It’s been noted ‘Big-Data’ is like teenage sex; everyone talks about it, nobody really knows how it’s done, everyone thinks everyone else is at it, so everyone claims they’re doing it.
The paper highlighted this week, from Guanghua Chi (et al) from the Big Data Lab, Baidu Research, Baidu Inc. seems though to have harnessed and put big-data to a particularly interesting use.
There’s been no reliable study to date, the authors claim (and I agree), that gets to the bottom of China’s ghost city phenomena. Sell-side analysts bus clients to more obvious examples, cynics refer to urban legend stories of static electricity meters (tosh rooted in a 2010 report from the normally reliable folk at Caixin, Static Electricity Meters), agents talk of city-by-city overhang based on sales versus inventory data (a wobbly denominator atop a dynamic numerator) and amateurs count dark holes at night.
The study in the paper you can access in full now at Ghost Cities uses data from Baidu to see where people actually are . In the process the difference between ‘real’ cities and ‘tourist’ developments is contrasted with a study of Rushan (a mostly beachfront development in Shandong) and Kangbashi (to where the government of Ordos moved in 2006), both well publicized ghost sites.
The paper has some nice graphics but Fig. 2 is the most useful and highlights the top-20 overbuilt cities in China; few readers will be familiar with more than a small handful of these towns and the main point is thereby made convincingly. The ghost city phenomena in China is almost exclusively a second, third and fourth tier city problem.
This conclusion chimes with my own experience and evidence from a number of non-partisan reports. That there is over-build in the residential property market in China has never been a contentious point; there is. However, where it’s located and why it therefore doesn’t (much) matter is the more important issue; this paper makes a valuable contribution to that analysis.
Happy Sunday