China's Ghost Cities: Failure or Strategy According to Wade Shepard

China's ghost cities are the large new districts built during the property boom of the 2000s and 2010s that stood empty for years: twelve-lane avenues without cars, residential towers with no lights on, closed shopping malls and stadiums without crowds. The American journalist Wade Shepard travelled through them for years and described them in Ghost Cities of China (2015), the book that turned the phenomenon into a global debate. His question is the one that remains open: are they the greatest planning failure in history or a long-term urbanization strategy that the Western gaze failed to read.

Why China built cities before it had residents

Shepard first explains why they were built. Since the fiscal reform of 1994, Chinese local governments have depended on land sales to fund themselves, so that urbanising is their main source of revenue. Add growth targets set from Beijing, the need to house hundreds of millions of rural migrants and a property market in which housing works as households' main savings asset, with families buying empty flats as investments. The result was a machine that produced complete cities before immediate demand existed, on the calculation that demand would come.

Ordos, Zhengdong and Tianducheng: the cases and how they evolved

The cases he describes became famous. Kangbashi, the new district of Ordos in Inner Mongolia, planned for a million residents and almost deserted in 2010. Zhengdong, the business district of Zhengzhou, photographed empty by the international press. Tianducheng, the replica of Paris with its Eiffel Tower near Hangzhou. And dozens of new towns in every province. Shepard revisited the same places years later and found what the photos did not show: many were filling up. Zhengdong passed a million residents, Kangbashi became the administrative and educational centre of Ordos, and the schools and hospitals moved from the old cities pulled population with them.

Failure or strategy: build first, populate later

Hence his nuanced reading. Ghost cities are, in many cases, cities under construction that fill over a decade or two, following the same pattern as Shenzhen or Pudong in Shanghai, also dismissed as empty in their early years. The Chinese strategy consists of building the infrastructure first and attracting the population afterwards by relocating institutions, universities and industries. But Shepard does not excuse the model its costs: local government debt, the expropriation of farmers with minimal compensation, housing bought as an asset and never lived in, and cities that, when demand did not come, were left half finished.

Evergrande and the limit of the model

Time has proved both halves of the argument right. Many new districts have consolidated and are now part of China's urban fabric, which has housed the largest internal migration in history. At the same time, the crisis of developers such as Evergrande since 2021 and estimates of tens of millions of empty homes have shown that the model based on land sales and housing as investment had reached its limit, with families paying mortgages on flats never completed. The Chinese government has since redirected policy toward housing that is lived in rather than speculated on.

For planning, ghost cities leave two lessons. The first is about time: a new city takes decades to fill, and judging it by its first years can be a mistake, as Shenzhen or Kangbashi show. The second is about the model: building for investment rather than for life produces cities that may or may not fill, but that in either case have displaced those who lived on the land, indebted those who financed them and treated housing as an asset before a home. Shepard makes clear that the final judgement is not about photos of empty avenues but about who pays and who lives there.