China has been on the verge of collapse for more than 20 years. In 2001, lawyer and commentator Gordon Chang published “The Coming Collapse of China.” The book famously predicted that China’s economic model would fail within a decade.

The decade passed. The prediction was revised, republished and absorbed into a durable genre that has survived every missed deadline. The crisis, by Chang’s and others’ readings, was always near.

Consider two other prominent figures on opposite ends of the China collapse spectrum. Nouriel Roubini earned the nickname “Dr. Doom” even before predicting the 2008 global financial crisis. In 2011, he warned that China faced a meaningful probability of a hard landing.

He pointed to its runaway debt, over-investment and infrastructure projects disconnected from real demand. The crash, he suggested, would come after 2013.

By 2015, as the hard-landing consensus reached its peak, Roubini reassessed the evidence. He rejected the collapse scenario and argued instead for a “bumpy landing” — growth slowing, but without systemic failure. The prediction changed because the evidence changed.

Peter Zeihan took a different angle. For more than a decade, the geopolitical strategist has argued that China’s economy and political system face structural collapse. Demographics will shrink the workforce. Export dependence will undermine growth. In books and interviews, Zeihan’s timeline shifts, but his conclusion remains remarkably consistent.

The challenges Zeihan identifies are real. China’s population is aging. Export markets are more contested. Yet the collapse he predicts has not materialized. China has responded to demographic pressures through automation and moved steadily up the industrial value chain.

The contrast is revealing: Roubini changed his mind while Zeihan moved the date.

Wall still standing

When the Shanghai stock market plunged in 2015, commentators warned of a hard landing. When property developer China Evergrande Group defaulted in 2021, comparisons with Lehman Brothers appeared almost immediately. Yet the collapse never arrived. China’s wall is still standing.

This is not to suggest that all is well. Household wealth remains tied to a declining property market. Youth unemployment rose to such a high level that authorities stopped publishing the figure. Export markets have become more difficult with rising protectionism in the West.

The people forecasting trouble were not inventing these problems – they were identifying genuine stresses. What they have repeatedly misjudged, though, was not the existence of stress but how it would propagate.

A prediction that repeatedly fails and is then quietly postponed changes its nature. It ceases to be a forecast and becomes a standing expectation that survives its own disconfirmation.

The question is no longer whether China faces serious challenges. It clearly does. The question is why the forecasts keep failing in the same direction. Error from bad data scatters. Some forecasts are too optimistic, others too pessimistic. Over time they average toward reality. China’s collapse forecast does not scatter, it leans.

The collapse deadline keeps receding. 2011 became 2012, then 2016. The hard landing was pronounced again after the 2015 stock crash and continued variously though the trade war, through the pandemic, through Evergrande’s default.

Across those same two decades, China’s economy grew, household incomes more than quadrupled, and the industrial base moved steadily up the value chain. A forecasting error that consistently points in the same direction reveals more about the observer than about the object.

Why the narrative persists

Three forces explain the durability of the collapse narrative. None requires anyone to be lying.

First, the conclusion is useful. An investor gets a reason to avoid Chinese assets that sounds like analysis, not anxiety. A government gets risk management in place of admitting a peer competitor has arrived. And the media gets a better story: a rising rival is complicated; a collapsing rival gets clicks.

When a conclusion is this welcome, confirming evidence is waved through while contradicting evidence is asked for its papers. Being wrong carries almost no professional cost. A forecaster can miss the same call for two decades and remain a sought-after authority. Nobody decides this consciously. It is what wanting does to looking.

Second, the assumption is old. A long tradition in Western thought holds that an economy cannot function without the institutions the West built: independent central banks, courts that constrain the state, free flowing information. Hold that assumption firmly enough, and Chinese growth begins to look like a trick borrowed from the future. Collapse becomes a deduction, not a prediction.

The conviction shows itself most clearly in the assumption that China could copy but never invent. Yet Chinese firms now lead in electric vehicles, batteries and renewable energy, among a growing list of other next-generation technologies. The premise has survived even as reality has steadily eroded it.

Third, the models were built elsewhere. The tools used to assess economic vulnerability learned their trade in systems where the state acts as referee. They watch private debt, leverage and property valuations. Those variables matter in China, but they do not transmit stress the same way.

Evergrande is the clearest example. Comparisons with investment bank Lehman Brothers seemed unavoidable. Yet Lehman’s collapse occurred within a system of largely independent creditors. China presents a different configuration: state-owned banks and government-directed restructuring have altered the pathways through which distress could spread.

The result was not the absence of crisis – it was a different kind of crisis. Developers defaulted, property values fell and growth slowed. But the chain reaction many analysts expected never materialized. The model identified genuine vulnerabilities. What it misjudged was the mechanism of propagation.

The opposite illusion

And yet none of this means the wall is sound. Those who warned about China’s weaknesses were right about many things. Developers did default. Property values did fall.

Demographics are turning, and they will not reverse easily. However, the use of “imminent“ has been wrong for 20 years. “Fragile“ has not been wrong at all. That is the honorable way to miss — the path Roubini took.

The same instrument that exposes the collapse narrative can be turned on its mirror image. If a China perpetually on the verge of collapse is useful to believe, so is a China that cannot fail. Both narratives rely on selective attention. Both risk mistaking conviction for evidence.

The models used to forecast China’s economy are de facto maps: compressions of a vast and complex reality into a handful of neat, legible variables. The territory itself remains far messier.

When the territory refuses to behave as the map says it should, the disciplined response is to question the map. The reflexive response is to question the territory: the data must be fake, the growth hollow.

So, again, why have all the collapse forecasts been so consistently wrong? In part, because forecasters have been watching and weighing the wrong indicators.

The new middle class was not built primarily in Shanghai or Shenzhen but in inland cities such as Chengdu, Hefei, Xi’an, and Zhengzhou, where income growth was driven by industrial expansion and infrastructure investment rather than by speculative gains in coastal real estate.

If a genuine systemic decline is approaching, it will appear there first – in household incomes and altered spending patterns among the inland middle class.

If those indicators begin to show a sustained contraction, the collapse thesis will finally have the transmission mechanism it has long lacked. If, on the other hand, they remain resilient, analysts and critics may again need to reconsider their flawed assumptions.

RELATED ARTICLESMORE FROM AUTHOR