Elementary economics tells you that in a competitive environment, the price of a new house should equal:
the price of land + construction costs + a reasonable profit for the developer
But in most cities, that sum is not even close to what buyers are paying.
Take Dallas, for example. If you live in central Dallas, and if you could magically add a quarter of an acre to your lot size, you'd add (on average) about $2,200 to the value of your house. (We know this from comparisons of similar houses on different-sized lots.) Do the same in central Philadelphia, and your house value increases by $8,400; in central Houston, it's more like $17,600. In that sense, central Dallas land is just about the cheapest urban land you can find in this country. Among large cities, only Atlanta, Boston, and St. Louis rank lower. In theory, that should be great news for Dallas housing prices. But it's not. A house that costs $100,000 to build typically sells for $140,000 in Dallas, maybe $120,000 in Houston, and under $90,000 in Philadelphia.
Aha! say the commentators. Housing prices must be driven by something other than fundamentals. Speculators, of either the rational or the irrational variety, are the obvious culprits.
Here's what's wrong with that analysis: Housing prices have to make sense on both the demand side and the supply side. No matter what you do or don't believe about the ability of crazed demanders to bid up prices, you still have to explain why competitive suppliers don't bid those prices right back down. In other words, if the housing market is so tight that builders are making a fortune, they ought to be flooding the market with new houses—and driving down prices.
In fact, buyers' behavior is relatively easy to explain. Most of the recent explosion in housing prices has been in cities like San Francisco and Santa Barbara—in other words, in really nice places to live. It's not unreasonable to believe that, as Americans grow richer, and as technology makes us more mobile, more and more of us want to move to California. And it's not unreasonable to expect that this trend will continue, so that even a very expensive house in the Bay Area can look like a good investment.
The great mystery is on the supply side. Instead of the traditional formula "housing price equals land price + construction costs + reasonable profit," we seem to be seeing something more like "housing price equals land price + constructions costs plus reasonable profit + mystery component." And, most interestingly, the mystery component varies a lot from city to city.
Even in cities like San Francisco, where there's little room to build and land is consequently dear (on the order of $85,000 per quarter acre, compared with $2,200 for Dallas), you can't use land prices to explain away housing prices. The mystery component in San Francisco housing—that is, the amount left over when you subtract land prices and construction costs from house prices—is the highest in the country.
Edward Glaeser of Harvard and Joe Gyourko of the University of Pennsylvania have computed these mystery components for about two dozen American cities. They speculate that the mystery component is essentially a "zoning tax." That is, zoning and other restrictions put a brake on competitive forces and keep housing prices up. (Read one of their papers here.)