Doug writes:
"The housing market is different from stock or bond markets. Turnover is relatively slow. The average holding period for a T-bond is something like 30 days. With heavily traded stocks, I'll bet the entire float turns over every few days (sorry, don't have stats at hand). Out of 108 million owner-occupied housing units in the U.S., about 1 million new and 5 million existing turn over in a year. Might that affect the "signature"?"
The short answer is "no" (with a caveat to follow later). The reason is that the time period doesn't matter. The hypothesis is that cooperative feedback will show itself no matter how long the period between interactions. Similar physics is used to predict changes in trend in everything from earthquake faults to failing aerospace materials to particle-physics experiments.
C Devine writes that "if the UK real estate market suffers from a crash, it's quite possible that this could stimulate a similar event elsewhere, including the US real estate market." In fact, the statistical work suggests that there is a historically unprecedented degree of correlation among world financial markets. The hypothesis suggests that as the *potential* for more participants to join in a market increases, the frequency of bubbles may also increase if new participants act as trend followers - in other words, if the bubbles themselves become attractors. I don't know how much that is a factor with real estate.
Now the caveat: larger markets, although they may have more "inertia" also have more potential participants. They may also be less subject to events which deform or undermine the growth of the feedback. Therefore Doug is exactly on track when he immediately suspects that the trade volume of the market could be an important factor. However, the hypothesis as I understand it holds that feedback loops should be able to overcome quite a bit of noise.
Also keep in mind that we are talking fundamentally about what should be observable in bubbles. None of this is deterministic.
peace,
boddi