The Biggest Bet in History

In January 1992, Deng Xiaoping climbed into a golf cart at Splendid China, an amusement park in Shenzhen, and set in motion what would become the largest land bubble in human history. He was eighty-eight years old, three years removed from the catastrophe of Tiananmen Square, watching as the reform project he had spent decades building was stalled by conservative rivals inside the Communist Party. His southern tour, a series of speeches across Shenzhen, Zhuhai, and Guangzhou, was a political act as much as an economic one, a declaration that reform was irreversible and that those who did not believe it should lose their positions. News of the tour was suppressed by state media controlled by his rivals and reached the mainland only two months later, through an unauthorised report in a local Shenzhen newspaper.

The speech worked. By the end of 1992, Jiang Zemin had endorsed Deng’s vision, and the fourteenth National Congress of the Chinese Communist Party introduced the concept of the “socialist market economy” to the world. Reform was back. And within it, largely unnoticed at the time, were the seeds of a property market that would dwarf every comparable episode in economic history.

The architecture of a boom

Understanding how China’s housing bubble grew to its extraordinary size requires understanding two decisions made in the 1990s that, combined, had consequences neither was intended to produce alone.

The first was the 1994 fiscal reform. For most of the preceding decades, China’s economic management had been highly decentralised: local governments kept the majority of the tax revenue they raised and bore most of the spending responsibilities. Beijing, by the early 1990s, was collecting only about 22 percent of total government revenues. Premier Zhu Rongji’s solution was a sweeping recentralisation. From 1994, the central government would take the bulk of tax revenues, with the local share falling from 78 percent to 44 percent in a single year. Spending responsibilities, however, were barely changed. In 1993, local government revenues had been sufficient to cover all of their expenditures. In 1994, the revenues they were permitted to keep covered less than 60 percent.

The result, as Stanford political scientist Jean Oi has argued, was a structural fiscal gap that has never been formally closed. Local governments were legally prohibited from running deficits or issuing bonds. They were simultaneously responsible for infrastructure, education, healthcare, social security, and environmental protection, with insufficient revenues to fund any of it. Something had to give.

The second decision was the gradual privatisation of housing. For most of the communist era, housing had been allocated by state employers and was woefully inadequate in size, quality, and availability. In 1994, state employees gained the right to purchase their existing homes at heavily subsidised prices. In 1998, a new law permanently severed the link between government employment and the provision of housing, opening the market to private developers. In the years that followed, China’s GDP grew at roughly 10 percent per year. Residential real estate investment surged at more than 20 percent annually.

What connected these two decisions was land. When local governments needed a new source of off-budget revenue to fill the fiscal gap that Beijing had imposed, land was the most accessible tool available. Revenue from land sales, the proceeds from granting developers use rights over state-owned territory, was classified as a separate, off-budget income that did not have to be shared with Beijing. From less than 10 percent of local government income at the end of the twentieth century, land revenues rose to roughly two-thirds by 2010. In some provinces, the dependence ran even higher: Zhejiang, for example, derived close to 70 percent of its fiscal resources from land sales.

Local governments thus had every incentive to keep land prices rising. Selling more land at higher prices was not merely profitable. It was the mechanism through which local officials funded hospitals, roads, schools, and their own administrative operations. The fiscal and property markets became structurally fused. Land finance was not a side phenomenon of China’s development model. It was the model.

The people who needed somewhere to put their money

On the demand side of the market, a separate set of forces was pushing in the same direction. China’s rapid growth created a large and growing class of households with money to invest for the first time in their lives. Their options, however, were extraordinarily limited.

Interest rates on bank deposits were kept deliberately low, ensuring that returns were often negative in real terms. The stock market, which reopened in Shanghai in 1990, delivered a volatile and largely miserable experience for retail investors. And while China had opened its goods markets to the world, the government maintained strict capital controls that made it very difficult for ordinary households to invest abroad. For most Chinese savers, property was not merely the best available investment. It was very nearly the only one.

The social pressures reinforcing this preference were considerable. As the communist welfare system was dismantled, state-provided housing, pensions, and healthcare gradually withdrawn, households needed to accumulate their own security. The average Chinese pension, according to the International Labour Organization, amounted to the equivalent of less than $25 per month in 2020. The country’s hukou system of household registration meant that the hundreds of millions of workers who migrated to cities from their birthplaces lost access even to the limited welfare they might have received at home. A family’s financial future was increasingly stored in its housing, to a degree unmatched almost anywhere else in the world. By 2018, an estimated 76 percent of Chinese household wealth was held in real estate, compared with around 41 percent in Japan and 28 percent in the United States.

Old cultural norms reasserted themselves almost as soon as the legal restrictions on property ownership were dropped. The pre-communist expectation that a man seeking a bride should own a home returned quickly, adding social pressure to the already powerful economic logic of property investment. By 2018, research from the Survey and Research Center for China Household Finance found that 44 percent of property purchases were made by buyers who already owned at least one home, and another 25 percent by those who owned more than two. Buying apartments to leave empty, capturing capital gains rather than rental income, became standard practice across the middle class.

The metrics of this speculation were striking. The price-to-rent ratio in Chinese cities ranked among the highest in the world: purchasing a property and renting it out took nearly 48 years to recover the initial investment, making the economics of buy-to-let essentially indefensible on any conventional calculation. In the 50 largest Chinese cities, the average price-to-income ratio reached 13.4, higher than San Francisco at roughly 10, and London at around 9, cities which are themselves considered expensive outliers in their own national housing markets. By some estimates, the total valuation of China’s residential real estate reached roughly twice that of the American market, despite China’s economy being around 25 percent smaller.

None of these metrics deterred buyers. What drove purchasing decisions was not the rental yield or the earnings multiple but the expectation, rational for a remarkably long time, and ultimately irrational, that prices would simply continue rising. China had, like Japan before it, produced its own Land Myth: the belief that property was a special asset, exempt from the normal logic of supply and demand, whose value the government would protect indefinitely.

The developers and their model

Into this environment came the property developers. Companies like China Vanke, founded in the early days of the Shenzhen stock exchange, and Country Garden, established in 1992, arrived early. Evergrande, founded in 1996 by Xu Jiayin, then a thirty-four-year-old former steel company manager who had taken inspiration from Deng’s southern tour and headed to Shenzhen to make his fortune, arrived slightly later but grew faster and borrowed more aggressively than almost any of its peers.

The developer business model was straightforward in structure, though extraordinary in its execution. Developers would acquire land from local governments, often at cheap prices on marginal plots overlooked by competitors, finance construction through a combination of bank loans, bond issuance, and pre-sales, and then use the proceeds to acquire more land. Speed was everything. In a market where land prices rose reliably and competitors were always moving to acquire the best plots, the imperative was to capture as much land as quickly as possible. The opportunity cost of hesitation was visible and immediate.

The financing structure grew increasingly elaborate. Chinese banks’ restrictions on lending to property developers drove companies towards international bond markets, where they became some of the largest issuers of US dollar debt in Asia. High-yield bonds from developers like Evergrande offered returns of between 10 and 20 percent, attracting fund managers around the world at a time when global interest rates had been driven to historic lows by the aftermath of the 2008 financial crisis. Between 2010 and 2020, Evergrande’s bond debts alone rose from less than 9 billion yuan to more than 230 billion.

The most ingenious, and ultimately most dangerous, financing innovation was the pre-sale. Developers discovered that they could sell apartments to buyers a year or two before completion, receiving the purchase price upfront while still in the construction phase. Buyers accepted a modest discount in exchange for locking in current prices before they rose further. For developers, pre-sales provided cheap, flexible capital that was far easier to access than bank credit. Between 2015 and 2019, pre-sale financing grew by 103 percent, becoming the single most important source of funding for the sector. By the eve of the Covid-19 pandemic, roughly 85 percent of Chinese property sales were of homes not yet completed.

The pre-sale model rested on an absence that might seem extraordinary in retrospect: there was no official escrow system preventing developers from using the funds from one buyer to complete construction for a different set of buyers from months or years earlier. Money flowed freely across projects. When developers sold apartments whose foundations had barely been laid, the cash received was routinely used to finish housing promised to a previous generation of buyers. The structure was not technically a Ponzi scheme, since the underlying assets, land and buildings, were real. But its functioning required continuous sales growth, rising prices, and uninterrupted access to fresh financing. A slowdown in any one of these conditions would create cascading pressures across the system.

Evergrande took all of these dynamics to their logical extreme. By its peak in 2017, the company’s assets had grown from around 5 billion yuan in 2004 to a peak of 2.3 trillion yuan, an average growth rate exceeding 50 percent per year for over a decade. Its founder, Hui Ka Yan, was China’s richest individual, ahead of technology billionaires like Jack Ma and Pony Ma. The company had diversified into electric vehicles, healthcare, insurance, and professional football. Guangzhou Evergrande topped the Chinese Super League every year from 2011 onward. What was less visible was that the company had long since crossed the threshold at which new debt was being taken on primarily to service old debt. The basic structure of a scheme requiring perpetual growth had been established. The company needed the market to keep rising not just to prosper, but to survive.

The financial analysis of Evergrande’s position revealed several layers of fragility. The official debt-to-assets ratio of the company had appeared to improve in the years before its collapse, falling from around 41 percent in 2017 to 24 percent by 2021. In practice, this improvement was largely cosmetic: the company had replaced formal borrowings with unpaid bills to suppliers, which do not appear as debt in the same part of the balance sheet. When trade payables were included, the true leverage ratio had barely moved. Meanwhile, the coverage ratio, the relationship between operating earnings and interest payments, had fallen below 1 by 2020, meaning the company’s revenues were no longer sufficient to cover its financing costs without selling assets or borrowing more. Roughly half of Evergrande’s formal debt at any given time needed to be rolled over within a year, making the company acutely vulnerable to any disruption in its access to new financing. When one factored in supplier obligations, that proportion rose to around 78 percent of total liabilities.

The developer’s asset position offered little comfort. Approximately 90 percent of its property inventory consisted of homes still under construction, not completed units that could be sold quickly to raise cash. The remaining 10 percent was finished housing whose sale prices were already declining. Producing a genuine assessment of Evergrande’s liquidity required recognising that the cash already received from pre-selling homes partly needed to be returned in the form of completed apartments to those buyers, rather than being available to repay creditors. The ability to sell assets quickly enough, and at prices high enough, to meet the company’s obligations in an environment of falling prices and market distrust was, at best, highly constrained.

The shadow financing layer

Behind the formal banking system and the bond markets, a further layer of financing had developed that made the overall picture considerably more complex. Shadow banking in China was not, as in the United States, a parallel system that grew up independently of the formal banks. It was a toolkit that the banks themselves developed to circumvent regulatory limits on how much they could lend to which sectors.

The government had traditionally managed credit creation through administrative instruction: quarterly lending quotas, caps on the proportion of deposits that could be extended as loans, restrictions on exposure to property developers or certain industries. Shadow banking was the response, arrangements between banks and trust companies, securities firms, or wealth management products that allowed credit to reach borrowers in forms that did not appear as formal loans on balance sheets. At their peak, these structures became so layered and complex that the People’s Bank of China began referring to the more elaborate versions as “Russian nesting dolls.” A crackdown beginning in 2016 substantially dismantled the shadow banking architecture over the following years, reducing the potential for contagion to spread from institution to institution if a major borrower collapsed. This was one of the reasons why Evergrande’s eventual default in December 2021, while significant, did not produce the Lehman-style cascade that some had predicted.

The same creative financing impulse shaped local government debt. Unable to borrow directly under the legal framework, local governments created Local Government Financing Vehicles, nominally separate entities that could borrow from banks or issue bonds on their behalf. By 2014, these vehicles existed in large numbers across every province. They issued bonds that were sold, often through the banking system, as wealth management products carrying what market participants understood to be an implicit guarantee from the local government behind them, even though no explicit legal obligation existed. An IMF analysis found that one province’s implicit debt ran 80 percent higher than its explicit, formally recorded obligations. The grey rhino, as China’s own financial authorities described local government debt in 2017, was considerably larger than the official numbers suggested, and had been growing for years with the tacit acceptance of Beijing.

The three red lines and what followed

By the mid-2010s, it had become apparent to China’s leadership that the property sector’s trajectory was unsustainable. Xi Jinping’s mantra, “houses are for living in, not for speculation,” introduced at the end of 2016, represented a clear statement of intent, though for several years it appeared to remain largely aspirational. Local governments, whose fiscal survival depended on buoyant land prices, had little incentive to cool their markets. Any time a city’s growth targets looked threatened, property was the obvious emergency lever to pull.

The decisive shift came in August 2020, when senior officials at the People’s Bank of China and the Ministry of Housing summoned major developers and introduced what became known as the Three Red Lines: caps on a developer’s liability-to-asset ratio, net gearing ratio, and the ratio of cash to short-term debt. The regulations came into force at the end of the year. Under the new rules, developers in violation of all three lines, as Evergrande was, could not increase their total borrowings at all. Those in violation of two could grow debt only marginally, and so on.

The effect on a company like Evergrande, which had relied on continuous borrowing to service existing obligations, was severe. The model that had worked for twenty years suddenly became inoperable. For about eight months after the rules came into force, Evergrande managed to keep going by deferring payments to its suppliers and contractors, a practice that accumulated billions in hidden obligations and halted construction on hundreds of projects for lack of materials. By the autumn of 2021, even this was insufficient. In September, the company failed to make interest payments owed to international bondholders. In December 2021, credit rating agencies formally declared Evergrande in default, just short of twenty-five years after Hui Ka Yan had founded it in Shenzhen.

Evergrande was not alone. In the weeks and months around its difficulties, several other developers, Fantasia Holdings, Sinic Holdings, Modern Land, Kaisa Group, announced that they could not meet bond payments. The analysis of these companies’ finances showed a common structural pattern: debt concentrated at the short end of the maturity curve, meaning that obligations needed to be continuously rolled over; coverage ratios that had been declining for years and in some cases had already turned negative; and asset bases dominated by housing under construction rather than completed inventory that could be converted quickly into cash. The offshore dollar bond market, in which Chinese developers had been among the largest issuers in Asia, seized up almost entirely. Companies that had been able to raise new dollar funding at relatively modest yields found that investors were no longer willing to lend to them at any price.

The direct exposure of China’s financial system to the property sector was considerable. By the time the three red lines came into force, bank lending to real estate, including both developer financing and household mortgages, accounted for a substantial portion of total credit in the system. Mortgage debt dominated this exposure: for every yuan lent directly to developers, roughly ten yuan was outstanding in household mortgage credit. Those households, in many cases, were servicing mortgages on apartments that developers were now struggling or unable to complete. A movement began on social media, and spread rapidly, in which buyers who had made pre-sale deposits and mortgage payments on stalled projects coordinated to stop making their loan payments, a peculiar inversion of a bank run in which, instead of depositors withdrawing from a bank, would-be homeowners declined to keep paying for homes they feared might never materialise.

The contagion through the property sector itself, however, proved harder to contain than the contagion within the banking system. New home sales in December 2021 were around 20 percent below the prior year. New construction starts in the same month were down 31 percent year-on-year, and continued to fall. By 2024, residential real estate investment had declined by roughly a third from its 2021 peak. The expectation that had sustained the market for three decades, that prices could only rise, was finally, visibly fracturing. The pre-sale model, which depended on buyers trusting that their deposits would be returned in the form of completed apartments, lost much of its credibility. Developers found themselves in a slow-motion liquidity crisis in which falling sales reduced their revenues, which reduced their ability to finish projects, which further reduced buyers’ willingness to purchase, and so on.

The local government trap deepens

The consequences for local government finances were severe and, in some respects, self-reinforcing. At the very moment when declining property sales were cutting into land revenues, the dominant source of local fiscal income, local governments were simultaneously being asked to absorb some of the consequences of the property crisis, mobilising state-owned developers to acquire distressed projects from companies like Evergrande and ensuring that stalled housing completions were finished.

Beijing’s initial reaction to the drop in land auction activity was, in practical terms, to redirect local government financing vehicles towards buying the land that private developers were no longer purchasing, essentially using indebted public entities to prop up revenues that were collapsing precisely because private demand had withdrawn. The fiscal logic was circular: the vehicles would borrow to buy land, in order to generate revenues that would service, among other things, the debt they were taking on to buy the land. The mismatch that had always characterised these vehicles, long-duration infrastructure investments funded with short-term debt, was being extended rather than resolved.

The revenue data illustrated the scale of the problem with some precision. Land use right sales fell by close to half in 2022 compared to the prior year. The price per unit of land sold dropped by around 28 percent in twelve months. For local governments that had structured their entire fiscal model around the assumption of continuous land sales at rising prices, this was not merely a revenue shortfall. It was a structural challenge to the sustainability of their finances.

The reform that might address the underlying problem, replacing volatile land revenue with a stable, broad-based property tax, had been under discussion for decades. Experiments in Shanghai and Chongqing had been running for years. The conclusion drawn from the Shanghai pilot was instructive: property tax revenues there amounted to roughly 10 percent of what land sales generated. Even the most optimistic implementation scenario for a national property tax would leave local governments significantly short of what they needed. And beyond the arithmetic, a national property tax posed an acute political problem: the people who would pay it most heavily were the urban middle class whose property wealth was the primary store of their household savings, and whose support for the party’s social contract depended substantially on the continued appreciation of that wealth.

The deeper economic costs

The property boom produced clear and immediate benefits: employment in construction, materials, and related industries; rapid urbanisation and the associated infrastructure investment; and the asset wealth of tens of millions of households. These were not trivial. China’s economic transformation over the three decades of the boom was genuinely extraordinary by any comparative standard.

But research conducted during and after the boom period has identified significant costs that were accumulating throughout, invisible in the headline growth figures. In the cities where land prices rose most rapidly, credit available to small manufacturing companies contracted as banks preferred the safer collateral of real estate over riskier unsecured business loans. One study examining 172 Chinese cities found that where property prices grew fastest, corporate investment by local firms fell by around 21 percent, total output by 36 percent, and measured productivity by 12 percent. The productive capacity of cities most caught up in the property boom was being eroded precisely because of that boom.

The most productive manufacturing companies also appeared to reduce spending on research and development during periods of local property market exuberance, diverting investment toward land speculation. One estimate suggested that productivity in the manufacturing sector between 1995 and 2010 would have been around 0.5 percentage points stronger per year in the absence of the real estate boom, a gap that, compounded over a quarter of a century, becomes very substantial. Meanwhile, for young would-be entrepreneurs, the combination of high property prices and the social expectation of homeownership before marriage discouraged risk-taking in business. Between 2000 and 2010, residential property prices in major cities rose on average by 9.4 percent per year, while the average return on Chinese companies ran at 5.6 percent. The incentive to acquire property rather than build a business was clear and rational.

The misallocation of capital was also visible in the physical landscape. By 2017, estimates suggested that more than a fifth of China’s urban housing stock sat vacant, apartments purchased as investment vehicles, left empty in the expectation of capital gains. The construction of this excess inventory consumed enormous quantities of concrete, steel, copper, and human labour. In certain parts of the country, particularly smaller tier-three and tier-four cities, the volume of housing built relative to the local population created surpluses that may take decades to absorb, if they are absorbed at all.

Where China stands

By the mid-2020s, China’s property market occupies an uncomfortable position. Residential real estate remains among the most expensive in the world by price-to-income standards. Despite years of declining activity, prices in major cities have not collapsed in the way that post-bubble markets in Japan or the United States eventually did. The government retains significant tools to manage the pace of adjustment: state-owned banks can be directed to extend credit to distressed developers, required reserve ratios can be adjusted, state developers can absorb inventory, and capital controls prevent the kind of sudden capital flight that can accelerate a property crash in more open economies.

But these tools manage the symptoms rather than the underlying conditions. The demographic trajectory, a rapidly aging population, a shrinking working-age cohort, declining birth rates, reduces the organic demand for new housing that once absorbed continuous construction activity. The savings rate, while still high, reflects in part the absence of alternative investment vehicles and the inadequacy of the pension system, conditions that are difficult to change quickly. The fiscal system continues to generate structural deficits at the local level that local governments fund through mechanisms, land sales, financing vehicle debt, shadow borrowing, whose sustainability depends on conditions that no longer prevail.

The path forward involves choices rather than solutions. Allowing property prices to fall significantly would impose large losses on the middle-class households whose financial security is stored primarily in their homes, but would make housing more affordable for younger generations and reduce the capital misallocated to empty apartments. Maintaining prices through state support preserves existing wealth but extends the period of stagnation and continues to burden the economy with resources tied up in unproductive real estate. Some combination of fiscal reform, genuinely redistributing tax revenues to local governments, formalising the property tax, shifting spending responsibilities to match revenue, is almost certainly necessary, but would require accepting a degree of central control over local spending that Beijing has historically been unwilling to relinquish, and would impose costs on precisely the constituencies whose support the party values most.

What is clearest is that the conditions that produced three decades of housing appreciation, rapid income growth, mass urbanisation, financial repression, and demographic expansion, are no longer in place. The Chinese economy will likely grow at something closer to 3 to 4 percent by the end of the decade, far below the double-digit rates under which the Land Myth was established and the middle class came to believe that property could only rise. When China’s large middle-aged cohort seeks to liquidate investment properties to fund retirement, it will look for buyers among a smaller and less wealthy younger generation that has been priced out of the market for most of its adult life.

The apartments scattered across China’s cities, gleaming in prosperous Shanghai and Shenzhen, emptying slowly in poorer inland cities, were described by one analyst as “bank accounts in the sky.” The phrase captured something essential about how Chinese households came to think about real estate: not primarily as places to live, but as the safest way to store and grow the savings that neither the pension system, nor the stock market, nor the deposit rate would reliably protect. For a long time, that calculation was correct. The question now is what happens when the bank account stops appreciating, and who, in a system so thoroughly built around the assumption that it always would, is left to absorb the difference.