Housing Supply and Demand
Of course, construction businesses do not care about surplus/shortage per se: they want to make money. The way that the market ensures that housing construction will continue is by having a housing price that stays at or above the cost of construction. Construction businesses respond by making more houses.
Note I do NOT claim that real housing prices will necessarily rise (more details on "The anatomy of a housing recovery"). They could rise -- perhaps significantly for a while (that depends on the reasons for the housing cycle in the first place, on which I'll write more later) -- but ultimately they'll be about where they are now. Moreover, I am confident that NEITHER real housing prices nor the pace of housing construction will return to their previous highs any time in the next five years.
Simply put, my definition of a "housing recovery" is not a huge increase in real housing prices (more details on "The anatomy of a housing recovery"). Of course, (absent a deflation) my definition of a housing recovery does mean that nominal housing prices will trend up.
Where I differ from some Gurus
If real housing prices fall significantly below their lows of this spring, or the pace of housing construction fails to pick up, then we have evidence I have missed something critical. Maybe demand turns out to be stagnant for a number of years. Maybe banks stop making loans. But I am skeptical that these forces will be powerful enough to overwhelm the basic supply and demand forces I have outlined.
In contrast, Professors Shiller and Krugman have been saying that nominal housing prices are likely to get significantly lower. Absent a deflation, it follows that they think real houses prices will fall even more than that.
Composition Bias and Possible Downward Wiggles
Someone emailed me that housing prices indices are plagued by "composition bias": the indices move up and down because the types of houses included in them change, and not because the price of any one of those houses actually moved in the direction of the index. Specifically, he suggested that housing prices might not have increased in the last month or two.
I agree that composition bias needs to be considered, and admire Professors Case and Shiller for the hard work they've done on this measurement problem (and appreciate the email -- I would like to thank the author by name but he asked to be anonymous). It is likely that housing price indices will show some downward wiggles merely due to composition bias.
However, if today we are at the point that composition biases, rather than a CRASH (a genuine and significant surplus of houses), are driving the price measures, then I rest my case: the crash is over, or at least so diminished that we cannot distinguish it from technical issues that may be boring, but sometimes demand attention from those of us doing quantitative work.