You'd think it'd be relatively hard to mislay a city. Cities are, by definition, very large things: losing one sounds like it should be much more challenging than losing, say, your car keys, or even your entire car.
But in China, there are entire cities which go relatively unaccounted for. This is the result of a combination of fast development, low regulation, and the fact that many Chinese cities are still practically empty.
This is why it's pretty hard to get a figure for just how many "ghost cities" – cities that exist, but lack residents – there are in China. Most of these cities have some residents, making "ghost" a relative, and pretty hard to define, term.
So Baidu (basically, China's Google) decided to work it out. The researchers behind a new study, "Ghost Cities: Analysis Based on Positioning Data in China", used location data from users' phones, plus mapping and building location data, to find areas with high volume of buildings, but a low density of people. The researchers also tried to discount vacant areas that were empty because of tourism: apartments only filled in high tourism season, for example.
Here are some areas of vacant housing they found in nine different cities:
Overall, the researchers say they found over 50 ghost cities, though they only revealed around 20 in the report, so as not to adversely effect the real estate market in the rest. (No comment.)
As the researchers admit in their conclusion, this isn't a conclusive study: it relies, first and foremost, on the idea that "Baidu users" are a good proxy for "people", and that areas with no Baidu activity are empty. Yet as they note, arrogantly and probably accurately:
With the ubiquity of smart mobile phones, Baidu users occupy the most proportion of the whole population.
This does seem to be one of the more accurate surveys of China's ghost cities produced so far, even if the researchers won't release all the details. That could mean greater accountability for the development firms who toss up concrete blocks and then fail to fill them.