David Leonhardt's excellent piece on house prices in the New York Times this morning asks a fundamental question about how to think about the future of house prices. If houses are a staple, they are currently overvalued by historical standards; if they are a luxury good, they are not.
Mr. Leonhardt's definition of a luxury good is one with an income elasticity equal to one (i.e., a good where the share of income spent on the good remains constant), whereas technically speaking, luxury goods have income elasticities greater than one (think nice vacations). But the point is still a good one--the income elasticity of demand for housing should tell us a lot about where house prices "should be" right now.
National house prices did follow income closely between 1970 and 2000, while according to Robert Shiller, they grew only by the rate of inflation before that. I wonder if the reason for the change is not that people wanted to spend a constant fraction of their income on housing, but rather that they wanted to spend a constant (or even increasing) fraction of their income on certain cities, such as Boston, New York, San Francisco and Los Angeles. These are all high amenity places, chock full of luxury goods, that have inelastic housing supply. It is possible that housing in, say, Wichita (sorry to the Kansans out there) is a staple, while Santa Barbara is a luxury.