Your correspondent is in Hangzhou, China. This city is not the largest by Chinese standards, but it is rich and influential. Alibaba is based here. In late February, local newspapers reported discounts by major property developers, news which quickly soured sentiment across Zhejiang province and then nationwide. If a major property bust plays out, Hangzhou is 'ground zero.'

Commentators have pounced (here, here, here and here) on the recent stall, and worried regulators are nudging banks to open their mortgage lines to first-time buyers. Yes, real estate has stumbled before (the last time was in late 2011) but since 2000 property has been almost a one-way ride. Now, however, developers are getting cautious:

We have reached a pause, perhaps even a juncture, in China's boom. Two years ago, urbanisation, a signature policy of Premier Li Keqiang, excited real estate developers into a land buying spree across China. Now their inventories are rising along with their financial gearing, just as buyer appetite rolls over. Rating agencies are getting nervous.

China is of course a huge and variegated market, but it's possible to discern a worrying divergence in fortunes. A few big metros are thriving while hundreds of smaller towns struggle in what one China strategist calls 'triumph of the megacities'. The gap has always been there, but it's getting visibly wider. Stories about 'ghost cities' have revived. These tales have an apocryphal quality, mixing amusement, amazement and horror. There used to be one or two, notably Ordos. Now the reports of woefully under-occupied 'new towns' grow: Caofeidian, Manzhouli, Changzhou, Shenfu, Yujiapu, Taizhou, Chenggong, and so on. From my personal observations, barren suburbs are out there, and have been for a while. But then, as they say, 'China is a big place.' How to put it all in perspective?

The investment bank CLSA recently conducted a survey of 609 projects nationwide. By physically visiting the sites by day and night, watching the lights go on, noting decorations and talking to guards, they calculated an estimate of 15% vacancy rate in apartments built since 2000. That implies about 10 million empty units nationally, considerably less than feared (one report had suggested implausibly that 64 million homes lay empty). Still, 15% is high, more than Las Vegas and considerably more than the US at 10% overall.

Adding construction already underway, CLSA say China's vacancy rate will hit 20% and could go even higher unless developers immediately curb new starts. Vacancy is heavier in small cities, already nudging 20%. The ghost cities are altogether worse.

One valid reason for vacancies is to satisfy 'upgrading' demand, as when a family buys a new unit but takes time to move in. This often takes many years in China. Revealingly, such affluent families aren't swapping one home for a better one; they're holding both. Chinese want to be multi-home owners, or even better, landlords. That motive is understandable enough, but obviously it cannot work en masse. Urban home ownership already is 90%.

That's the challenge of China's urbanisation. Those who can afford to own homes already do so (or even more than one). But poorer new migrants below the bottom of the housing ladder look up at an impossible financial hurdle. True, smaller cities are roughly twice as affordable as large cities, but mega-metros attract young Chinese with bright lights and better jobs, so smaller cities are actually seeing their populations contract. It wasn't meant to be this way.

Chinese authorities obviously can't repeat the experience of, say, Japan and Korea, where a third of the entire nation crowds into its capital. Instead of creating congested, unmanageable and potentially unstable super-conurbations, planners prefer a more continental dispersion of the populace, like in America. Still, the forces for concentration are powerful. High-speed rail links should foster regional economies, but may have the opposite effect by encouraging migration and commuting to larger metros. Perhaps 'cluster cities' will be the optimal solution.

Why does this matter? And isn't ongoing Chinese urbanisation anyway meant to one of the two great realities of the century? Probably, but with 4 billion square meters of housing currently under way (that's about 40 million apartments) and starting more than 1 billion m2 annually, Chinese residential developers may be building too much, and mostly in the wrong places – even assuming continuing urbanisation. If 20 million farmers migrate every year, they'll need 30m2 each; that's 600 million m2 of floorspace.

There are of course other sources of demand, like upgrading and redevelopment, but obviously urbanisation won't grow forever. Indeed, officials are saying demand may be peaking, and that's what the market today worries about. CLSA, which is not known as especially bearish, thinks small cities may need to collapse construction by a shocking 60% over the next few years. These cities consume 70% of materials used in all residential construction, which accounts for half of all steel use. Do the maths: that's a 20% impact on steel demand.