An Organized Credit Creation Experiment

In August 2015, the Ministry of Housing and Urban-Rural Development (MOHURD) and the China Development Bank (CDB) issued a document requiring that the proportion of monetized resettlement for shantytown renovation should not be lower than 50%. However, just two years later, in August 2017, MOHURD issued another document requiring the "control of the proportion of monetized resettlement." For the same policy tool, the shift from "must use" to "cannot use" took only two years.

During this 24-month window, 3.38 trillion yuan of base money was injected into tier-3 and tier-4 cities. During this period, housing prices in these cities rose by 49%, and Country Garden's sales grew tenfold. The market overheating was not a natural occurrence but a precisely opened window of opportunity.

The Journey of 3.38 Trillion

Before 2015, real estate bailouts typically relied on commercial banks lowering down payments or interest rates. However, traditional methods faced an insurmountable obstacle: the risk control systems of commercial banks. Bankers knew well that tier-3 and tier-4 cities with net population outflows lacked repayment capacity, so credit funds could never penetrate these markets on a large scale. Data from late 2014 confirmed the banks' judgment: 80.67% of the national commercial housing inventory was concentrated in tier-3 and tier-4 cities, while the per capita disposable income in these cities was only one-third of that in tier-1 cities. The risk control models of commercial banks dictated that the places most in need of inventory reduction were the ones least likely to obtain conventional credit. To break through this line of defense, a new monetary channel had to be opened.

In April 2014, the central bank launched Pledged Supplementary Lending (PSL), an independent pipeline that bypassed the risk control systems of commercial banks. The flow of funds was clear: the central bank printed money for the CDB, the CDB loaned it to local governments and their financing platforms (LGFVs), local governments provided cash compensation to relocated households, and those households rushed into sales offices with cash in hand. In this entire chain, no link required an assessment of the borrower's repayment capacity, because the repayment obligation ultimately fell on the local government, whose repayment capacity depended on future land transfer revenue. This was a self-consistent cycle: using future land revenue to guarantee today's home purchase funds, while the very purpose of today's home purchase funds was to maintain land prices. This was not just fiscal stimulus; it was a targeted credit creation experiment.

In June 2015, the State Council issued a document establishing a three-year plan for shantytown renovation (2015-2017), aiming to renovate 18 million units. Two months later, MOHURD set the floor for the monetized resettlement ratio at 50%. The logic of this combination was tightly linked: there was a timetable, quantitative targets, and a funding channel. The proportion of monetized resettlement soared from 9% in 2014 to 29.9% in 2015, reached 48.5% in 2016, and exceeded 53.9% in 2017. Every percentage point jump corresponded to hundreds of billions of base money injected into a market that originally lacked purchasing power.

PSL was not just capital; it was base money. When 3.38 trillion yuan of PSL funds were injected into the system and amplified by the multiplier effect of commercial banks, the broad purchasing power created far exceeded the face value. After a PSL loan became a deposit for a relocated household, banks could issue new loans based on that deposit, stacking layer upon layer. This explains why those originally stagnant tier-3 and tier-4 markets suddenly saw a flood of liquidity: what was injected was not one-time capital, but monetary seeds capable of self-reproduction.

The official narrative called this "inventory reduction," but this cannot explain a simple logical paradox: if the goal was merely to reduce inventory, the most effective market mechanism would have been to lower prices.

Prices did not fall; instead, local governments and developers chose to maintain high prices and inject massive amounts of capital to create new demand. The only reasonable explanation was "protecting land prices." In 2015, land transfer fees still accounted for about 40% of local fiscal revenue. Once housing prices were allowed to clear through market mechanisms, the value of land collateral would collapse, and loans on bank balance sheets backed by land would collectively become bad debts, thereby triggering a local debt crisis. Reducing inventory through price cuts was logically sound but politically unfeasible. Inventory reduction was merely a means; preserving the valuation system of land finance was the true goal of this credit experiment.

Adaptive Evolution in the Credit Flood

In a normal business environment, a company's sales growing tenfold in six years is usually seen as a management miracle. Country Garden grew from 47.5 billion in 2012 to 550 billion in 2017, a 10.58-fold increase. Business schools credited this to the management innovation of the "high turnover" model.

This is a misreading.

Country Garden's "456 model" (opening for sale 4 months after land acquisition, recovering funds in 5 months, and reinvesting funds in 6 months) was not a universal management efficiency but a biological adaptation to a specific environment of high-concentration credit. It is like the tube worms at deep-sea hydrothermal vents; these organisms cannot survive in normal-temperature seawater but thrive in 300-degree sulfur-rich hydrothermal fluids. Tier-3 and tier-4 cities from 2015 to 2017 were such a special environment filled with PSL "hydrothermal fluid."

In this environment, the source of funds for homebuyers was not labor savings but credit injections from relocation compensation. For developers, the core competitiveness was no longer product quality or cost control, but the extreme speed of capital turnover. Whoever could turn cement into mortgageable assets faster could take a larger share of the PSL pie. Country Garden later upgraded the 456 model to a 345 model (opening in 3 months, recovering in 4, and reinvesting in 5). This was less an improvement in efficiency and more an acceleration of a survival race. When all competitors are shortening their turnover cycles, staying in place means being eliminated.

With 58% of its sales coming from tier-3 and tier-4 cities, Yang Guoqiang did not invent more advanced construction technology; rather, he most keenly captured the flow of the credit flood and evolved a survival mechanism that turned his company into a super-conduit. This mechanism required every link to be in a state of extreme tension: design cycles were compressed to the limit, construction quality gave way to progress, and pre-sales began while the building was still being piled. Because the margin for error was zero, a delay in payment for one project could drag down the entire capital chain, as the land payment for the next project relied on that specific inflow.

The foreshadowing of the tragedy lay precisely here: when the PSL tap was tightened in 2018 and environmental parameters returned to normal, this extreme adaptability instantly became a fatal vulnerability, much like tube worms leaving a hydrothermal vent. Country Garden did not die from operational failure, but from environmental normalization.

The Ponzi Logic of Inventory Reduction

This experiment left behind a confusing balance sheet: from the data, the policy seemed successful. In 2016, shantytown monetization helped digest 250 million square meters of commercial housing inventory. In those county towns where relocated households flooded into sales offices, inventories that had been backlogged for years were indeed cleared, local government land could be sold again, and a new round of land transfer fees was recorded.

However, during the same period, developers' new land reserves in tier-3 and tier-4 cities reached as high as 476 million square meters. While digesting 250 million in inventory, 476 million in new inventory was created.

This is the paradox of "inventory reduction": to digest old inventory, high land prices and high expectations must be maintained; and high land prices and high expectations inevitably incentivize developers to create more new inventory. In June 2016, housing prices in tier-3 and tier-4 cities rose by a cumulative 9%; by April 2018, this figure had become 49%. Driven by price signals, the supply side, which should have contracted, expanded frantically. Inventory reduction turned into inventory replenishment. Every "successfully" cleared city used higher land prices to attract developers to buy land, and every newly auctioned plot of land created future inventory pressure.

This was not a deviation in policy execution but an inherent feature of a Ponzi structure. In any system that relies on debt expansion to maintain asset prices, every attempt at deleveraging eventually requires the introduction of even greater leverage to prevent a collapse. The inventory crisis of 2014 gave birth to shantytown monetization; shantytown monetization created the credit contraction crisis of 2018; credit contraction gave birth to the "Three Red Lines" in 2020; and the "Three Red Lines" directly triggered the debt crises of Evergrande and Country Garden. Every dose of medicine created the next illness that required a larger dose.

Now, the bill for this experiment has come due.

The land market in 2024 shows the final outcome: the share of land acquisition by central/state-owned enterprises and LGFVs has reached 85.3%, while private real estate enterprises account for only 8.5%. This is not just a cyclical low, but an extinction of a species. The inventory clearance cycle in tier-3 and tier-4 cities has stretched to 36 months, meaning that even if supply stopped completely, it would take three years to digest existing inventory. Those families who bought houses with relocation money in 2016-2017 are now sitting on assets with continuously declining prices and almost zero liquidity.

Evergrande's total liabilities at the end of 2022 were 2.44 trillion yuan, with a two-year loss of 812 billion yuan. These astronomical figures mark the total bankruptcy of the "high turnover, high leverage, high debt" model. The future of Chinese real estate development will return to the logic before the 1998 housing reform: dominated by state capital, existing as a quasi-public service, with profit margins strictly limited and financial attributes stripped away. The thirty-year experiment of "real estate marketization" is coming to an end—not because the experiment failed, but because it was too successful.

That "Golden Age," ignited by PSL, fueled by the frantic bets of the Country Gardens, and ultimately paid for by the relocated households of tier-3 and tier-4 cities, did not fail. The experiment successfully completed its historical mission of primitive capital accumulation and was then ruthlessly formatted by the system.

The Life Cycle of Land Finance

On January 1, 1994, the tax-sharing system was implemented, and local governments lost one-third of their fiscal revenue overnight. On July 3, 1998, the State Council announced the end of welfare housing allocation, transforming land from an administrative resource into a commodity. On August 31, 2004, all commercial land was forced onto the auction block. These three dates, like three valves, combined to complete the ignition sequence of a rocket booster. From that moment on, the fate of local governments was tied to a one-way combustion curve: it took only 23 years to go from 50.7 billion to 8.7 trillion; it took only 3 years to fall from 8.7 trillion back to 4.87 trillion.

Three-Step Ignition

The birth of land finance was not a spontaneous innovation by local governments, but the result of a decade-long implicit contract gradually being fulfilled. The starting point was the 1994 tax-sharing reform. Under the strong push of Zhu Rongji, the central government's share of fiscal revenue jumped from 22% to 55.7%, while the local share plummeted from 78% to 44.3%. However, expenditure responsibilities did not move upward along with fiscal powers; local governments still had to bear the spending responsibility for almost all public services such as education, healthcare, and infrastructure. Their share of national fiscal expenditure did not fall but rose, gradually climbing from 72% to the current 86%. The massive gap between revenue and expenditure forced local governments to find new sources of wealth, and the compensation given by the central government was an option that seemed worthless at the time: land transfer proceeds would belong to the local governments.

This option was almost impossible to exercise in 1994. At that time, housing had not yet been commodified, and the per capita living area of urban residents was only 9 square meters. The vast majority of urban workers still relied on their work units to allocate public housing. There was no demand for land, and naturally, no price to speak of. It was not until 1998, when the State Council issued Document No. 23 to end welfare housing and launch housing commodification, that real estate developers officially appeared as buyers. Even so, land prices under the early negotiated transfer model remained low, and backroom deals were common. In 2001, national land transfer fees were only 49.2 billion yuan. This figure seemed insignificant for filling the fiscal gap.

The real tipping point was the "August 31 Deadline" in 2004. The Ministry of Land and Resources mandated that all commercial land must be subject to "bidding, auction, and listing" (zhaopaigua), completely closing the back door of negotiated transfers. The market-based pricing mechanism instantly released the financial attributes of land: developers' bidding pushed up land prices, high land prices raised housing price expectations, and high housing price expectations attracted more developers to participate in auctions. National land transfer fees were 49.2 billion in 2001 and reached 549.2 billion in 2006, a 10.2-fold increase in five years. The "bidding, auction, and listing" system, known as the "Second Land Revolution," essentially provided a nuclear fission-style positive feedback loop for land finance.

These three steps were all indispensable. Without the hunger created by the tax-sharing system, local governments would have had no incentive to mine land value; without the market demand created by housing reform, land could not be monetized; without the price competition introduced by auctions, land could not be sold at high prices. Land finance was not an "accidental institutional innovation," but the result of the central government tacitly allowing local governments to fill the gap through land monetization after exchanging fiscal power for tax power. In 1994, the central government took away one-third of the local cash flow and, in return, gave the local governments a piece of empty land. The final return on this deal was staggering: 50.7 billion became 8.7 trillion.

The Booster's Combustion Curve

Mainstream narratives often describe land finance as a "deformed dependence" or "drinking poison to quench thirst," which is a hindsight perspective. Returning to 1998, China's urbanization rate was only 36%, infrastructure stock was extremely scarce, and local governments had neither the legal right to issue municipal bonds nor a taxable property tax base. Land finance was the only feasible large-scale financing solution at the time: essentially, it was a one-time discounting of the urbanization dividends of the next 70 years to pay for current infrastructure construction costs.

Judging by the results, this booster completed its mission. Between 1998 and 2021, China built the world's largest high-speed rail network (40,000 km), expressway network (160,000 km), and urban rail transit system (over 8,000 km). In 2020 alone, land transfer fees reached 8.4 trillion yuan, exceeding the total national fiscal revenue of 1998. The financing foundation for these infrastructure assets is precisely the cumulative tens of trillions in land transfer fees and their derived LGFV (Local Government Financing Vehicle) debts. Land finance is not a parasite, but a booster. The problem is not that land finance was invented, but that it had no built-in termination condition.

The operation logic of land finance is similar to a solid-fuel rocket. Once ignited, the thrust curve is entirely determined by the fuel stock and the combustion cross-section; the only thing the operator can do is wait for the fuel to burn out. The 50.7 billion in 1998 was the moment of ignition; after the 2004 auction mandate, it entered the main combustion phase, with annual increments remaining at the trillion-yuan level. It first broke 1 trillion in 2007, reached 4.29 trillion in 2014, and soared to a historical peak of 8.7 trillion in 2021. Every node corresponded to a resonance between accelerated urbanization and credit expansion. The subsequent fall is not a "deceleration," but a ballistic freefall after the fuel is exhausted: a 23.3% drop in 2022, another 13.2% drop in 2023, and another 16% drop in 2024, with a cumulative evaporation of 44% over three years.

The supply of high-quality state-owned construction land available for transfer is limited. As the urbanization rate pushed from 36% to 65%, core plots in first-tier cities were nearly exhausted. While there is volume in third- and fourth-tier cities, the land has lost price support due to population outflow. From 2017 to 2021, land transfer fees grew by about 1 trillion yuan annually, but the surface regularity masked a deep change: growth had become entirely dependent on the expansion of the area sold rather than the increase in unit price. To maintain the total volume, local governments had to push increasingly distant and remote plots onto the auction block, and the quality and premium rates of marginal land deteriorated. This is exactly the characteristic of the final stage of solid fuel combustion: thrust remains, but efficiency has declined, and there is no possibility of "igniting it again."

After the Fuel is Exhausted

In 2024, land finance dependency dropped from its 2020 peak of 43.2% to 17.3%. Reformists might see this as a successful transition to "breaking land dependence," and conservatives might interpret it as a victory for regulatory policies. However, both forms of optimism are built on a misreading of the data. The decline in dependency is not because local governments have found alternative sources of income, but because the numerator (land transfer fees) has physically collapsed.

The reason the denominator (local comprehensive financial capacity) did not shrink drastically in tandem is due to the strong hedging of central government transfer payments. In 2023, central transfers to local governments exceeded 10 trillion yuan (10.3 trillion) for the first time, and remained at a high of 10.2 trillion in 2024. The proportion of transfer payments in local comprehensive financial capacity has soared from 28.6% in 2021 to 36%. Coupled with the abnormal growth of non-tax revenue (mainly from fines, confiscations, and the disposal of state-owned assets), the balance sheet was barely maintained. This is not a "transition," but a transfer of the local deficit to the central treasury. When central transfer payments become the primary source of local finance, so-called "local finance" has actually lost its independence. The average fiscal self-sufficiency rate of the 31 provinces is only 44.7%, meaning that for every yuan a local government spends, 55 cents must be requested from the central government.

The mathematical end has arrived. In 2024, land transfer fees were 4.87 trillion yuan, leaving an annual gap of about 3.8 trillion compared to the peak. No single type of tax can fill this vacuum. Property tax is difficult to implement in the short term due to falling asset prices and political resistance; the extraction of non-tax revenue has approached the limits of social endurance, and the public resentment triggered by the "fine economy" is spreading across the country. Meanwhile, the central government itself faces deficit pressure, and the sustainability of 10 trillion yuan in transfer payments is being questioned.

The final way this gap will be paid for is through the contraction of public services. The regional divergence in fiscal self-sufficiency is shocking: Shanghai is at 84.8%, Beijing at 75.9%, while Tibet is less than 10%, and Qinghai is only 17%. For provinces with self-sufficiency rates consistently below 30%, service downgrades are likely to occur before 2027. Reductions in bus routes, stagnation in municipal road maintenance, and the normalization of wage arrears for grassroots civil servants will no longer be occasional news, but the background of daily life. The drop in land finance dependency from 43% to 17% is like the root system of a tree shrinking by 50%: it is not a successful pruning, but the entire tree withering.

The Impossible Trinity

In Chinese local finance, there is an official term called the "Three Guarantees." The original order was "guaranteeing wages, guaranteeing operations, and guaranteeing basic livelihood." It was later quietly adjusted to "guaranteeing basic livelihood, guaranteeing wages, and guaranteeing operations." Placing "basic livelihood" at the top precisely indicates that "basic livelihood" is the part most easily compressed. In 2024, when the Minister of Finance summarized the fiscal situation at a State Council Information Office press conference with the eight characters "generally stable, locally tight," the true meaning of "locally tight" had already been annotated by the remaining 333 buses on the streets of Baoding and the phrase "guaranteeing three and striving for six" circulating within Nanjing's public sector system.

Multiple Choice

Local governments face an irreconcilable trinity: providing public services, repaying existing debt, and not incurring new debt. Only two of these three can be chosen.

Each choice comes with different political costs. Choosing "guaranteeing services + repaying debt" means new debt must be borrowed to maintain operations, touching the political red line of the central government strictly prohibiting new implicit debt. Choosing "guaranteeing services + no new debt" could lead to defaults on existing debt, triggering the collapse of "LGFV faith" and the spread of financial risks. Choosing "repaying debt + no new debt" leads to a contraction of public services, the cost of which is shared by grassroots civil servants, citizens, and enterprises. Currently, China has chosen the third combination, not because it is optimal, but because the first two are politically unacceptable.

In November 2024, the Ministry of Finance launched a 12-trillion-yuan debt resolution plan (including a 6-trillion-yuan special bond quota, 4 trillion yuan over 5 years at 800 billion yuan annually, and 2 trillion yuan in shantytown renovation debt), clarifying the priority: repay debt first, then provide a safety net, and finally investment-oriented spending on public services. In the same year, the Central Economic Work Conference proposed "resolving local fiscal difficulties" for the first time, and the State Council required "firmly holding the bottom line of the Three Guarantees." Together, these three directives reveal the actual distribution of weight: the 12 trillion yuan is mainly for debt repayment, and the "Three Guarantees" are merely a safety net, not full protection. When the grassroots are explicitly told to "guarantee three and strive for six," it effectively means the first option in the trinity has been abandoned.

The official narrative describes the entire process as a successful practice of "preventing and resolving major risks." In 2024, local implicit debt dropped from 14.3 trillion to 10.5 trillion, a reduction of 3.8 trillion in one year, which appears encouraging on the surface. However, the essence of debt resolution is debt replacement rather than reduction: replacing high-interest, short-term implicit debt of Local Government Financing Vehicles (LGFVs) with lower-interest, longer-term explicit bonds. Credit card debt is converted into a mortgage; the monthly payment indeed decreases, and interest expenses are lowered, but the total local debt balance has not decreased by a single cent. Furthermore, the rigid principal and interest repayment constraints of explicit debt are stronger, making it impossible to maneuver through delays or extensions. In 2024, the LGFV debt ratio in 60% of prefecture-level cities rose rather than fell compared to 2023; the speed of replacement could never keep up with the rolling interest. By the end of 2024, local government statutory debt was 47.5 trillion yuan, plus 10.5 trillion yuan in implicit debt, and the full-caliber debt of LGFV platforms, totaling approximately 60 trillion yuan. Changing the repayment method does not make this number smaller. Debt resolution alleviates liquidity risks but increases the rigid pressure on the budget.

The Three Faces of the Cost

The choice of the impossible trinity is not an abstract macro concept but is specifically projected onto three interconnected social levels. The three types of phenomena mentioned above are not isolated "social problems" but are reactions of the same fiscal dilemma in different fields.

The first face is "long-distance fishing." When local governments must repay rigid debt and have lost land transfer fees—their core source of income—they can only extract cash flow from the outside. Public security organs in less-developed regions go to developed regions to file cases across jurisdictions and conduct "law enforcement for revenue" against private enterprises, which is essentially a hunting behavior driven by fiscal hunger. At a State Council special study meeting in 2024, Li Qiang's speech on this issue indicated that the problem has escalated from "individual corruption" to a deep-seated matter requiring the Premier's attention.

Data confirms this judgment. In 2024, national non-tax revenue reached 4.473 trillion yuan, a year-on-year increase of 25.4%; during the same period, tax revenue was 17.4972 trillion yuan, a decrease of 3.4%. More notably, non-tax revenue in the single month of December grew by 93% year-on-year. If law enforcement efficiency truly doubled suddenly at the end of the year, a commendation meeting should have been held instead of a rectification meeting. Fine and confiscation income grew by 14.8%, with an absolute amount of about 400-500 billion yuan, backed by a system that treats enterprises as an alternative tax base. The proportion of non-tax revenue in local comprehensive income has approached 20%, whereas in a normal fiscal system, this ratio is usually below 5%. When the growth rate of fine and confiscation income in a region far exceeds the growth rate of tax revenue, such behavior is no longer law enforcement but hunting.

The second face is the atrophy of public transportation. In Baoding, a city with a population of ten million, bus capacity plummeted from a peak of 1,300 vehicles to 333, a drop of 74%. Liberals tend to attribute this to "government inefficiency," while the establishment explains it as "temporary difficulties." Both arguments avoid the core issue: the atrophy of Baoding's public transport is not a management error but a mathematical result of the impossible trinity. When land transfer income disappears and debt interest must be paid on time, bus subsidies become the first budget item to be cut. In 2024, the national bus passenger volume was 38.67 billion trips, only 56% of the 69.176 billion trips in 2019. The decline in passenger flow and the reduction of routes form a vicious cycle: the longer the waiting time, the fewer the passengers; the fewer the passengers, the fewer the routes; the fewer the routes, the longer the waiting time. Even in Jiading, Shanghai, bus vehicles decreased by about 20% over two years; the atrophy is no longer confined to economically underdeveloped areas.

The third face is the internal backlash within the system. In major economic provinces like Jiangsu with a GDP exceeding 10 trillion yuan, civil servant salary cuts have reached 15-20%. Some regions have proposed "guaranteeing three and striving for six," meaning striving to pay six months' salary in a year. Teachers' salaries in Zhengzhou have decreased by 45% compared to 2018. In early 2025, news broke that civil servants would receive an average monthly raise of 500 yuan, which public opinion once interpreted as a "signal of recovery." This news instead exposed the helplessness of reality: first owe months of salary, then announce a nominal raise, while actual annual income is still declining. The 59 million personnel in government agencies and public institutions nationwide are experiencing a cognitive shift: the "iron rice bowl" is still made of iron, but there is less rice inside. The system's promise has not been withdrawn, but the proportion of fulfillment is decreasing year by year.

The Arithmetic of 2027

The impossible trinity will reach its peak pressure test in 2026-2027. The intersection of three trend lines is a geometric necessity, unrelated to emotion.

The 12 trillion yuan in debt resolution funds is concentrated for disbursement between 2024 and 2028, meaning the debt repayment pressure for these five years is artificially concentrated into a single window. At the same time, the shrinkage of land transfer fees continues, with an expected further drop of 10-15% in 2025, bringing the cumulative decline to over 50%. Although central transfer payments have exceeded 10 trillion yuan, the growth rate is slowing (10.2 trillion yuan in 2024, slightly down from 10.3 trillion yuan in 2023), and they must prioritize covering high-risk provinces such as Guangxi, Jilin, Heilongjiang, Yunnan, and Gansu, where total debt ratios exceed 500%. Rigid debt repayment expenditures are rising, land revenue is falling, and transfer payments are hitting a ceiling. The three lines will intersect at the same point around 2027. By then, the concentrated repayment period for debt resolution will not yet have ended, while land revenue will have evaporated by more than half from its peak. These two forces simultaneously squeeze the budget space for public services.

After the intersection point, provinces with a fiscal self-sufficiency rate below 30% will experience deep-seated gaps in the "Three Guarantees." Salary arrears for grassroots civil servants will change from "a few months" to "normalized," the maintenance of county-level bus systems will change from "difficult" to "unsustainable," and the maintenance cycle for municipal roads will change from five years to ten years or even be shelved indefinitely. Some cities have already adopted "intermittent lighting" measures for street lamps, using physical semi-darkness to offset the deficit on the budget sheet. As mentioned earlier, the average fiscal self-sufficiency rate of 31 provinces is less than 45%, with only Shanghai being above the median line of national fiscal expenditure. When there are only two blankets but three people shivering in the cold night, no matter how they are arranged, someone will always freeze. The linear system of equations for the three variables ultimately has only one solution; whether one is pessimistic or not is unimportant. The endgame of the impossible trinity is not a sudden crisis but a long, gradual process of public service decline.

The Real Ledger of Guaranteed Home Delivery

The bill for the first batch of "Guaranteed Home Delivery" projects in Shangqiu can be described by a simple division problem: 17 residential projects, 9,200 housing units, a total investment of 920 million RMB, averaging 100,000 RMB per unit.

What can 100,000 RMB do? According to current construction quotas, this amount is only enough to cover the basic construction and installation costs—including concrete, steel reinforcement, water and electricity pipelines, and exterior wall coating—required to bring a stalled project to a "roughcast" (shell) delivery state. It does not include fine interior decoration, landscaping, marble patterns in elevator halls, fitness equipment, children's play facilities, or property management rooms. Homebuyers paid between 800,000 and 1.5 million RMB at the market peak from 2019 to 2021, but what they ultimately receive is a physical structure costing 100,000 RMB, consisting of little more than four walls and a roof.

So, where did the price difference go? The answer reveals the real ledger behind the "Guaranteed Home Delivery" policy.

Between Two Weeks

The introduction of the "Guaranteed Home Delivery" policy had a clear temporal milestone: in mid-July 2022, a document titled "Notice of Mandatory Mortgage Payment Suspension" gained widespread attention on social networks. Property owners from over 300 unfinished projects across more than 20 provinces and cities nationwide collectively announced that if their projects could not resume work, they would stop repaying their bank mortgages. This wave of mortgage strikes spread, involving funds totaling trillions of RMB, with Henan ranking first with 61 projects. Just two weeks later, on July 28, the expression "Guaranteed Home Delivery, Stabilizing People's Livelihood" appeared for the first time in the resolutions of the Central Politburo meeting.

Such a short time gap clearly indicates the causal relationship: before this, the problem of unfinished buildings was regarded as a breach of contract in a market economy, falling within the scope of civil disputes; after this, it was redefined as a political task concerning social stability. The momentum driving this transformation stemmed from the financial risks that the mortgage strike wave might trigger, with the plight of homebuyers being secondary: once millions of middle-class citizens collectively defaulted, the mortgage assets of banks would rapidly deteriorate, and their safety buffers would be insufficient to support them.

The original intention of the "Guaranteed Home Delivery" policy was to maintain stability; protecting consumers was not the primary consideration.

This original intention determined the basic logic of the policy: in subsequent implementation documents, "Stabilizing People's Livelihood" always took precedence over "Guaranteed Home Delivery." The primary goal of the policy was to calm the situation, while fulfilling commercial promises was secondary. When homebuyers expected government intervention to allow them to reclaim the "high-quality community" promised in their contracts, the goal of policymakers was merely to eliminate the triggers for mass incidents. On April 30, 2024, the Central Politburo meeting further adjusted the wording, changing "Guaranteed Home Delivery" (of buildings) to "Guaranteed Housing Delivery" (of units), and the Ministry of Housing and Urban-Rural Development established special work teams at the national, provincial, and municipal levels. The change in name indicates an upgrade in stability maintenance requirements, though the intensity of protection did not necessarily increase accordingly.

The aforementioned misalignment of goals explains why "Guaranteed Home Delivery" often stops at "delivering the building." For the sake of stability maintenance, delivering a roughcast house without greenery, without supporting facilities, and even with temporary water and electricity connections has the same effect as delivering a finely decorated luxury home: as long as the keys are handed to the owner, the "unfinished" label is torn off, and the reason for collective mortgage strikes disappears.

Where the Five Trillion Went

Official data shows that as of the end of 2023, more than 3 million units out of 3.5 million "Guaranteed Home Delivery" projects have been delivered, reaching a delivery rate of 86%. In 2024, another 3.38 million units are planned for delivery to complete the set goals. However, the comforting nature of these figures lies in their evasion of the definition of the denominator.

Only projects identified as "savable" are included in the statistics for the 3.5 million units. Those "zombie projects" that are truly insolvent and whose asset value has returned to zero have never entered the statistical denominator. Evergrande has contract liabilities of 600 billion RMB but delivered only 60,000 units in the full year of 2024 (a significant drop from 234,000 units in 2023); Country Garden has contract liabilities of 400 billion RMB, with approximately 200,000 units still to be delivered. None of this data is included in the 86% denominator. In other words, the 86% success rate is calculated after excluding "failing" projects.

The mismatch on the funding side is even more significant. Wu Ge, Chief Economist of Changjiang Securities, estimates that the total funding gap for "Guaranteed Home Delivery" is approximately 4 trillion RMB. Cumulative special loans launched at the central level amount to only 350 billion RMB (200 billion in 2022 + 150 billion in 2023). Even including subsequent supporting financing, the actual injected funds are about 550 billion RMB, covering only 14% of the 4 trillion RMB gap.

The logic of this 14% fund allocation is similar to a battlefield medical triage system. At a mass casualty scene, military doctors divide the wounded into three categories: Red (immediate surgery for survival), Yellow (can wait), and Black (unsavable, no resources allocated). The allocation rules for "Guaranteed Home Delivery" funds are exactly the same: limited special loans are only directed toward "Red" projects, while for those "Black" projects where assets cannot cover debts, the system chooses to abandon them. The difference is that in battlefield triage, those labeled black are the minority, while in "Guaranteed Home Delivery," those abandoned are the majority.

So, what exactly are the "White List" mechanism and the "5 trillion RMB financing" mentioned in media reports?

The 5 trillion RMB "White List" financing disclosed by the National Financial Regulatory Administration is mostly used for the extension and adjustment of existing loans, and very little is used as new loans for civil engineering works. Banks face a realistic dilemma: they must respond to policies supporting real estate enterprises, ensure the safety of depositors' funds, and meet their own risk control requirements. The optimal solution is to "trade time for space," extending development loans that were originally about to expire and listing them as "normal category loans" on the books to avoid the "non-performing loan" classification. The main role of this 5 trillion RMB is to maintain the quality of bank assets and prevent development loan defaults from turning into bad debts; it rarely translates directly into cement and steel on construction sites. Differences in local implementation rates confirm this: Henan's White List credit was 110.1 billion RMB, with actual disbursement of 66.6 billion RMB (a 60% implementation rate), while Guangdong's implementation rate was 62%. Approval does not equal disbursement, and disbursement does not equal use for construction.

The so-called 5 trillion RMB in "Guaranteed Home Delivery" financing is, at its core, self-protection for the financial system.

The 100,000 RMB House

After funding priority is given to banks, can the remaining money still be used to build houses? Policymakers proposed the "Last-In, First-Out" principle: to attract new capital injections into unfinished projects, it is stipulated that newly injected supporting financing enjoys the highest repayment priority, ahead of original creditors. Meanwhile, the new housing policy at the end of 2022 allowed a maximum of 30% of pre-sale regulatory fund replacements to be used to repay the project's maturing debts.

Combining these two rules clarifies the priority order of fund flows for unfinished buildings: banks with new loans come first, suppliers second, and homebuyers last. Homebuyers believe "Guaranteed Home Delivery" aims to ensure the delivery of real estate, but in reality, it preserves bank loans.

This allocation logic leads to a decline in delivery quality: limited funds prioritize the exit of "last-in" capital, followed by the wages of construction workers (the stability maintenance baseline), leaving very little for actual construction. The cost of "100,000 RMB per unit" in the Shangqiu project is by no means an isolated case; it is an inevitable result of this allocation logic. Calculated at 100,000 RMB per unit, for a residence with a construction area of 100 square meters, the investment per square meter is only 1,000 RMB, barely enough to erect the structural skeleton and basic enclosure.

While Evergrande carries 600 billion RMB in contract liabilities, it delivered only 60,000 residential units in 2024; Country Garden still has about 200,000 units to be delivered. As time passes, the asset quality of projects awaiting delivery declines daily, and "downgraded delivery" has become the norm. Future "Guaranteed Home Delivery" projects will deliver a "habitable concrete structure." The standards of commercial housing have long since vanished—no landscaping, no entrance lobby, no video intercom, and perhaps not even formal electricity. Homebuyers will receive a key and open a door to a roughcast concrete box. Meanwhile, the proportion of existing home sales has risen from 12.7% in 2020 to 30.84% in 2024, as the pre-sale system is gradually being abandoned.

This is the ending: homebuyers paid 1 million RMB and ultimately received a building worth 100,000 RMB. The 900,000 RMB that evaporated in the middle did not disappear into thin air. During the frenzy of 2019 to 2021, this money turned into land transfer fees in local government fiscal accounts, interest on developers' USD bonds, and massive dividends under the high-turnover model. That piece of land was valued at 500 million RMB in 2020 but is now worth almost nothing; "Guaranteed Home Delivery" cannot fill this hole. "Guaranteed Home Delivery" is merely using a thin sheet of iron to cover the hole, preventing more people from falling in.

The Path to Zombification

In 1993, an official from Japan's Ministry of Finance stated during a parliamentary session: "Land price adjustments are orderly, the banking system is sound, and there is no risk." At that time, the non-performing loan (NPL) ratio reported by the Japanese banking industry was only 2%. However, by 2002, the true NPL ratio was exposed at 8.6%, with 105 trillion yen in bad debts piled high, forcing Heizo Takenaka to launch the largest financial reform since World War II. During those nine years, zombification became the primary characteristic of the economy. In 2025, the overall NPL ratio of China's commercial banks is 1.50%, the NPL ratio for the real estate industry is 4.94%, and the proportion of "special mention" loans is 1.47%. The 5 trillion yuan in "white list" loans are "mostly pushed forward in the form of extensions," and 60 out of 77 defaulted real estate enterprises have chosen "restructuring" over liquidation. These phenomena are remarkably similar to Japan in 1995.

The Illusion of a Gentle Decline

A "soft landing" is the official narrative of current Chinese real estate policy: slow price adjustments, no acute crisis, and the market returning to equilibrium on its own. Japan told a similar story between 1991 and 1997. At that time, land prices in Japan's six major cities fell by an average of 5-8% annually. The government believed the decline was "controllable," and banks avoided the appearance of bad debts through loan extensions and reclassifications; everything seemed under control. However, in November 1997, the collapses of Hokkaido Takushoku Bank and Yamaichi Securities proved that six years of "orderly adjustment" had merely pushed the problem from the surface into the depths. The proportion of zombie firms soared from 5% in 1991 to 15% in 1997, further swelling to 25.5% by 2001. The banking NPL ratio exploded from a superficial 2% to 8.6% (105 trillion yen), severely suppressing the productivity and job-creation capacity of the entire economy.

The manifestations of a soft landing and zombification are almost identical in the first five years.

The difference emerges after the sixth year: a soft landing can restore growth, while zombification continues to drain economic vitality. China has now entered the fourth year of this process (2021-2025) and has yet to show any signs of restored growth. From the peak in 2021 to the end of 2024, housing prices have cumulatively fallen by approximately 33%, with declines exceeding 40% in some cities. The price-to-income ratios in Beijing, Shanghai, Guangzhou, and Shenzhen exceeded 30 or even 55 at their peak, whereas in Tokyo when the bubble burst, it was only 18. In other words, China requires a longer adjustment cycle to return to the mean. In the first ten months of 2025, real estate development investment fell by 14.7% year-on-year, marking five consecutive years of negative growth. A balance sheet recession is not a "cyclical" phenomenon but a long-term process of credit contraction and demand shrinkage that lasts until excess debt is cleared or written off. The argument for a so-called cyclical correction is difficult to sustain. When policymakers attempt to limit price drops through administrative means to protect bank balance sheets, they are essentially forcing the vitality of the entire economy to pay for ineffective assets that cannot be cleared. Japan referred to this state as the "Lost Decade," later revised to the "Lost Two Decades," and eventually stopped revising it altogether because people had forgotten what "normal" looked like.

The Chinese Version of Evergreening

Japanese economists referred to the practice of banks continuously lending to insolvent companies in the 1990s as "evergreening": not because the companies had vitality, but because the loans never matured. Research by Caballero, Hoshi, and Kashyap in the 2008 American Economic Review pointed out that when banks keep zombie firms on life support, not only do the zombie firms themselves struggle to recover, but more importantly, healthy firms are crowded out of the market because zombie firms occupy land, labor, and credit lines without being able to generate profit.

The 5 trillion yuan in "white list" loans in China, which are "mostly pushed forward in the form of extensions, with incremental financing being the minority," is a precise replica of evergreening.

Among 77 defaulted real estate enterprises, 60 chose "restructuring" rather than liquidation, and 20 approved restructuring plans resolved 1.2 trillion yuan in debt. These enterprises are still "alive" in a statistical sense, but they are dead in an economic sense, and they are squeezing the survival space of peers who still have the ability and willingness to repay debts. In 2024, Vanke could still obtain bank loans by pledging project companies, but since the beginning of 2025, it has not disclosed any similar announcements. When even a "star student" with a state-owned background finds it difficult to obtain loans, the financing channels for small and medium-sized developers have long been completely closed. Evergrande was ordered to liquidate by a Hong Kong court in January 2024 and was delisted from the Hong Kong Stock Exchange in August 2025. In October 2025, Country Garden announced that its controlling shareholder agreed to convert a $1.14 billion loan into equity. The essence of a "debt-to-equity swap" is that the bank gives up the possibility of debt recovery and turns from a creditor into a passive shareholder, further locking up resources that could have been used for new credit.

The official NPL ratio for the banking industry is 1.50%; the credibility of this figure depends on the strictness of the definition of "non-performing." The NPL ratio for the real estate industry has reached 4.94% (Bank of China's metrics), and the real estate NPL ratio for Qingdao Rural Commercial Bank has surged from 7.17% to 21.32%. Risk is transmitting from the corporate sector to the household sector: the NPL ratio for personal loans at the six major banks rose across the board by 0.16 to 0.45 percentage points in 2024. The NPL ratio for personal loans at ICBC rose from 0.70% to 1.15%, an increase of 64%. The proportion of "special mention" loans (a precursor to NPLs) is 1.47%, which, combined with NPLs, totals about 3%. The transition period for the new "Measures for the Risk Classification of Financial Assets of Commercial Banks" in 2024 is coming to an end, and some banks will be forced to reclassify "special mention" loans as "non-performing." The official 1.5% NPL ratio looks more like accounting treatment under the grace of a transition period; the true risk level may be 2-3 times the official figure.

The four major AMCs (Asset Management Companies), which were supposed to act as bad debt processors, are themselves in trouble. Huarong was renamed CITIC Financial Asset, while Cinda, Orient, and Great Wall were transferred to Central Huijin; the fire brigade itself is on fire. China Cinda has deployed 63.9 billion yuan since 2022 to resolve 125 real estate risk projects, but 63.9 billion yuan is less than 0.2% coverage relative to the tens of trillions in debt stock. Every year, the entire industry disposes of about 3 trillion yuan in non-performing assets, but the speed of new NPL formation is almost equal to the speed of disposal, and the inventory has not decreased.

The Takenaka Moment of 2028

The core of zombification is a problem of time allocation: do you choose to endure three years of severe pain today (forced write-offs, bank recapitalization, corporate bankruptcies), or choose fifteen years of long-term bleeding in the future (declining productivity, shrinking investment, degrading public services)? Japan's data provides the answer: the 12 years of "patience" from 1990 to 2002 resulted in a total disposal cost of 47.1 trillion yen, with only 50% recovered. Scholars estimate that if forced clearing had begun in 1992, the total cost might have been only half as much.

China's choices—extensions, white lists, and ensuring the delivery of pre-sold homes—are identical to Japan's choices in 1993: acknowledging the existence of the problem but being unwilling to pay the short-term political price.

Supporters often say "China has a stronger state capacity to manage risks," which is exactly what Japan's Ministry of Finance said in 1993. Strong administrative capacity can indeed delay the outbreak of a crisis, but it cannot reduce the total cost; it can only spread the total cost over a longer timeline. The results of this distribution are already apparent: household sector debt is about 80 trillion yuan, accounting for 60% of GDP, with mortgages accounting for over 55%, and the debt-to-income ratio is as high as 144.8%. These figures indicate that even if housing prices stabilize, the debt repayment burden of residents will continue to suppress consumption for the next decade. The repair of balance sheets requires not confidence, but time and write-downs.

If China currently corresponds to Japan in 1995 (the 4th year after the bubble burst, mid-stage of the denial phase), then the "Takenaka Moment" will arrive as early as 2028 (corresponding to Japan in 2002, the 12th year after the bubble). The so-called "Takenaka Moment" refers to the point when political leaders resolve to force the write-off of bad debts. This moment in Japan was triggered jointly by Junichiro Koizumi and Heizo Takenaka in October 2002, at the cost of several major banks accepting state capital injections and thousands of companies being forced out of the market, with the unemployment rate climbing to 5.5% at one point. The short-term cost was brutal, but only after undergoing this surgery did the Japanese economy return to positive growth after 2005. By then, the real estate NPL ratio of Chinese banks may rise from the current 4.94% to 8-10% (corresponding to Japan's 8.6% in 2002), triggering mandatory capital injections and bad debt write-offs. In the three years leading up to that (2025-2028), China will experience a zombie period of "living but not growing": housing prices will fall by an average of 4-5% annually without crashing, bank bad debts will continue to rise without exploding, developers will default without liquidating, and the economic growth rate will gradually slide from 5% toward 3%. This is not the process of a soft landing, but the process of a failed soft landing. A verifiable sign will be: if the real estate NPL ratio of banks breaks 8% before 2027, and the first national bank requires a state capital injection due to real estate bad debts, the "Takenaka Moment" will be forced forward. The total cost of delay is always higher than immediate clearing.

Debt Restructuring Map

On January 6, 2023, Guizhou Zunyi Road and Bridge Construction Group issued an announcement stating that it had reached an agreement with banks: 15.6 billion yuan in bank loans would be extended for 20 years, with the interest rate reduced from 7.5% to 3.5%. The tone of the announcement was calm, appearing like an ordinary commercial contract modification; no one mentioned "default." However, with the interest rate halved and the term tripled, in terms of net present value, the banks have effectively written off half of the principal. This is the first public de facto write-down in the history of Chinese local debt, despite being packaged as a "negotiated extension." Similar practices are spreading: a listed company in Yunnan confirmed a 26% impairment of accounts receivable in 2024, and Guangxi, Gansu, and Qinghai are also lining up. The 12 trillion yuan debt resolution plan provides a buffer for local governments, but the true clearing has yet to arrive; four-fifths of the 67.82 trillion yuan in LGFV debt cannot cover interest through operations. Zero default is merely a political commitment, until the cost of maintaining that commitment exceeds the price of abandoning it.

The Political Economy of Zero Default

LGFV (Local Government Financing Vehicle) standardized bonds have so far maintained a record of "technical zero default," based on which the market has formed an "LGFV faith," believing that the government will not allow LGFVs to default. This view confuses cause and effect; "zero default" does not represent credit quality but rather that the political cost of default is too high. Once a standardized bond in a certain province defaults, the financing channels for all LGFVs in that province would immediately close, and a liquidity crisis would spread rapidly. To avoid a systemic shock, local governments have built a sophisticated yet fragile defense system: they would rather use new debt to pay off old debt, rather let non-standard financing default first, and rather let bank loans be extended for 20 years with interest rates halved, all to maintain the rigid repayment of standardized bonds.

The cost is extremely high. Rhodium Group data shows that among the 67.82 trillion yuan of LGFV interest-bearing debt, the operating cash flow generated by four-fifths of the debt cannot cover interest expenses. The scale of 67.82 trillion yuan has exceeded 52% of China's GDP, equivalent to every Chinese person carrying approximately 50,000 yuan in LGFV debt. Many highways built with this debt have average daily traffic volumes of less than 30% of their designed capacity, industrial park vacancy rates exceed 60%, and the occupancy rates of new urban areas are less than 20% of the planned population. Projects begin losing money the day they are completed, and interest is paid entirely by borrowing new debt. Most platforms are already financially bankrupt and can only maintain a "living dead" state by borrowing new to pay old. Between 2018 and 2022, Guizhou experienced 108 non-standard default events. Trust and leasing companies bore the first wave of losses, acting as "human shields" to protect standardized bonds from breaking. Non-standard defaults do not trigger cross-default clauses, do not cause public market shocks, and do not make news headlines. This is a carefully arranged "controlled bloodletting": letting peripheral creditors quietly bear the losses to preserve the dignity of core financing channels.

Ultimately, this structure is not a credit system but a Ponzi structure guaranteed by provincial-level credit. The source of debt repayment is no longer project returns but the principal of the next creditor; as long as the chain of borrowing new to pay old remains unbroken, the game can continue. However, with land transfer fee revenues falling sharply (with drops exceeding 50% in some provinces), local finances as the ultimate guarantor are also becoming unsustainable. Shanghai's fiscal self-sufficiency rate is 84.8%, while Qinghai's is only 17% and Tibet's is less than 10%. The debt ratios of 12 key provinces (Tianjin, Inner Mongolia, Liaoning, Jilin, Heilongjiang, Guangxi, Chongqing, Guizhou, Yunnan, Gansu, Qinghai, Ningxia) generally exceed 400%, with Guizhou exceeding 600%. When fiscal support capacity is exhausted, "zero default" is no longer a matter of will, but a matter of mathematics.

The Zunyi Model and the Write-down Map

The Zunyi Road and Bridge case provides a logic for resolution, involving 15.6 billion yuan in bank loans extended for 20 years, with only interest paid and no principal for the first 10 years, and the interest rate reduced from 7.5% to 3.5%. According to international financial market standards (such as ISDA), when a debt restructuring results in a net present value (NPV) loss of more than 20%, it constitutes an "event of default." Calculating with a 3.5% interest rate and a 20-year term, the NPV impairment of this debt is approximately 50%. Zunyi Road and Bridge was not an "extension" but a de facto default with a 50% NPV impairment; it is just that both parties agreed not to use the word "default" after reaching a consensus.

This case established the rhetorical path for Chinese-style debt restructuring: do not call it default, call it a "negotiated extension"; do not say write-down, call it "trading time for space"; do not mention losses, call it "interest rate concessions." When a bank accepts a 3.5% interest rate to renew a 7.5% loan for 20 years, the bank has effectively written off half of the principal, only pretending on its accounting statements that there is no loss.

The 12 trillion yuan debt resolution plan introduced in November 2024 (6 trillion yuan limit swap + 4 trillion yuan new special bonds + 2 trillion yuan shantytown renovation debt forgiveness) is often misinterpreted as a savior. These 12 trillion yuan correspond to 67.82 trillion yuan in LGFV interest-bearing debt, a coverage rate of only 18%. More importantly, swapping is not remission; replacing 7% LGFV non-standard financing with 3% local government bonds does indeed save about 240 billion yuan in annual interest costs, but the principal remains exactly the same. It is like a family with an annual income of 500,000 and a debt of 5 million; the bank agrees to lower the credit card interest rate from 18% to 8%. The monthly payment is indeed reduced, but the debt is still 5 million—it just means dying a bit slower. The 12 trillion yuan is a painkiller, not a scalpel. The true scalpel is writing down the principal, and that requires someone to bear real pain.

The geographic diffusion path of write-downs is already clear. The first circle is Guizhou (already occurred with Zunyi Road and Bridge in 2023); the second circle is Yunnan, where an A-share listed company already confirmed a 26% impairment provision for LGFV accounts receivable in 2024; the third circle is Guangxi, Gansu, and Qinghai, where fiscal self-sufficiency rates are generally below 25% and debt ratios exceed 500%; the fourth circle is the three Northeastern provinces and Inner Mongolia, where economic contraction and population outflow have caused the debt base to solidify while repayment sources continue to decrease. The logic of diffusion is not geographic proximity but fiscal self-sufficiency: when a province's own revenue cannot cover debt interest, central government transfer payments become de facto debt subsidies. The 12 key provinces designated by the State Council's Document No. 35 are the 12 buffer zones waiting in line for write-downs.

The Statistical Disappearing Act of 2027

State Council Document No. 35 requires that all LGFV platforms must complete market-oriented transformation or exit by the end of 2027. As of June 2025, more than 60% of LGFV platforms have announced their "exit." This figure seems optimistic, provided one does not look too closely at the definition of "exit."

The so-called exit means that the platform is no longer included in the statistical scope of "implicit government debt" and no longer performs government financing functions. The debt has not disappeared; a 10 billion yuan loan will not decrease by a single cent just because the borrowing entity changed its name from "Chengtou Company" to "State-owned Capital Operation Company." The actual effect of "exit" is that the local government has shed its implicit guarantee responsibility for these debts; the debt has changed from "what the government should pay" to "the enterprise's own business." When all LGFVs are required to exit in 2027, 67 trillion yuan in debt will change from "implicit government debt" to "SOE operating debt." The day implicit debt hits zero on the statistical reports is not the day the debt disappears, but the day the debt is redefined.

This brings about a massive pricing paradox. Currently, the reason banks and bond markets give these platforms low interest rates of around 3% is because they default to the belief that these platforms have government backstops. When the 2027 deadline arrives and all platforms are completely severed from the government in legal and statistical terms, these platforms will face the market as independent SOEs. Why should an enterprise with no government credit endorsement and an asset return rate of less than 1% obtain 3% financing? The market will reprice, and 3.5% interest rates will return to 7% or even higher. At that time, extension plans designed based on low interest rates will collapse on multiple fronts. A verifiable signal: if the interest rate for public market bond issuance by any "exited" LGFV platform in a key province breaks 6% before the end of 2026, that will be the first brick in the collapse of "LGFV faith." After losing implicit government endorsement, the true risk of 67 trillion yuan in debt will be accurately priced by the market for the first time. At that point, local governments will have only two choices: either resume implicit guarantees (equivalent to admitting "exit" was just window dressing) or allow defaults to occur (equivalent to admitting "LGFV faith" was an illusion). Whichever path is chosen, it will demonstrate that the current narrative of "exiting without defaulting" is unsustainable, and no one will like that price.

Shrinking Public Services

At six o'clock on the morning of February 23, 2023, citizens of Shangqiu, Henan, found a new line of red text on bus stop signs: Due to severe losses, all urban bus routes will be suspended starting March 1; over 2,000 drivers have not received wages for six months. After this notice sparked heated discussion on social media, it was deleted within less than an hour, and a municipal government task force rapidly moved in to take over. A subsequent official statement claimed they would "overcome difficulties to ensure operations do not stop." This was not just a piece of crisis PR, but more like an omen: the local government's task force arrived faster than the buses, but the funds were nowhere to be seen. In July of the same year, Baoding Public Transport reduced its fleet from 1,300 to 333 vehicles, with the first batch of 591 pure electric buses all scrapped due to battery aging; in the same month, Mohe Public Transport announced a "suspension of operations" only to retract it the same day. The commonality among these cities is not "poor management," but rather the cutting off of fiscal supply. In 2024, 2,175 bus routes were cut nationwide, and the number of buses decreased by 32,000—this is by no means an isolated case. As land transfer fees fell by a cumulative 44% over three years, bus subsidies became the first budget items to be slashed.

Public Transport Bears the Brunt

Statistical data from the China Urban Public Transport Association in 2024 shows that nationwide bus routes decreased by 2,175, and the bus fleet size decreased by over 32,000 vehicles, accounting for approximately 6% of the total. The contraction is not limited to peripheral cities like Shangqiu or Mohe; first-tier cities are also quietly scaling back, with Guangzhou suspending 32 routes within six months and Beijing suspending 24. Since 2021, waves of bus operation suspensions have broken out in Hebei, Shandong, Henan, Jiangsu, Guangdong, Liaoning, Hainan, and other regions, with over 20 public transport companies owing back wages. The public generally attributes the cause to "declining passenger flow," "battery aging," or "poor corporate management." These reasons seem to follow commercial logic but mask the underlying fiscal problem: public transport has never been a business; it is a public welfare service indirectly supported by land transfer fees.

According to Ministry of Transport data, fare revenue from Chinese urban buses covers only 10%-30% of operating costs, with the remaining 70%-90% gap long dependent on local fiscal subsidies. During the golden age of land finance, this money came from surpluses in land transfer fees; however, as land transfer revenue fell from 8.7051 trillion yuan in 2021 to 4.8699 trillion yuan in 2024, with 22 provinces seeing negative growth in land revenue (Jiangsu -23.3%, Zhejiang -27%, Guangdong -28.9%, Hunan -36.1%), the fiscal logic has undergone a profound change. For local officials, within a limited pool of funds, repaying interest on Local Government Financing Vehicle (LGFV) debt is a legal red line, civil servant salaries are the bottom line for maintaining stability, and bus subsidies have become the easiest part to cut.

The wage arrears at Shangqiu Public Transport are not due to a lack of passengers, but because the Finance Bureau stopped allocating funds; Baoding Public Transport compressed 72 routes to 49 not to optimize efficiency, but because it could not afford new vehicles to replace electric buses with dead batteries. The contraction of public services follows a clear priority: projects requiring long-term cash flow subsidies (such as buses and off-peak subway services) are cut first, followed by projects with high maintenance costs (such as street lighting brightness, road repairs, and landscaping), while projects involving personnel quotas (bianzhi) are last. Nationwide, street lighting electricity costs exceed 40 billion yuan annually; against the backdrop of fiscal tightening, various regions have adopted saving measures such as "staggered lighting" or "intermittent lighting." Bus suspensions are not a result of market failure, but the first clear signal of fiscal austerity.

Tightening Belts May Become the Norm

"Tightening one's belt" (Guo jin ri zi) has become a high-frequency term in the 2024 budget reports of governments at all levels. Hunan Province announced a 40% reduction in office maintenance funds; Longhua District in Shenzhen, one of the wealthiest administrative districts in the country, explicitly cut daily office expenses by 50% and training/research expenses by 70% in its budget documents, stating that "in principle, no more funds will be allocated for greenery," making even potted plants in civil servants' offices a luxury. These figures seem to reflect rational fiscal discipline, as if everything will return to normal as long as this period is weathered.

However, the data reveals another reality: the superficial savings are merely an expedient measure; the long-term collapse of the revenue structure is the fundamental problem. The expected growth in fiscal revenue for the full year of 2025 is only 0.1%. In the first 10 months, local general public budget revenue was 10.5 trillion yuan, while expenditures reached 19.1 trillion yuan, resulting in an operating gap of 8.6 trillion yuan alone—a hole that cannot be filled even if all annual land transfer revenue is included. At the same time, the downward trend in land transfer revenue is almost irreversible, as real estate inventory far exceeds demand for the next decade. Atop a repeatedly shrinking fiscal cake sits a massive population supported by public finance. China currently has approximately 80 million people supported by public finance (including 7 million civil servants, 13 million personnel managed in reference to the Civil Service Law, 31 million in public institutions [shiye bianzhi], and 40 million off-budget temporary workers), plus another 30 million retirees. As zero tax growth becomes the norm, the prioritization of the "Three Guarantees" (guaranteeing salaries, operations, and basic livelihood) becomes a cruel arithmetic problem: every extra cent spent on personnel expenses means one cent less for public service expenditures.

To bridge the gap, a more covert method of extraction is emerging. In 2024, against the backdrop of a 3.4% decline in national tax revenue, non-tax revenue grew by 25.4% year-on-year, reaching approximately 4.47 trillion yuan. Non-tax revenue mainly includes administrative fees, fines and confiscations, and revenue from the paid use of state-owned resources. When the government's revenue model shifts from "tax-base sharing" to "fee extraction," residents face double pressure: reduced public services and increased hidden burdens. This situation is similar to a decision made by a property management company due to financial difficulties: stopping garden maintenance, reducing security patrols, lowering elevator maintenance frequency, while simultaneously raising property fees by 25%. Owners receive fewer services but pay more. This is the reality for residents in third- and fourth-tier cities in 2024-2025: bus wait times are longer, streetlights are dimmer, green belts wither with no one to replant them, potholes go unrepaired, while parking tickets are issued more frequently and various administrative fees are quietly increasing.

Debt Interest Prioritized Over Teachers' Salaries

In this round of fiscal tightening, civil servants in Beijing and Shanghai have generally seen salary cuts of about 20%, and police salaries in Tianjin, Nanjing, and Dalian have also decreased by 20%. In a prefecture-level city in the southwest, the monthly take-home pay of a section-level (keji) civil servant has fallen below 4,000 yuan, and year-end bonuses have dropped from 50,000-100,000 yuan in the past to zero. Performance-based pay for some middle school teachers in Guangzhou has been suspended, with annual income decreasing by 45% compared to the 2018 peak; some regions have even demanded the return of "excessively distributed" bonuses from earlier years. Twenty provinces and cities, including Shandong, Jiangxi, Beijing, and Chongqing, have launched institutional reforms, with administrative quotas (bianzhi) uniformly reduced by 3%, indicating that some will lose even the qualification to have their salaries cut.

These personnel supported by public finance are victims, but they are not at the bottom of the chain. In this zero-sum game, the returns for one role are rigid: the creditors. In 2024, local government debt interest payments totaled approximately 1.2 trillion yuan, and this money has the highest payment priority. Civil servant salaries can be discounted, bus subsidies can be canceled, and teacher performance pay can be suspended, but bank interest cannot be short by a single cent or delayed by a single day. Debt default means a collapse of credit ratings and a rupture of the refinancing chain, a risk that local governments cannot afford. Limited fiscal funds are prioritized for repaying debt interest, with the remaining funds then allocated to public services.

This crowding-out effect is particularly evident in third- and fourth-tier cities. Central government urban renewal funds are explicitly tilted toward megacities and super-large cities, while third- and fourth-tier cities are not on the support list. In 2025, land transfer revenue in third- and fourth-tier cities is expected to see the largest declines (some exceeding 30%), with a very low possibility of recovery (net population outflow + excess inventory). If by 2027, the national bus fleet decreases by more than 20% (about 100,000 vehicles) compared to the 2023 peak, and the first case occurs of a prefecture-level city closing compulsory education schools on a large scale due to fiscal difficulties, it will mark the transition of public service shrinkage from the "saving" phase to the "disintegration" phase. Third- and fourth-tier cities are gradually becoming a vast public service desert, yet they are home to over 60% of China's population.

The Intersection of Population and Inventory

The 2020 Seventh National Census showed that the permanent resident population of Hegang was 890,000, a decrease of 210,000 compared to 2000. During the same period, the number of primary schools in the city shrank from 170 to 36, and the number of enrolled students dropped from 69,400 to 30,200. In 2019, it was discovered that a 55-square-meter apartment in Hegang could be purchased for just 19,000 RMB. This news sparked heated discussions online but was quickly forgotten. Few realized that Hegang was by no means a joke, but rather a warning sample. By 2024, the populations of 173 out of 303 cities nationwide were decreasing, accounting for more than 57%. Goldman Sachs predicts that the demand for new urban housing in China will remain below 75% of its 2017 peak for the long term, and the household formation rate will drop from 8 million households per year in the 2010s to 1.5 million by 2050. The inventory absorption period for housing in third- and fourth-tier cities has reached 30.3 months, while ultra-broad inventory requires 44.5 months to digest. When the population decline curve intersects with the inventory accumulation curve, the assumption of "gradual digestion" will be shattered, and a vicious cycle lacking a self-correction mechanism has already begun.

The Death Cross

In technical stock analysis, a "death cross" refers to a short-term moving average falling below a long-term moving average, signaling a trend that is difficult to reverse. China's third- and fourth-tier cities are experiencing a real estate version of the "death cross": the population decline curve and the housing inventory curve collided between 2022 and 2024. On one hand, the populations of 173 out of 303 cities nationwide are decreasing, and the household formation rate has dropped from 8 million per year to 1.5 million. On the other hand, the absorption period in third- and fourth-tier cities has climbed to 30.3 months (narrow definition), while the ultra-broad inventory absorption period has reached 44.5 months. In 2024, the national population decreased by 1.39 million, with 9.54 million births and 10.93 million deaths, resulting in a natural growth rate of -0.99‰. Even more concerning are some leading indicators: in the first three quarters of 2024, the number of marriage registrations nationwide was 4.747 million pairs, a year-on-year decrease of 943,000 pairs, which foreshadows a further decline in the birth population for 2025-2026.

From provincial data, Shandong decreased by 428,000 people in one year (a cumulative decrease of 900,000 over three years), Heilongjiang decreased by 330,000, Henan decreased by 300,000 (falling for four consecutive years), Hunan decreased by 290,000, and Liaoning decreased by 270,000. In Henan, 15 out of 17 prefecture-level cities saw population decreases, a ratio as high as 83.3%; in Shandong, 11 out of 16 cities experienced population declines. Goldman Sachs estimates that population shrinkage will lead to a reduction in housing demand by 500,000 units per year in the 2020s, a figure that will increase to 1.4 million units per year in the 2030s. Meanwhile, in 2024, the population of Shenzhen increased by nearly 200,000, Guangzhou by 150,000, and Guiyang by 200,000. The distribution of China's population changes is extremely uneven, showing a trend of polarization: a few "winner" cities attract population inflows, while the majority of "loser" cities lose both residents and housing prices.

In 2024, national sales of new commercial housing totaled 9.675 trillion RMB, a year-on-year decrease of 17.1%, indicating that the speed of developers' capital recovery is even faster than the rate of population outflow. The absorption period is a ratio with inventory as the numerator and sales area as the denominator. Current market concerns regarding the "30.3 months" are mainly concentrated on the excessive inventory, while ignoring the sharp shrinkage in sales area. In 2024, sales in third- and fourth-tier cities dropped to 53% of their 2021 peak. The national sales area of new commercial housing decreased by 12.9% year-on-year, new construction starts decreased by 23%, and development investment decreased by 10.6%. The calculation of the absorption period is based on an optimistic assumption: that the current sales speed can remain constant. However, as net population outflow becomes the norm, the denominator of sales area will shrink in the long term. If the denominator shrinks by another 10% each year, the absorption period will exceed 60 months in three years. In other words, even if not a single new house is built within five years, the inventory will still be difficult to digest.

When the denominator of the absorption period approaches zero due to population loss, the absorption period will mathematically tend toward infinity. In cities with population inflows, inventory is a temporary backlog that can be digested by trading time for space; in cities with population outflows, inventory becomes a permanent sediment, and time will only cause those properties to depreciate further. The stage of supply-demand mismatch has passed; today, the two have become completely disconnected.

Houses No One Will Ever Buy

In 2023, the urban housing-to-household ratio was 1.07, 1.12 in third- and fourth-tier cities, and as high as 1.35 in the Northeast region. Behind these figures are 65 million to 90 million vacant housing units, of which 31 million units are completed but unsold, and 50 million to 60 million units are purchased but unoccupied. The national housing vacancy rate is approximately 16.7%. A study published in Nature Cities in 2024 showed that the housing utilization efficiency in highly urbanized areas of China fell from 84% in 2010 to 78% in 2020, with the decline being faster in old urban areas. Professor Long Ying of Tsinghua University once pointed out: "The government does not know exactly what China's housing vacancy rate is." This ambiguity in data serves as a buffer during market upturns but becomes an amplifier of panic during downturns.

An "absorption period of 30.3 months" is just the tip of the iceberg. If properties under construction but unsold are included (broad inventory), the total reaches 1.96 billion square meters, with an absorption period of 28.9 months. If properties approved but not yet started are added (ultra-broad inventory), the total is 3.27 billion square meters, and the absorption period extends to 44.5 months. The degree of market divergence is already quite evident: in January 2025, the absorption period in first-tier cities fell back to 12.6 months, close to a healthy range; meanwhile, the absorption period in third- and fourth-tier cities continued to climb at a high of 30.3 months. Compared to April 2019, when the absorption period in third- and fourth-tier cities was only 10.5 months, it has increased nearly threefold in five years. The absorption periods in cities such as Nanping, Ganzhou, Xiangyang, Taizhou, and Yantai have exceeded 50 months. The red line set by the Ministry of Natural Resources is 36 months, beyond which land supply must be suspended, yet these cities have long crossed that line while new housing continues to be delivered. Goldman Sachs estimates that completing projects that have started but are unfinished requires 1.1 trillion USD (approximately 8 trillion RMB), which is more than 25 times the announced inventory buyback schemes.

The IMF predicts that China's basic housing demand will drop by 35%-55%. Goldman Sachs predicts that by 2030, annual demand will drop to 11.5 million units, 8 million units by 2040, and only 6 million units by 2050. When the demand side shrinks by 10%-15% annually, any "absorption period" calculated based on current speeds is merely a self-deceiving number. For many third- and fourth-tier cities, those houses may never be purchased by anyone. By 2050, when the household formation rate drops to 1.5 million per year, the number of new households in China will not even be enough to fill the existing empty houses. In 2024, national housing development investment was still as high as 10.028 trillion RMB (a year-on-year decrease of only 10.6%), and although new construction starts decreased by 23%, they still amounted to 738.93 million square meters. The demand side is shrinking at double-digit speeds, while the inertia of the supply side has not yet stopped.

Hegang is Not a Joke

Hegang is by no means a joke, but rather a warning: its permanent resident population was 1.1 million in 2000 and dropped to 890,000 in 2020, a decrease of 210,000 over 20 years. The number of primary schools plummeted from 170 to 36, the number of enrolled students fell from 69,400 to 30,200, and housing prices are only 2,157 RMB/square meter, with a record low of 345 RMB/square meter. The fiscal gap soared from 180 million RMB in 2001 to 11.8 billion RMB in 2020, with a revenue-to-expenditure ratio of less than 25%; every 0.04 RMB of fiscal revenue needs to cover 1 RMB of public expenditure. This is the complete process of a city's functional death: young people leave, schools close, housing prices collapse, finances go bankrupt, and eventually, even the funds to maintain basic operations are difficult to sustain.

In 2019, the news of buying a 55-square-meter house for 19,000 RMB in Hegang sparked heated discussion. Five years later, cities with housing prices below 1,000 RMB have spread from the Northeast all the way to Sichuan, Hunan, Hubei, Guizhou, and Guangxi; the so-called "joke" is becoming a reality faster than people expected. In 2024, Qiqihar lost 129,000 people in one year, and Suihua lost 99,000. The 157 shrinking cities nationwide are moving along this trajectory. Even the four municipalities directly under the central government saw no population growth in 2024: Shanghai decreased by 72,000, Beijing by 26,000, and Chongqing by 9,600. When a city's long-term population outflow exceeds 10%, its housing market crosses the "sluggish" stage and goes straight to "zero."

This triggers a self-reinforcing vicious cycle: population outflow leads to shrinking housing demand, inventory accumulation triggers falling house prices (house prices in third- and fourth-tier cities continue to fall by 3%-5% annually), land transfer fees subsequently plummet (land revenue in third- and fourth-tier cities has dropped by more than 30%), and local governments are unable to maintain public services (school mergers, bus service suspensions, halving of street lights). The decline in service quality drives more population outflow. This chain lacks a self-correction mechanism; once a net population outflow trend is formed, it accelerates due to the shrinkage of public services. Goldman Sachs predicts that actual housing prices may not bottom out nationwide until 2027, with another 10% drop required. But this is only a national average; for third- and fourth-tier cities with long-term population outflows, the concept of bottoming out may not exist. If by 2028, the number of cities nationwide with a housing-to-household ratio exceeding 1.2 surpasses 100 (currently about 60), and the first batch of regional bank bad debt events caused by property values hitting zero occurs, it will mark the evolution of the "death cross" from an individual city phenomenon to a systemic risk. By then, those affected will far exceed the residents of third- and fourth-tier cities; all depositors and taxpayers will ultimately pay for the bank's bad debts.

The Bill for the Triple Identity

Evergrande's debt of 2.4 trillion yuan, Country Garden's debt of 1.14 trillion yuan, and the bankruptcy of 1,660 real estate enterprises. These staggering figures are often categorized as "corporate crises," as if the losses were confined to the boardrooms of aggressively expanding companies. However, the limited liability system was originally designed to cut losses; what Xu Jiayin lost was equity, not his entire personal fortune. The real problem has never been who lost money, but who is unable to stop the bleeding. The evaporation of $18 trillion in household wealth, the monthly mortgage payments on 8 million unfinished homes, the suppressed returns on deposits due to low interest rates, and the future tax burden to bail out local government debt—these losses are ultimately concentrated on the same group of people: China's urban middle class. They are simultaneously homebuyers, savers, and taxpayers, paying for the crisis in every capacity, yet none of these identities offers a possible exit.

Water Flows Downward

The distribution of losses in a financial crisis follows the laws of physics: water flows downward, and losses accumulate at the lowest elevation. Who sits at the lowest elevation depends on who has the least negotiating leverage.

Developers stand on the hillside, where the limited liability system acts as a flood levee. Xu Jiayin has lived in more luxury mansions than most people will see in a lifetime. When the flood hits, what he loses is the investors' money; his personal wealth is insulated through legal tools such as family trusts. The overseas assets of Yang Huiyan's family did not vanish because of Country Garden's losses. Between 2021 and 2024, 1,660 real estate enterprises went bankrupt or were deregistered, and at least 10 listed developers were delisted. This is not just death; it is also a form of liberation, as bankruptcy is the legal right granted to enterprises to stop their losses. In 2024, national sales of new commercial housing fell by approximately 50% from their 2021 peak, and the cumulative amount of defaulted domestic and offshore bonds by real estate companies exceeded 5000 billion yuan. The developers' losses are real, as are the unpaid wages of hundreds of thousands of construction workers and the accounts receivable of small and medium-sized suppliers. But the limited liability system ensures that the losses of shareholders have a ceiling, while those who cannot protect themselves with legal tools face no such limit.

Banks stand halfway up the mountain. The 53 trillion yuan in real estate-related loans (including 38 trillion in mortgages and 13 trillion in development loans) constitutes the largest single exposure on the asset side of the banking industry. To cooperate with debt resolution and the "ensuring delivery of housing" (Baojiaolou) initiative, the banking industry's net interest margin (NIM) has been squeezed from 2.2% to a historic low of 1.53%, ROE has dropped from 12% to 8.1%, and ROA has fallen to 0.6%. The non-performing loan (NPL) ratio for real estate at the Bank of China is 4.94%, while the industry average is about 3.2%—a doubling in three years compared to approximately 1.5% in 2019, which has indeed weakened profitability. Defaults on real estate-related trust products are frequent; institutions like Zhongrong Trust, which managed hundreds of billions, have lost their ability to make payments. Bad debts in the trust industry, totaling 2-3 trillion yuan in real estate exposure, are being transmitted to the banking system. However, the central bank, as the lender of last resort, always stands behind the banks. As long as liquidity is maintained, the mountainside will not be submerged.

Homebuyers stand at the bottom of the valley. Real estate accounts for over 70% of total Chinese household assets. Secondary housing prices in 70 cities have fallen for more than 24 consecutive months, and a cumulative decline of 40% to 50% means the bulk of a lifetime's savings has evaporated. Unlike corporations, China has no personal bankruptcy protection mechanism. 1,660 real estate companies can go bankrupt, and the "Big Five" banks will not go bankrupt, but the homebuyers' mortgage debt will never disappear. Whether the house is unfinished or the asset has already turned into negative equity, the monthly mortgage must be paid. Water flows downward, and losses accumulate on those with no negotiating power.

Who is Being Saved?

300 billion yuan—this is the quota for the affordable housing re-lending facility launched by the central bank, interpreted by the market as the "National Team" buying up housing inventory.

The flow of this money is worth deconstructing: the central bank lends to commercial banks, commercial banks lend to local state-owned enterprises (SOEs), and SOEs acquire existing inventory from developers. The beneficiaries are also worth analyzing: developers receive much-needed liquidity to repay debts, banks swap potential bad debts for loans backed by government credit, and homebuyers are not on the list of direct beneficiaries. More critical is the comparison of scale: Goldman Sachs estimates that to digest current inventory and complete projects already started, approximately $1.1 trillion (about 8 trillion yuan) is needed. 300 billion yuan is about 4% of the required funds; it is like a house on fire, and the fire department has brought a cup of water.

The logic of the 4 trillion yuan "white list" loans is similar; this money is lent to developers for "ensuring the delivery of housing." For homebuyers, what they receive is not the option of "returning the house and getting a refund," but a promise to "wait, the house will be finished": continue paying the mortgage to sustain an expectation of delivery that may be delayed by several years. Policies such as reducing the down payment ratio from 30% to 15% and lowering existing mortgage rates by 50 basis points have indeed eased the monthly payment burden, but for families already in a negative equity position, easing the mortgage is not the same as recovering the principal. The proportion of completed home sales rose from 12.7% in 2020 to 30.84% in 2024; the policy is mending the fold after the sheep have been lost. For the owners of 8 million unfinished homes who have already fallen into the trap, the sheep are already gone. After the "924" policy in September 2024, transaction volumes in first-tier cities briefly doubled, while third- and fourth-tier cities showed almost no reaction. The policy is like a shot of adrenaline; it can make the heart beat a few more times, but it cannot cure organic lesions.

More hidden than the real estate bailout is the 10 trillion yuan local government debt swap. The media calls it "debt resolution" (Huazhai), a verb chosen exquisitely because "resolving" is neither "repaying" nor "reducing." Just as chemical changes follow the law of conservation of mass, the total amount of debt has not decreased by a single cent; only its form has changed. High-interest, short-term implicit debt (with interest rates of 7%-10%) is being swapped for low-interest, long-term explicit debt (with interest rates of about 2.3%), with maturities extended from 3-5 years to 15-30 years. Of the 78 trillion yuan in LGFV debt (58% of GDP), only 14.3 trillion yuan has been recognized by the Ministry of Finance as "implicit debt," while the rest remains off-balance-sheet through various accounting maneuvers. The essence of intergenerational transfer is not to resolve risk, but to postpone the eruption of risk until after current officials have retired.

Triple Identity

The Wall Street Journal estimates that Chinese households have lost about $18 trillion due to falling house prices. For the 41.5% of urban households owning two or more properties, the investment decisions made by leveraging at the peak in 2020-2021 have irreversibly destroyed household balance sheets. But this is only the first blow.

The same group of people also takes on the role of savers. To maintain the survival of banks under low interest margins, deposit rates have been suppressed below 1.5%. The balance of national household savings deposits exceeds 130 trillion yuan; for every 1 percentage point reduction, more than 1 trillion yuan in interest is transferred annually from the pockets of savers to the banks' income statements. This is essentially a "financial repression tax" levied on savers, using their interest losses to subsidize the banks' bad debt provisions. S&P Global Ratings estimates that the actual non-performing asset ratio of China's banking industry could be as high as 7.0%, far higher than the officially disclosed 3.2%. The difference of 3.8 percentage points is temporarily hidden through accounting maneuvers such as extensions, restructuring, and special-mention loans. The balance of non-performing loans has reached 3.38 trillion yuan, and defaults in the trust industry's 2-3 trillion yuan real estate exposure are frequent. When approximately 10% to 15% of the 38 trillion yuan in mortgages are on the verge of negative equity, these loans no longer correspond to high-quality assets but to a melting block of ice. The banks' income statements look acceptable today, but a time bomb is buried in their balance sheets. Homebuyers have already felt the shockwave of the explosion; the banks just haven't admitted it yet.

The number of foreclosed properties surged from 380,000 in 2020 to about 800,000 in 2024. If the annual increase in foreclosures exceeds 500,000 by 2027, and the first regional bank restructurings are triggered because the scale of negative equity exceeds the banks' provision coverage capability, that will be the signal that loss distribution has moved from "implicit transfer" to "explicit eruption." At that point, only two choices will remain on the table: let the banks recognize the losses (shareholder equity goes to zero or even deposit haircuts), or let the central bank print money on a massive scale to dilute the debt (harming all currency holders). No matter which path is taken, the bill will be paid by the same group: China's urban middle class. Over the past decade, they were encouraged to "use six pockets to buy a house"; in 2022, a mortgage boycott broke out across more than 300 housing projects; now they endure the double blow of asset shrinkage and unfinished homes; and in the future, they will continue to bail out the remains of the feast through low interest rates and high tax burdens. In a crisis, the silent majority bears the largest share, yet possesses no negotiating leverage.

Three Paths

Goldman Sachs predicts a bottom in 2027, UBS believes the drag will narrow by 2026, and a Reuters survey suggests a return to slight positive growth by 2027. The prevailing market sentiment resembles a soothing melody: the worst phase has passed, and now it is simply a matter of waiting for time to heal the wounds. However, this type of optimism relies on a fragile assumption: that policy intensity reaches at least the historical average of past crises, while the external environment remains stable. Goldman Sachs itself admits that China's current stimulus intensity is lower than the 21 historical cases compared in its model. If the aforementioned premises cannot be met, the timing of the bottom will be postponed—perhaps not to "2027," but to "2029" or even later. China's real estate crisis is not a linear process with a clear endpoint, but a dynamic multiple-choice question. The options before policymakers are not an inevitable recovery, but three distinct paths: slow bleeding, active liquidation, or a full-blown crisis. Which path is chosen is not a natural evolution of market forces, but depends entirely on the speed and scale of policy intervention.

Three Types of Surgery

Conservative treatment is the current strategy, which involves letting the market clear naturally over a long period by restricting supply and fine-tuning demand. The cost is long-term economic malaise and functional degradation. The GDP deflator has been negative for nine consecutive quarters, reaching as low as -1.3% in the second quarter of 2024—a classic manifestation of deflation. Deeper damage is reflected in the active contraction of balance sheets: the balance of individual housing loans has seen negative growth for nine consecutive quarters, with the total stagnating at 38 trillion yuan. The household sector is undergoing a painful deleveraging process, with the growth rate of total retail sales of consumer goods falling from 17.1% in 2011 to 8% in 2019, and further down to 3.5% in 2024. This strategy of "trading time for space" is essentially allowing the wealth of the middle class to gradually evaporate during a protracted decline to avoid violent shocks to the financial system. The patient survives, but may never regain health.

Radical surgery would require excising the necrotic parts, which demands that the government commit fiscal resources equivalent to 5-10% of GDP to acquire existing housing inventory and provide subsidies to homebuyers, rather than merely injecting liquidity into developers. The lesson from Japan lies in procrastination: land prices fell by 54% between 1991 and 2005, and GDP grew by only 2.6% over the twenty years from 1997 to 2017 (an annualized rate of 0.13%). China's advantage is that its decision-makers are familiar with the "balance sheet recession" theory proposed by Richard Koo; theoretically, they could avoid Japanese-style hesitation. But this is only theoretical. The cost of implementing radical surgery is the transformation of private sector bad debts into explicit public debt, which could cause the government debt-to-GDP ratio to rapidly exceed 100%. This is a massive political gamble, requiring policymakers to make a difficult choice between short-term acute pain and long-term stagnation.

If conservative treatment drags on for too long and radical surgery fails to be initiated, emergency resuscitation will become the only option. This is the path of a full-blown crisis, where the risk is not limited to real estate itself but lies in its derivative financial chains. Currently, only 19% of Local Government Financing Vehicles (LGFVs) can rely on their own cash flow to pay interest; the rest need to roll over debt by borrowing anew. Meanwhile, the volume of maturing offshore LGFV bonds has surged by 69.4%, reaching $48.2 billion. The IMF explicitly stated in its 2025 Financial Sector Assessment Program (FSAP) that financial stability risks are "elevated and rising." Once a provincial-level LGFV defaults or a systematically important city commercial bank is taken over, the crisis will rapidly shift from chronic blood loss to acute organ failure. At that point, the cost of rescue will no longer be 5% of GDP, but rather manifold monetary over-issuance and currency devaluation. Path A (slow bleeding) is not an independent path; its endpoint is often Path C (full-blown crisis): as bank provisions are exhausted, chronic illnesses eventually turn into acute emergencies.

Bottoming Out is a Set of Prerequisites

Goldman Sachs' prediction of "home prices bottoming in 2027" is based on a median analysis of 21 historical global real estate crashes. Data shows that the median real estate downturn lasts about 6 years, with a price decline of approximately 30-40%.

The aforementioned predictive model ignores a key variable: in those 21 crashes, the vast majority of governments implemented robust fiscal stimulus during the crisis. Goldman Sachs admits that China's current policy support intensity is lower than the average of these 21 historical cases. This indicates that the core prerequisite for "bottoming in 2027"—that policy intensity is not lower than the historical average—does not hold true in China. If policy intensity remains below the average for a long time, the downward cycle is likely to exceed 6 years, and the bottom may not appear before 2028. UBS's forecast is more cautious: real estate sales and new construction starts will still decline by 5-10% in 2026, continuing to drag down GDP growth by 0.5 to 1 percentage point. "Bottoming out" is never a fixed point in time, but a set of prerequisites that must be met. If these conditions are missing, the timing of the bottom will be postponed.

Both optimists and pessimists suffer from projection bias. For example, optimists point out that transaction volumes in Tier-1 cities doubled after September 2024 and that Shenzhen's population grew by 200,000, believing "the worst moment has passed." But they overlook market divergence: stabilization in Tier-1 cities does not represent national stabilization; populations in 173 cities are still shrinking, and the inventory turnover period in Tier-3 and Tier-4 cities is still climbing above 30 months. Pessimists believe China will repeat Japan's "Lost Decades," but they overlook the difference in speed: it took only 2 years for China's real estate investment as a share of GDP to drop from 10% to 6%, whereas it took Japan 20 years to complete this process.

China is not making a slow landing; it is attempting to open a parachute during a free fall.

The pain brought by rapid adjustment is more concentrated, but it may also shorten the clearing cycle, provided that policy can catch the falling momentum. Richard Koo pointed out that China "knowing what a balance sheet recession is" is a huge advantage, but knowing the diagnosis is one thing; acting according to the diagnosis is another. Inventory buyback funds have only covered 4% of the market gap, debt swaps have merely extended maturities without reducing the scale, and the "whitelist" policy protects projects rather than homebuyer confidence. Mathematically, a 4% coverage rate combined with the assumption that "policy makes no major mistakes" is equivalent to a test of faith. China's current state is: holding the CT scan and discussing whether to take medicine while organ function deteriorates.

Between 1991 and 2001, Japan repeatedly introduced "insufficient" stimulus packages, each briefly boosting confidence followed by deeper disappointment, until Junichiro Koizumi implemented structural reforms alongside quantitative easing in 2001, at which point the market truly bottomed. China's policy rhythm in 2024-2025 is almost repeating Japan's mistakes from 1993-1997: every policy introduced is half of what the market needs, and every time the policy effect fades, the next round of policy faces even more stubborn pessimistic expectations.

Signals and Choices

Over the next three years, three verifiable signals will determine which outcome China moves toward.

Signal A is the confirmation of Path A (slow bleeding): by the end of 2026, home prices in Tier-1 cities turn positive year-on-year, but Tier-3 and Tier-4 cities continue to fall by more than 15%, and the GDP deflator remains negative. This would mark the solidification of a national split and the entrenchment of deflationary expectations. Most market participants are currently betting on this, believing it to be the "soft landing" with the lowest cost.

Signal B is the initiation of Path B (active liquidation): the government announces an additional fiscal stimulus package of more than 5% of GDP, with funds flowing to homebuyers or used for large-scale inventory acquisitions rather than to developers or infrastructure. This requires decision-makers to be willing to bear the risk of a short-term debt surge in exchange for long-term balance sheet repair.

Signal C is the alarm for Path C (full-blown crisis): a provincial-level LGFV publicly defaults for the first time, or a city commercial bank is forcibly taken over, or the annual increase in foreclosed properties exceeds 500,000 units. This is the moment when systemic risk turns from an undercurrent into an open flame.

The most likely evolutionary route at present is not a single path, but a slide from Path A toward Path C.

If the status quo is maintained, Path A may continue until mid-2026. Because policy intensity is consistently insufficient to reverse deflationary expectations, the interest payment capacity of LGFVs will further deteriorate, and the banking system's provisions will be exhausted, eventually forcing the system to slide into Path C. The room for choice then will be narrower than it is now, and the cost will be higher. Every year that "radical surgery" is delayed, the probability of "emergency resuscitation" rises further. The difference is: a 20-year landing allows time to find a runway; a 2-year fall only allows time to find a parachute. Whether the parachute can open depends on the speed of decision-making, not the wind speed.