« Move on -- this isn't true here | Main | Play *Jeopardy* at MarginalRevolution »
The best two sentences I read this morning
In this paper, we present a simple model of housing bubbles that predicts that places with more elastic housing supply have fewer and shorter bubbles, with smaller price increases. However, the welfare consequences of bubbles may actually be higher in more elastic places because those places will overbuild more in response to a bubble.
That's from Ed Glaeser, Joseph Gyourko, and Albert Saiz. Here is the paper, ungated here, here is Mark Thoma's commentary.
Posted by Tyler Cowen on July 26, 2008 at 07:19 AM in Economics | Permalink
Comments
It's an interesting paper that largely agrees with some thoughts that I've had. One thing I'd distinguish in a more complicated model is between short-term and long-term elasticity of housing supply. Some states and locales have time-consuming or complicated regulation for housing permits but eventually do grant most starts. In those places, the short-term supply is fairly inelastic, because it takes some time for the regulations to be satisfied, but the long-term supply fairly elastic.
I'd argue that in some cases such places get the worst of both worlds. Bubbles last longer and have larger price increases because it takes a long time for supply to respond to demand. However, they will still manage to overbuild a great deal in response to a bubble, perhaps even more than in more elastic places.
In summary, I postulate that in such locales:
1) Prices begin to rise in a bubble.
2) In response, additional supply is planned to meet (bubble-related) demand.
3) However, the regulation process takes a long time, and in the meantime prices continue to rise.
4) Even more supply is planned and regulatory process started in response to additional price increases.
5) Enough additional supply is finally finished together with other factors to end the bubble, and prices start to decrease.
6) However, additional housing starts that have already gone through most of the regulatory process are still finished.
7) This drives down housing costs in response even further than in a locale with quicker regulation.
I'd argue that this model explains some places like, e.g., Northern Virginia. The amount of overbuilding certainly seems greater here than in North Carolina or Texas, where smaller price increases occurred.
Posted by: John Thacker at Jul 26, 2008 9:26:58 AM
If you *know* housing is overpriced, is there any effective way to sell it short?
Posted by: Cyrus at Jul 26, 2008 11:53:43 AM
"...places with more elastic housing supply have fewer and shorter bubbles, with smaller price increases."
Phoenix, Las Vegas, and the exurbs of Metro LA have led the nation in both price increases and, now, price falls. All three areas have been exceptionally elastic - housing stock increased dramatically. These may in fact be the most elastic major metro areas in the USA. And yet they have been - so far at least - the worst affected by the housing bubble.
Posted by: Mercutio.Mont at Jul 26, 2008 12:37:18 PM
I love it. And what variable would be most responsible for changes in elasticity in the real estate market? Could it be... money supply?
Posted by: Michael F. Martin at Jul 26, 2008 4:40:39 PM
From the paper: "Ideally, we would model the decision of existing owners to sell as an optimization problem, but this would greatly complicate the model."
Why not model the decision to sell as a Poisson distribution in frequency rather than modeling in an exogenous shock? Then the problem becomes a linear optimization in frequency of supply and demand.
Posted by: Michael F. Martin at Jul 26, 2008 4:51:00 PM
"While the utility from living in the area is time invariant, buyers do recognize that at each future date they will sell their homes with probabilityλ and then receive the sales price of the house. Individuals who bought at time t understand that with probability 1− e−λj they will have sold their house and left the city by time t+j . This forward-looking aspect of the model means that the current willingness-to-pay for housing will reflect expectations of future prices."
Isn't it simpler to just model individual utility as a function of time? I suppose they thought this would be too complicated. But it's easy enough if you start with utility as a function of frequency -- which is observable -- and then fourier transform back into time assuming (to begin with) random distribution in phase. Market cycles then synchronize phase. You get the same dynamics but without the aphysical assumptions about exogenous shocks. The phases are randomly distributed because at some moments in time price signals become suspect and phase distributions start to randomize.
Posted by: Michael F. Martin at Jul 26, 2008 4:58:39 PM
The measure of supply-side conditions they use is a measure of the amount of buildable land that depends solely upon geography. Notably, it does not include any information about regulatory barriers to construction. They do report that they also experimented with a measure of the strictness of the local regulatory environment. They report (in a footnote) that they obtained "qualitatively similar" results from those regressions. Since zoning may be endogenous while geography is fixed, they prefer using the latter.
Still, I'd like to see them report the results of omitted-variable tests to see if the zoning variable has any explanatory power beyond that already contributed by the geography-based measure. The correlation between the two is only 0.4, so multicolinearity is not likely to be a problem. It is, after all, easy to think of supply responses that may be ruled out by zoning but not by geography, such as replacing neighborhoods of older, widely-spaced houses with rows of townhouses and condos.
Posted by: Jeff Hallman at Jul 27, 2008 9:18:46 AM
The measure of supply-side conditions they use is a measure of the amount of buildable land that depends solely upon geography. Notably, it does not include any information about regulatory barriers to construction. They do report that they also experimented with a measure of the strictness of the local regulatory environment. They report (in a footnote) that they obtained "qualitatively similar" results from those regressions. Since zoning may be endogenous while geography is fixed, they prefer using the latter.
Still, I'd like to see them report the results of omitted-variable tests to see if the zoning variable has any explanatory power beyond that already contributed by the geography-based measure. The correlation between the two is only 0.4, so multicolinearity is not likely to be a problem. It is, after all, easy to think of supply responses that may be ruled out by zoning but not by geography, such as replacing neighborhoods of older, widely-spaced houses with rows of townhouses and condos.
Posted by: Jeff Hallman at Jul 27, 2008 9:19:28 AM
Geez, now why did that post twice?
Tyler shoud also have read a bit further into the paper. The quote
"the welfare consequences of bubbles may actually be higher in more elastic places"
applies to a model in which buyers' expect prices to go up forever. When that assumption is changed to buyers expect rising prices only until a period when prices don't go up, at which time they revert to construction costs, the duration of bubbles is reduced. The net welfare effect of elasticity is then ambiguous, as greater elasticity induces a more rapid supply response to a shorter bubble.
Posted by: Jeff Hallman at Jul 27, 2008 9:34:03 AM
Phoenix, Las Vegas, and the exurbs of Metro LA have led the nation in both price increases and, now, price falls. All three areas have been exceptionally elastic - housing stock increased dramatically. These may in fact be the most elastic major metro areas in the USA.
I would strongly disagree, Mercutio. While all three did have increase in housing stock, at least in the case of Phoenix and Las Vegas population increased yet faster than housing stock. This suggests that supply was not all that elastic compared to demand.
Compare with Dallas and Charlotte, where supply did increase at a rate commensurate with population; two cities whose markets are still seeing month-to-month increases and have barely been down off their peak, partially because they never went up much.
In any case, I still think that my point about short-term versus long-term elasticity is important as well. There are markets where new housing stock is built quickly, where new housing stock is not built, and where almost all housing proposals are built eventually, but the regulatory process takes a long time.
Posted by: John Thacker at Jul 30, 2008 12:16:03 PM
