China's Phantom Cities: Why Millions Counted as Urban Live on Farms

In 2021, a Chinese vlogger uploaded a short video documenting the daily routine of Wang Xiansheng, a corn farmer in Heshun County, Shanxi Province. Wang plowed a hillside, tended his crop, and talked about the weather. The clip would have been unremarkable except for one detail: his national ID card listed his permanent residence as “Jinzhong City.” Online commenters seized on the mismatch. Here was a man working ancestral farmland, yet according to official documents he was an urban resident. The video did not expose a clerical error. It captured a feature of China’s administrative geography that reshapes how the world reads its development numbers. Wang Xiansheng is a farmer who lives in a city because, in China, “city” often means an entire prefecture-level administrative unit—not a contiguous built-up area. Jinzhong City, a prefecture in central Shanxi, governs two urban districts and nine counties, including Heshun. The prefecture’s urban core sits roughly 100 kilometers from Wang’s village. Under China’s statistical system, all 3.2 million people registered within Jinzhong’s borders count toward the “urban” population of that prefecture, regardless of whether they live in apartment towers or mud-brick farmhouses. This is not a definitional quirk unique to Jinzhong. Prefecture-level cities are the default building block of Chinese territorial administration. They function as regional governments, merging what many other countries would separate into municipal and county jurisdictions. The result is that a single “city” can sprawl across thousands of square kilometers and contain landscapes that range from dense high-rises to alpine pastures. Chongqing, often cited as the world’s largest city by population, is a municipality that covers 82,400 square kilometers—an area larger than Austria—and includes entire rural counties. Only a fraction of its 32 million residents live in the urban core. The gap between administrative boundaries and lived urban experience is not a secret. Scholars such as You-tien Hsing have detailed how China’s territorial governance produces a particular kind of urban-rural hybrid, one that the official classification system struggles to capture. The National Bureau of Statistics defines “urban population” based on where people are registered, not on the density of the built environment or the economic activities they pursue. When the urbanization rate is reported as roughly 65 percent, that figure includes millions of people whose daily lives are agrarian, not metropolitan. For financial analysts and economic researchers, this distinction carries weight. Consumption forecasts, infrastructure demand models, and market-sizing exercises often treat urbanization as a proxy for rising incomes, changing diets, and demand for housing and services. If the denominator of “urban” contains a large rural tail, those proxies become less reliable. A prefecture-level city with a high urbanization rate may still have vast areas where household expenditure patterns resemble those of villages, not city dwellers. The spatial distribution of purchasing power, labor markets, and logistics costs can diverge sharply from what the headline number implies. The issue is not that China’s statistics are inaccurate; they are internally consistent but defined on terms that differ from the ones many external analysts assume. International comparisons become especially fraught. When an analyst places China’s 65 percent urbanization rate alongside, say, Brazil’s 87 percent or India’s 36 percent, the numbers are not measuring the same thing. Brazil’s urban classification is based on municipal seat status and administrative function, India’s on a combination of population size, density, and non-agricultural workforce—each with its own inclusions and exclusions. Without adjusting for these definitional gaps, cross-country comparisons can mislead. What, then, should a careful user of China’s urbanization data do? One approach is to descend to finer spatial scales. County-level and township-level data often reveal the internal heterogeneity that prefecture-level aggregates hide. Satellite-derived measures, such as nighttime light intensity or built-up area classifications, provide an independent check on how “urban” a place really is. Researchers have also developed alternative urban extent datasets that map continuous settlement patterns rather than administrative borders. None of these solutions is perfect, but together they offer a more textured picture. Wang Xiansheng’s story is not an indictment of a statistical system; it is a reminder that maps and administrative lines are human artifacts, not natural facts. The prefecture-level city is a governance tool that has enabled China to manage a vast territory, but it also embeds a particular spatial logic into every dataset that flows from it. For anyone who uses those datasets to make investment or policy decisions, the first step is to ask not just what the number says, but what geography it is actually describing. The farmer on the hillside is, administratively speaking, a city resident. Acknowledging that paradox—without dismissing it as deception or error—is the beginning of spatial literacy.