cannot be converted into high returns, families' willingness to invest in education will decline, leading to a slowdown in human capital accumulation.

When South Korea was at a similar stage of GDP per capita, its youth unemployment rate was controlled at 8-10%, and its education system was tightly integrated with industrial upgrading. What China currently faces is a "sandwich trap" where a shortage of high-end industrial positions coexists with a shortage of low-end labor.

The $14,005 Threshold

The World Bank sets the threshold for high-income countries at a GNI (Gross National Income) per capita of $14,005; China's current figure fluctuates between $13,000 and $14,000 (World Bank 2024 World Development Report). On the surface, crossing it seems just a step away.

In the past 50 years, only 34 out of 108 middle-income economies worldwide have successfully crossed this threshold. The list of successes includes South Korea (50 million people), Poland (38 million people), and Chile (19 million people). No economy with a population exceeding 100 million has completed this step after World War II.

China's special difficulty lies in its scale. South Korea could rely on the global market to absorb excess capacity and cross the threshold through export-led growth. China's exports now account for more than 14% of the global share, and even exceed 30% in several manufacturing fields. Physical space is no longer sufficient: the global market can hardly accommodate an export-oriented economy ten times the size of South Korea expanding in the original way. The rise of trade protectionism is, in the final analysis, a physical reaction of the outside world to China's scale.

If TFP growth cannot return to above 2% by 2027, the middle-income trap will no longer be an academic hypothesis but a clinical diagnosis. The criteria are very specific: the core lies in whether GNI per capita can stand stably above $14,005 by 2028 (excluding exchange rate fluctuations) and whether the ratio of consumption to GDP can recover from the current 39.9% to over 45%. If these two indicators are not met, China is likely to remain for a long time at a position "one centimeter before the high-income threshold."

The Curse of 39.9%

All discussions about productivity eventually converge on one number: 39.9%. This is the share of private consumption in GDP in 2025 (Michael Nicoletos 2026). In contrast, the global average is about 60%, the United States is 68%, and even Japan, with its high savings rate, is at 55%. Even more unsettling is that 39.9% is not only lower than international standards but even lower than the historical average since 1952 (49.4%). Seventy years have passed, and the proportion of wealth that Chinese people receive from what they have created has actually decreased.

This indicates that the massive wealth brought by productivity improvements over the past few decades has not been proportionally converted into residents' purchasing power but has been redirected back to the investment side through various mechanisms (land finance, low interest rates, corporate retained earnings).

A death spiral in physics: to maintain growth, the system invests resources into infrastructure and capacity building (investment). Expanding capacity requires more consumption to absorb it. Low resident income leads to insufficient consumption and overcapacity, forcing the system to make more investments to create demand. Continuing to add salt to an already saturated solution results not in a more concentrated solution, but in precipitation.

When TFP growth stagnates, the marginal effect of pure capital accumulation (investment-driven) has approached zero. To break this wall, residents must be allowed to take more, and the government and enterprises must take less. This is no longer an economic calculation, but a political one. Until then, 39.9% is the most precise measurement of the physical limit of the Chinese economy.

The Counterforce of Exports

On February 14, 2026, WTO Director-General Ngozi Okonjo-Iweala broke the conventions of diplomatic rhetoric at the Munich Security Conference. Facing a room full of trade representatives from various countries, she directly cited global trade data released that week: "The export-oriented model that has driven China's growth for the past 40 years can no longer drive growth for the next 40 years. A trade surplus of $1.2 trillion is unsustainable because the rest of the world cannot absorb it."

In that same week, the Penn Wharton Budget Model at the University of Pennsylvania showed that although the effective U.S. tariff rate on China had retreated from its peak of 51% in May 2025, it remained high at 30%. The EU's anti-subsidy duties of 17-36% on Chinese electric vehicles had been in effect for a year. Even in Latin America, considered a "friendly market," Mexico imposed tariffs of up to 50% on Chinese automotive products, household appliances, and clothing (Mexico imported 625,187 vehicles from China in 2024). Brazil eliminated tax exemptions for parcels under $50 and increased import taxes on electric vehicles. Turkey went even further: starting January 7, 2026, it completely abolished tax-free allowances for cross-border e-commerce, prompting Temu and Shein to immediately suspend all sales to Turkey.

The $1.189 trillion trade surplus recorded by China in 2025 exceeded the entire 2024 GDP of Saudi Arabia. Exports grew by 5.5% and contributed one-third of that year's economic growth—the highest share since the 1997 Asian financial crisis—while imports grew by only 0.5% during the same period, nearly stagnating.

Newton's Third Law applies to international political economy just as it does to physical mechanics: for every action, there is an equal and opposite reaction. The export shockwave of $1.189 trillion has triggered an equivalent amount of protectionist backlash globally.

The Vulnerability of 1.189 Trillion

Under normal circumstances, a trade surplus at the level of $1.189 trillion would be seen as the ultimate proof of manufacturing competitiveness. However, in the context of 2026, it serves as a dangerous warning signal.

The essence of a surplus is that domestic production exceeds domestic consumption. When a surplus reaches more than 10% of GDP, there are only two possibilities: either the country is a resource exporter with a small population like Saudi Arabia or Norway, or it is a manufacturing nation forced to pour its capacity outward due to severely shrinking domestic demand. Saudi Arabia exports oil from the ground; China exports products from factories. The difference is that oil exports do not shut down factories in importing countries, whereas manufacturing overcapacity does.

The data paints a picture of severe imbalance: in 2025, exports grew by 5.5%, while imports grew by only 0.5%. Most of that meager 0.5% increase was for components and raw materials used in processing trade, with very little involving final consumer goods. Household consumption as a share of GDP remained stuck at 39%, far below the global average of 60%. The operation of China's economic machine is increasingly dependent on the purchasing power of the outside world, while the supporting force of the internal cycle is daily diminishing.

When a massive economy has a trade surplus exceeding one-tenth of its own GDP, primarily composed of industrial goods, it moves beyond the realm of a mere trade partner and transforms into a systemic risk. The WTO Director-General's warning was not alarmist; it was a statement of physical limits: the global market does not have enough space to accommodate a second "World's Factory," especially when that factory's local customers stop placing orders. The larger the surplus, the deeper the dependence on the external environment, and the lower the system's anti-fragility.

The Report Card of Dual Circulation

The "Dual Circulation" strategy was officially proposed in May 2020, with the core logic of "making the domestic cycle the mainstay." Four years later, a precise report card can be given based on 2025 economic data.

The report card shows that the strategy failed to land as expected. The share of exports in GDP growth climbed from about 20% in 2020 to 33% in 2025, reaching its highest point since 1997. Meanwhile, household consumption as a share of GDP dropped slightly from about 40% to 39.9%. Retail sales growth for the full year of 2025 was only 4%. The internal cycle failed to become the mainstay; its relative share even contracted.

This phenomenon stems not from execution errors but from a structural "Impossible Triangle." To jumpstart domestic demand, the proportion of household income in GDP must be increased. In a model with a fixed GDP growth target, increasing the household share necessarily means reducing the share of government and enterprises—specifically, reducing infrastructure investment and manufacturing subsidies. Yet, investment remains the primary means of maintaining GDP growth, solving employment, and repaying local debt.

Decision-makers face a brutal arithmetic problem: to maintain economic growth of around 5%, they must rely on high investment. High investment leads to capacity expansion, and without a substantial increase in household income, the domestic market cannot absorb this new capacity. Thus, the only way out is export.

Four years after "Dual Circulation" was proposed, the Chinese economy looks more like a one-way valve than ever. Massive capacity pressure can only be released outward. The internal cycle has failed to turn; it hasn't even had its wheels fitted.

Plaza Accord 2.0 Will Not Happen

History rhymes but does not repeat. Although Japan's trade frictions in the 1980s are often used as a reference for China's current situation, such analogies ignore a key structural difference.

In 1985, Japan's trade surplus with the U.S. was about $560 billion, accounting for about 4% of its GDP. This imbalance eventually led to the signing of the "Plaza Accord," and the yen appreciated by 50% within two years. Japan released the "counterforce" of trade pressure through yen appreciation, at the cost of a domestic asset bubble and the subsequent "Lost Decades."

In 2025, China's trade surplus reached $1.189 trillion, accounting for about 6% of GDP—more than 20 times the scale of Japan's back then. China maintains strict capital controls and an exchange rate formation mechanism dominated by the central bank. The RMB will not be forced to appreciate by 50% like the yen was. Decision-makers have already indicated they would rather endure tariff barriers than relinquish control over the exchange rate.

This leads to a new physical phenomenon: when the exchange rate pressure release valve is welded shut, the counterforce does not simply disappear; it inevitably seeks a new breakthrough point.

The breakthrough points are tariffs and non-tariff barriers. Effective U.S. tariffs on China surged to 51% at one point in 2025, the EU imposed anti-subsidy duties on electric vehicles, and even developing countries like India, Brazil, and Indonesia began to build high walls. Japan's pressure back then was released through price; China's pressure now is released through quantity. The channels are different, but the laws of physics are the same.

Loss-Making Globalization

The solar industry provides a perfect micro-sample of what happens when "investment-driven" hits "market limits."

The financial statements of Tongwei Co., Ltd., the world's largest producer of high-purity crystalline silicon and solar cells, are a roller-coaster thriller. Net profit was 13.58 billion yuan in 2023, turned into a loss of 7.04 billion yuan in 2024, and is projected to lose 9 to 10 billion yuan in 2025 (PV Magazine 2026). Total losses for the entire Chinese clean energy industry in 2025 are estimated to reach $60 billion (Morningstar 2026). It is precisely this industry, with a combined loss of $60 billion, that contributed one-third of that year's GDP growth.

Enterprises lose at the micro level, while the nation grows at the macro level.

This model produces a first-order effect: Chinese companies capture the global market at prices below cost (TOPCon module FOB prices as low as $0.096/watt). At this price, a high-tech solar panel sells for less than a large pizza—the difference being that pizza shops don't need government subsidies to survive.

Then comes the second-order effect: local solar manufacturing in Europe and the U.S. is destroyed. Qcells closed its factory in Germany, Meyer Burger's expansion plans in the U.S. stalled, and the diversity of the global supply chain vanished.

Most dangerous is the impending third-order effect. When Chinese companies are finally forced to cut production or go bankrupt due to long-term losses, a massive gap will appear in the global solar module supply. By then, local capacity in Europe and the U.S. will no longer exist, and rebuilding the supply chain will take years. Overcapacity has moved beyond China's own problem; it is reshaping the global industrial structure, hijacking the global energy transition onto an extremely fragile, loss-driven single source of supply.

Shein's experiment in Brazil exposed another layer of vulnerability. In 2023, Shein pledged to invest $150 million with the goal of partnering with 2,000 Brazilian factories to create 100,000 jobs. But by the end of 2023, only 336 had signed. In February 2026, Reuters reporters visited all 12 states where Shein once had partners and found only one factory still producing for Shein. Fernando Pimentel of the Brazilian Textile Association summarized: "Working in Brazil is a different thing than working in China." The core competitiveness of the Chinese export model is built on a network of thousands of closely collaborating factories in the south. Such a supply chain cannot be replicated anywhere else. When exports encounter barriers, "transferring capacity" is by no means a realistic option.

The 20% Ceiling

There exists an invisible but hard physical ceiling in political economy: when a country's industrial exports exceed a 20% share of the global market, protectionism shifts from a policy option to a survival instinct.

When joining the WTO in 2001, China's share of industrial exports was only 5%, but by 2025, that figure had exceeded 20%. In the industrial categories tracked by Oxford Economics, China is the largest exporter in nearly 60% of categories. Historically, no country, including 19th-century Britain or 20th-century America, has been able to maintain a free trade environment after reaching this level of market share.

An IMF working paper from January 2026 pointed out an interesting phenomenon: "Export Front-loading." During windows where tariff expectations are rising, Chinese companies accelerate shipments. The abnormal surge in export data in the second half of 2025 was a result of this front-running behavior, though this is only a temporary escape.

If the trade surplus does not drop below $800 billion by 2028, the coverage rate of global trade barriers against China will exceed 50%. This prediction is based on simple physical deduction: when the water flow is too large and the pipe is too narrow, the pipe will not only clog but burst. In 2024, 4.6 billion small parcels entered the EU, 91% of which came from China, with the number doubling annually. Starting July 2026, the EU will impose a flat tariff of $3 per item on all cross-border parcels, eliminating the previous $150 tax-free allowance. Indonesia banned imports of 12 categories starting January 1, 2026, effectively blocking Temu from entering the market. Tariffs and non-tariff barriers set by the U.S., EU, India, Brazil, Mexico, Turkey, and Indonesia already cover about 30% of total exports. As the surplus continues to stay at the trillion-dollar level, it is only a matter of time before this proportion breaks 50%.

For the Chinese economy, the real challenge has moved beyond how to manufacture more products and has transformed into how to place that capacity when the world no longer buys it.

The Elderly in the Room

On February 15, 2023, thousands of white-haired elderly people gathered outside Zhongshan Park in Wuhan, with some singing "The Internationale."

The trigger for the event was the reduction of monthly allocations to personal medical insurance accounts from 260 yuan to 83 yuan, a decrease of 68%. One protester told the Financial Times: "This is robbery. The government wants to use my money to subsidize others." Another told the New York Times: "The socialist country was won by our older generation." Police cordoned off nearby subway stations, and amidst the jostling, some attempted to climb over barricades. Netizens dubbed this action the "Silver Movement," echoing the "White Paper Movement" where young people held up blank sheets of paper three months prior.

The Wuhan medical insurance reform has its technical rationality: transferring funds from personal accounts into a collective pool to expand the scope of outpatient reimbursement. However, what retired elderly people felt was not an "expanded reimbursement scope," but rather "177 yuan less per month." For urban retired employees with an average monthly pension of 3,326 yuan, 177 yuan represents a 5.3% reduction in income. Meanwhile, for rural elderly people with an average monthly pension of only 179 yuan (CFR 2025), 177 yuan is nearly equivalent to an entire month's income.

The gap between urban and rural areas is not 15-fold, but as high as 18-fold. The situation in Wuhan is not isolated; it is a microcosm.

In 2025, China's births totaled 7.9 million, while deaths reached 11.29 million, a net decrease of 3.39 million. This is the largest annual population decline since the Great Leap Forward famine (Gulf Times/NBS 2026). The figure of 7.9 million needs to be understood in a historical context: it is equivalent to the birth levels of 1738, the early years of the Qianlong Emperor's reign in the Qing Dynasty. However, China's total population in 1738 was approximately 150 million, whereas it is now 1.4 billion. Using an 18th-century birth scale to support a 21st-century retired population is an arithmetic problem that is almost unsolvable.

In the same year, the population over 60 reached 323 million. The total balance of the social security fund was 10.2 trillion yuan, which sounds like a massive figure. But if distributed equally among 323 million elderly people, it amounts to only 31,600 yuan per person. Calculated at an average monthly pension of 3,000 yuan for urban employees, this "huge sum" can only sustain payments for nine months. The Chinese Academy of Social Sciences (CASS) predicted in 2019 that pension surpluses would be exhausted by 2035. Six years later, this prediction has not been altered by any official data; the only variable is that the birth rate is declining faster than the pessimistic models of that time. The delayed retirement plan has already been launched: the retirement age for men will be extended from 60 to 63, a plan to be completed over 15 years, with an average delay of about two months per year.

The net annual population decrease of 3.39 million stands in stark contrast to the policy adjustment of delaying the retirement age by two months each year. The tear gas on the streets of Wuhan has become the first footnote to this arithmetic problem.

The Arithmetic of 10.2 Trillion

The 10.2 trillion yuan (approximately $1.46 trillion) social security fund balance (People's Daily 2026) is often cited as evidence of system security. However, this view overlooks a critical issue: assessing the security of a pension system cannot rely solely on how much water is left in the reservoir; one must focus on the flow rate difference between the inlet and the outlet.

In 2025, social security and employment expenditures grew by 6.7% year-on-year, reaching 4.4 trillion yuan (Yicai Global 2026), reflecting the flow rate of the outlet. Meanwhile, the situation at the inlet is even more severe: the base of the contributing population (births) plummeted from 17.86 million in 2016 to 7.9 million in 2025, meaning the supply of new labor will decrease by 56% over the next 20 years.

The current balance can only maintain existing payment levels. CASS's 2019 model predicted the fund would be exhausted by 2035, provided the total fertility rate remained around 1.6. However, the total fertility rate in 2025 has fallen below 1.0. When the number of contributors decreases faster than the number of recipients grows, the depletion of reserves will accelerate exponentially. Without large-scale transfers of state-owned assets or an economic growth rate maintained above 4% to support wage growth and the contribution base, 2035 may be an overly optimistic timeline.

The so-called "abundance" is actually a static misjudgment that ignores deteriorating flow.

The Reservoir's Inflow

The pay-as-you-go pension system is essentially similar to a legal Ponzi scheme: the pensions of current retirees are paid by current workers, rather than from past accumulations. This structure works well when the population pyramid is triangular: five workers support one retiree, with each person only needing to yield 20% of their output.

China's demographic structure is transforming into an inverted pyramid. The dependency ratio is rapidly shrinking from 5:1 toward 2:1. When two workers must support one retiree, the burden on each worker will soar from 20% to 50%. This is not just a matter of management efficiency, but a matter of fluid dynamics. Viewing the pension system as a reservoir, the inflow depends on births from 20 years ago, while the outflow is determined by the current rate of aging.

From 2016 to 2025, the source of inflow (births) decreased by 56%, while the demand side (population over 60) surged from 264 million to 323 million. Adding water to a bucket with a hole in the bottom, the rate of filling can never keep up with the rate of leaking; the water level will only continue to fall. The delayed retirement policy attempts to alleviate this pressure by extending working years, but a pace of delaying by two months each year, compared to the rate of worsening dependency ratios, merely reduces the force of impact when hitting the wall, without being able to change the outcome of hitting it.

The 18-Fold Gap

The massive gap between urban and rural pensions is a projection of the household registration (hukou) system onto the financial level. The average monthly pension for urban employees is approximately 3,326 yuan, while the average monthly basic pension for rural residents is only about 179 yuan (CFR 2025), a disparity of 18-fold.

This gap brings about vastly different living conditions. For example, Wang Fengqin, 70, a retired farmer in Wudaogang Village, Heilongjiang, receives a monthly pension of 2,000 yuan, but a single hospitalization could cost 1,000 yuan. Wudaogang Village has shrunk from 400 households in 1976 to about 70 today. Wang Fengqin's two sons live in Hegang, 15 kilometers away; one drives a truck and the other is a firefighter, each earning about 3,000 yuan a month and each having one child (Reuters/KTWB 2023). Zhang Zhidong, a 78-year-old retired teacher in Harbin with a monthly pension of over 3,000 yuan, says it is "enough for food, but not enough for medical treatment."

The situation in Heilongjiang can be seen as a warning; this former industrial powerhouse experienced a pension deficit as early as 2016, becoming the first province in the country unable to pay pensions on its own. In 2021, the province had only 100,000 births and 460,000 deaths; over the past decade, the population decreased by 17%, and the labor force shrank by about one-third. In January 2026, the Social Security Center of Daowai District, Harbin, issued a notice: 1,014 deceased individuals were still receiving pensions, and families were required to return the funds within 30 days, or face court enforcement (China Daily 2026). Deceased individuals receiving pensions is not a system loophole, but an early signal of system collapse.

A deeper crisis lies with the 200 million flexible workers (MOHRSS 2026), including delivery riders, ride-hailing drivers, and livestreaming hosts. JD.com's announcement in March 2025 that it would pay "five insurances and one fund" for about 150,000 full-time riders became news precisely because it is an exception rather than the norm. Meituan piloted pension insurance subsidies in Quanzhou, Fujian, and Nantong, Jiangsu, with the first batch of riders receiving about 490 yuan. Although the number of personal pension accounts opened is close to 30 million, the actual contribution rate is only about one-third. Banks use small cash rewards to promote account openings; people open their piggy banks, look at the balance, and close the lid again.

The inter-provincial pension transfer payment system (Central Adjustment Fund, established in 2018) is transforming from a zero-sum game into a negative-sum game. In the past, wealthy provinces like Guangdong, Beijing, and Zhejiang subsidized Heilongjiang. As aging spreads nationwide, the original "net contributor provinces" are one by one becoming "net recipient provinces." The national ratio of active employees to retirees has dropped from 10:1 a decade ago to 5:1, and is expected to fall to 4:1 by 2030 and 2:1 by 2050. When Jiangsu and Zhejiang also begin to face revenue-expenditure gaps, this mechanism of tearing down the east wall to patch the west wall will have no walls left to tear down. Retired elderly people in Wuhan have already taken to the streets once; next time, it may not be just Wuhan.

The Bill for Getting Old Before Getting Rich

Japan is often seen as a reference point for China's aging, but this comparison may be misleading. When the proportion of Japan's population over 65 reached 29.3%, its GDP per capita was approximately $40,000 (Oxford Economics). The proportion of China's population over 60 will surpass 30% before 2035, at which time optimistic estimates for GDP per capita are only $15,000 to $18,000.

Japan was able to "grow old while rich." With its massive net foreign assets and a comprehensive social security system, Japan maintained social stability during its "lost three decades," with life expectancy and quality of life even continuing to improve. China, however, must address aging challenges that even developed countries find difficult, using the fiscal strength of a developing nation. The Japanese government's social security spending accounts for one-third of its budget, and its debt-to-GDP ratio reaches 260%; this is room for maneuver built on a high-income foundation, which China lacks.

If the basic pension replacement rate for urban employees (pension as a percentage of pre-retirement wages) falls below 35% before 2030, and rural basic pensions are not substantially increased to over 500 yuan per month, the pension system may face a crisis of trust. By then, the majority of the more than 300 million elderly people will find that social security can only maintain a survival baseline rather than a dignified life. By 2040, China's population over 60 will reach 402 million, a figure exceeding the projected total population of the United States at that time, which is 379 million (NewKerala/US Census Bureau). A single country's elderly population exceeding the total population of another superpower will be a geopolitical landscape never before seen in human history.

Robots Don't Pay Pensions

China is the world's largest industrial robot market; in 2024, more than half of the robots installed globally were in China, and over 140 companies are competing in the humanoid robot track (ainvest.com 2026). This is widely seen as a technological antidote to a shrinking labor force.

But there is a logical contradiction here: robots can solve the labor supply problem (who does the work), but they cannot solve the problem of pension funding sources (who pays the contributions). Under the current tax and social security systems, robots are not considered taxable entities. When a factory replaces 100 workers with 100 robots, output may remain the same or even increase, but the pension fund loses 100 sources of contributions.

Unless the government imposes a "social security tax" on robots (which currently has no global precedent and is highly controversial), the higher the degree of automation, the faster the contribution base of the pension system will shrink. Technological progress and social security form a contradiction here: robots are a boon for manufacturing but a hidden danger for the pension system. Hoping that the "silver economy" will offset the costs of aging is equally unrealistic, because the premise of the silver economy is that the elderly have sufficient purchasing power, and the pension crisis is precisely what is eroding that power.

Entropy Increase: The Physical Limits of a System

In 1988, Soviet economist Abel Aganbegyan wrote an absurd fact in the New Left Review: The Chelyabinsk Tractor Plant produced tens of thousands of bulldozers annually for the Far North, yet "not a single bulldozer could last a single operating season, and the annual repair cost for each unit was several times its original price." Another shoe factory, in order to meet the production targets set by the State Planning Committee, produced all shoes in only one size. Aganbegyan’s diagnosis was calm and precise: "Output in 40% of industrial sectors is actually declining, and the living standards of two-thirds of the population have begun to slide."

Three years later, the Soviet flag was lowered from the Kremlin.

On January 17, 2025, the National Bureau of Statistics announced a full-year GDP growth rate of 5.0%. Data released the same day showed that fixed asset investment fell by 3.8%, the first decline since records began in 1996; real estate investment fell by 17.2%; and December retail sales grew by 0.9% year-on-year, a three-year low. The trade surplus reached a historic high of $1.189 trillion, while import growth was only 0.5%. New solar installations reached 315GW, refreshing the global record, while new approvals for coal power also hit a record high of 161GW in the same year. The number of births fell to 7.9 million, and the population over 60 exceeded 323 million. The balance of local government debt increased by 15% year-on-year to 54.82 trillion yuan (Yicai Global 2026).

The bulldozers in Chelyabinsk could not last a single season, Tongwei's photovoltaic modules are sold below cost, and Zunyi Daoqiao's loans have been extended for 20 years. Output increases, yet value evaporates. This is not a phenomenon unique to the Soviet Union as described by Aganbegyan, but a manifestation of the laws of physics.

The Second Law of Thermodynamics has a colder name: Entropy Increase.

Dissipative Structures and Four Energy Sources

In 1971, Nicholas Georgescu-Roegen introduced thermodynamics into economic analysis in The Entropy Law and the Economic Process. The economic process is essentially a one-way flow that transforms low-entropy resources (ordered, usable) into high-entropy waste (disordered, unusable). Growth is not a perpetual motion machine; ultimately, it is a process of consuming order.

Ilya Prigogine’s theory of dissipative structures brought a glimmer of hope to this gray law: an open system can maintain or even increase its degree of order by continuously inputting energy and matter from the outside and discharging internal entropy.

Over the past forty years, the Chinese economy was able to maintain a staggering state of low entropy (high growth, high investment returns, social stability) by relying on the continuous injection of four powerful "negative entropy flows." The young labor force in the demographic dividend brought a highly ordered supply of energy, and the technology catch-up dividend made imitating existing technologies much less entropy-intensive than exploring unknown ones. As an open system, the globalization dividend allowed the Chinese economy to continuously absorb capital and demand from the outside. Furthermore, the institutional dividend released productivity suppressed by the planned system, allowing the order of resource allocation to be reorganized.

Looking back at the 2025 dashboard: the population has turned to a net decrease, meaning the energy input channel is not only closed but has begun to reverse-extract energy (the burden of elderly care). The growth rate of Total Factor Productivity (TFP) has dropped from 4% before 2010 to below 1% (Rhodium Group 2024 estimate), and the marginal returns of technological imitation have almost bottomed out. Trade barriers have escalated, and the "de-risking" of Western markets is effectively an attempt to re-close the Chinese economy. As for the institutional dividend, the trend of "the state advances while the private sector retreats" has increased internal friction within the system.

Continuing to add load to an already overloaded circuit results not in greater power, but in a tripped breaker. When the four external input channels that maintain order narrow simultaneously, the increase of internal entropy accelerates irreversibly.

Simultaneous Equations

Policymakers are accustomed to categorizing problems: demographic issues belong to the Health Commission, debt issues to the Ministry of Finance, and export issues to the Ministry of Commerce. This linear thinking is ineffective in the stage of entropy increase.

Population shrinkage reduces the pension contribution base, the pension gap squeezes fiscal space, fiscal tightening drags down investment, falling investment weighs on employment, worsening employment suppresses consumption, insufficient consumption leads to capacity spilling over into exports, surging exports trigger tariff retaliation, and after external markets close, policy returns to investment-driven growth. However, the return on capital is already extremely low, and every unit of growth consumes more debt, eventually forming a closed loop.

Any single-dimensional "solution" will trigger new problems elsewhere. To alleviate debt, the central government requires local governments to compress investment, leading to an economic slowdown and worsening employment. To stimulate employment, exports are encouraged, inviting tariff retaliation. To cope with aging, retirement is delayed, intensifying competition for youth employment. Policymakers are not facing nine separate targets, but a system of simultaneous equations. Under current parameters, this system has no positive real solution.

Japan's Mirror, the Soviet Union's Shadow

"China is not Japan" is a common placebo used by economists. The reasons are substantial: China's per capita GDP is only $12,000, far lower than Japan's $28,000 in 1990 (at the then-exchange rate), leaving huge room for catch-up; it has a unified market of 1.4 billion people; and the government's mobilization capacity far exceeds that of Japan.

However, the most dangerous misjudgment lies precisely here: after the bubble burst in 1990, Japan's nominal GDP fell from $5.45 trillion in 1995 to $4.2 trillion in 2023, decreasing rather than increasing over 30 years. Japan stagnated in a state of "affluence." Relying on vast net overseas assets (Japan has been the world's largest net creditor nation for 33 consecutive years) and a sophisticated social security system, Japanese society maintained extremely high stability during the "Lost Decades." Tokyo subways remain punctual, convenience stores still operate 24 hours a day, and life expectancy continues to rise. Stagnation is a luxury that only wealthy countries can afford.

The risk China faces is "stagnating while poor" on the basis of "getting old before getting rich." If the physical limit is hit when per capita GDP is $12,000, it means there is insufficient wealth reserve to buffer the aging tsunami. Retired elderly in Wuhan took to the streets over a 177-yuan reduction in monthly medical insurance; in Japan, this amount isn't enough to buy a bowl of ramen.

A more appropriate historical reference might not be Japan, but the Soviet Union of the 1980s.

Soviet GDP growth slid from 6% in the 1950s all the way to 2% in the 1980s. More fatally, Total Factor Productivity (TFP) turned to negative growth after the 1970s (classic study by Easterly & Fischer). The Soviet Union's Eleventh Five-Year Plan (1981-1985) poured 827 billion rubles into fixed asset investment, a 15% increase over the Tenth Five-Year Plan, yet industrial output growth fell from 4.4% to 3.7%. More input yielding less output perfectly matches the micro-absurdities recorded by Aganbegyan at the beginning.

In 1986, Gorbachev proposed the "Acceleration Strategy" (Uskoreniye) at the 27th Party Congress, calling for an 80% increase in investment in machine building. Two years later, this strategy quietly disappeared, replaced by "Restructuring" (Perestroika). Between "Acceleration" and "Restructuring" lay only two years and the collapse of an empire.

This script has a sequel in the modern era. In the first three quarters of 2025, Russia's GDP growth was only 0.6%, driven almost entirely by the defense sector, with military spending accounting for 7.3% of GDP, "equivalent to late Soviet levels" (in the words of Defense Minister Belousov). To suppress inflation, the Russian central bank maintained the benchmark interest rate at 21%, a level that squeezed out almost all private investment. At AvtoVAZ, Russia's largest automaker in Tolyatti, Lada sales fell by nearly 25% in 2025, forcing shortened work hours and a search for export markets in Yemen and Iran. The northern steel giant Severstal reported zero profit tax for the first three quarters, causing corporate income tax revenue in the Vologda region to plummet by about 30%. The Samara region's 2026 budget deficit reached 30.1 billion rubles, exceeding 10% of fiscal revenue; the Murmansk region's deficit was 23.7 billion rubles, nearly one-fifth of revenue. National wage arrears grew 2.3 times in 2025, reaching the highest level in nine years. Russian crude oil prices fell below $35 per barrel, while the budget assumed $69. Thirty-four years after the collapse of the Soviet Union, the successor is still running on the same track: state-led investment, suppressed private vitality, and statistical figures disconnected from perceived reality.

In 2025, China, in terms of state-owned enterprises dominating resource allocation, diminishing marginal returns on investment, and rigid information feedback mechanisms, structurally resembles the characteristics of the late Soviet Union. The Second Law of Thermodynamics says: the entropy of a closed system only increases. Economists say: an open system can maintain low entropy by inputting energy. Policymakers propose the "Dual Circulation" strategy. Physicists would ask a more fundamental question: Is the system becoming more open, or more closed?

If China repeats Japan's script at $12,000 per capita, it won't be called the "Lost Thirty Years"; it will be called the "Thirty Years We Never Had the Chance to Own."

Three Scenarios: Where is the Wall?

Based on the convergence speed of physical constraints, three evolutionary scenarios for the next decade can be deduced.

Baseline Scenario (50% probability): Stagnation just after crossing the line. GDP growth gradually slides from 5% in 2025 to 3-3.5% in 2030, and further to 2-2.5% in 2035. Per capita GDP barely crosses the high-income country threshold (approximately $14,005) between 2028 and 2030, then falls into long-term stagnation. Affected by deflation, nominal GDP growth will remain lower than real GDP growth for a long time. This is a "boiling frog" state, where social pain is intense but still bearable.

Pessimistic Scenario (30% probability): Cliff-like adjustment. This could be triggered by the sudden breaking of a key constraint. It might be trade disruption due to the Taiwan Strait situation, a chain of defaults triggered by the failure of local debt swaps, or a secondary bottoming out of real estate piercing bank balance sheets. In this scenario, GDP growth falls below 2% before 2028, or even turns negative. The youth unemployment rate breaks 25%, and social instability rises sharply.

Optimistic Scenario (20% probability): Structural breakthrough. This would require a "phase transition" in the physical sense. The government implements major institutional reforms: completely abolishing the hukou system to release consumption potential, substantially advancing the mixed-ownership reform of SOEs to improve TFP, and shifting fiscal spending on a large scale from infrastructure to resident social security. If this occurs, TFP growth returns to above 2%, the share of consumption in GDP increases from 39.9% to 45%, and the Chinese economy successfully switches tracks.

To judge which track the economy is on, one must watch for the following falsifiable signals: If the GDP deflator remains negative for eight consecutive quarters before 2027, the probability of the pessimistic scenario rises to 60%. If the first official default of a standard local government bond occurs, the probability of the pessimistic scenario rises to 80%. If the share of household consumption in GDP breaks 43% before 2028, or if the growth rate of private fixed asset investment consistently exceeds that of state-owned enterprises, the probability of the optimistic scenario rises.

Resilience or Inertia

The Chinese economy still has chips to deal with challenges.

Central government debt as a percentage of GDP is only about 24% (Ministry of Finance data); even including all implicit debt, the broad government debt ratio is about 120% (IMF estimate), still lower than Japan's 260%. The market scale of 1.4 billion people remains a physical fact. In fields like new energy, electric vehicles, and parts of AI, technological accumulation is real.

The question is whether this resilience is used to drive transformation or to maintain the status quo.

If the central government increases leverage only to provide blood transfusions for ineffective infrastructure projects, it is merely delaying the onset of the disease. If the vast market is divided by the urban-rural dual structure, consumption potential will forever remain just "potential." If technological progress is confined to a few fields under the whole-nation system while the vast majority of private enterprises are stuck in low-end involution, innovation will be difficult to translate into productivity for the whole society.

The principles of physics show no mercy. If resilience is used to resist necessary adjustments, resilience becomes inertia. The greater the inertia, the more energy is required to change direction, and the greater the destructive force upon final impact.

The Ending: Time is No Longer a Friend

Over the past forty years, time was China's friend. As long as problems were pushed back, the compound interest of growth would turn big problems into small ones, and small ones into none. Zunyi Daoqiao's loans could be extended for 20 years, the pension gap could be filled by delaying retirement, and the carbon emissions of coal plants could be offset by the 2060 commitment.

Now, time has become the enemy: every day, the labor force decreases; every day, debt interest accumulates; every day, the price of Tongwei's PV modules falls; every day, the pension gap in Heilongjiang widens. Based on the analysis of the previous nine chapters, the intersection of multiple physical limits will appear between 2028 and 2032: the critical point when pension surpluses are exhausted, the final deadline for local government financing vehicle (LGFV) debt swaps, and the political ceiling for export shares.

It took Gorbachev two years to move from "Acceleration" to "Restructuring." The question is not "whether it will hit the wall"; based on current acceleration and distance, the wall is already in sight.

The conclusion of the Second Law of Thermodynamics is that a closed system eventually reaches heat death: a state of complete disorder, no temperature difference, and no flow. To avoid this ending, the Chinese economy must reopen the system—not only to international markets but, more importantly, to the suppressed creativity and consumption demand within the country. The market scale of 1.4 billion people is a physical fact, and the 39.9% consumption share is also a physical fact. The former represents potential, while the latter has become a chain. The key to unlocking the chain is not in Washington, but in Beijing.