I stumbled across a bunch of charts the other day and this one in particular caught my eye: http://www.businessinsider.com/chinese-land-prices-2010-7#likewise-price-increases-are-killing-income-increases-14 (Credit goes to Jing Wu, Joseph Gyouko and Yongheng Deng at NBER)
Just to give you some idea of what a price to income ratio of over 20.0x in Shenzhen means, here in the U.S. California had a price to income ratio of right about 10.0x before prices started plummeting to about half of their all time high. Other markets still had decent sized corrections with price to income ratios of only 4-5x. Generally, much north of 4x is considered to be on the edge of pushing it nationally here while some markets can sustain, over time, more like 5-6x. I've seen some countries that can apparently sustain near 7x, but even for those countries that seems to be the point at which prices stagnate and wait for incomes to catch up.
Returning to China, it is true that there is regional variation, but that has been true in every real estate bubble in history. Here LA, New York, Miami, Las Vegas, Phoenix, Milwaukee, Cleveland, and Detroit all had very different dynamics. In the U.K., London, Liverpool, Manchester, Brighton, and Newcastle had varying degrees of excess. The same applied in Spain in the 2000s and Japan in the 1980s. It's a very non-compelling argument to say the least. What's more is that price to income levels mask another trait of real estate excess which is that those on the bottom end of the income spectrum often get caught up in the frenzy and buy much more house than they can afford in otherwise not particularly overpriced markets such as Atlanta.
It is unclear how much further this has to go, but perhaps Alex can expand with some first hand details on what he saw there.