Let us first stipulate that houses themselves are unique entities, but similar enough that we can consider them comparable if not fungible. (In this, they are like any financial asset.) Let us further restrict them in acknowledging that there are significant transaction costs: one’s liquidity preference should not prefer housing to, say, cash-equivalents or marketable commodities, but they are arguably more liquid than a piece of sports memorabilia or antiquarian books or the like. (Given leisure and choice, one sells a copy of Action Comics #1 before one sells a residence; given a need for liquidity and reduced price uncertainty, this preference may change. The implications of this are left as an exercise.)
Assume a standard lifecycle model with intergenerational wealth transfer; we should perhaps stipulate credit constraints in the first (accrual) period, but that assumption can be relaxed during the second (consumption/transfer) period.
As a complicating factor—or rather, to reflect reality—stipulate that in areas that lack significant emigration, transfer of housing assets commonly is done as a continuous flow (that is, the final stages of transfer are from “empty nest’ to Gen2 living with and supporting Gen1 to Gen2 possessing the domicile as the overlap ceases to be), while areas with notable emigration patterns tend to monetize the housing asset and transfer the cash-and-marketable-securities equivalent to the next Generation.
As a further complicating factor to the above, lands that tend to lack emigration may become areas from which people immigrate. Similarly, some areas may become loci of emigration as employment patterns shift and true Structural Change promulgates through a system.
In such a model, the housing component serves the dual purpose of providing housing and being a vehicle for savings. Not a good vehicle for savings, under normal circumstances, but a piece of equity that can be passed to the next generation—either as a lump sum cash equivalent or as a domicile, depending on the externalities discussed in the Action Comics #1 parenthetic.)
This effectively transforms all housing purchases into rentals—rent for the generation, passing equity accrued to the next in whichever form is optimal.
From that point, it seems intuitive that if the incentive structure changes significantly—if, for instance, one can realize a substantially greater return from owning and selling than from renting—that a bubble might develop. And as more people are incented to do what they ordinarily would not—to sell a house, to “trade up,” or to otherwise find ways to realize the gains that come from a relative mispricing in the market—that what might become a bubble either (1) will return toward its original trajectory, (2) will revert toward its original trajectory (though it may find a different [higher or lower] equilibrium point, or (3) grow until it is a clear bubble, that will burst.
Note that we can see graphically the difference between the non-bubble states ( and ) and the bubble state—the first two would show a sign of mean reversion (a decline), while the Bubble State continues to rise. I wonder if there might be any evidence for one or the other of those hypotheses?
Maybe someone at the Federal Reserve Board of Boston should throw the above into some standard life-cycle equations and re-evaluate the presentations at this conference, including Glaeser, Gottleib, and Gyourko (PDF)?