Running the Numbers on Housing

Lots of economists are debating the best government policy for dealing with the housing crisis. Glenn Hubbard thinks we ought to refinance people's loans and Brad Delong agrees. Arnold Kling disagrees because he sees a supply overhang (simply too many houses) as does Megan McArdle. I think it's time to look at some numbers.

My basic thesis is that the demand for houses is very inelastic. Basically each household needs a housing unit of some sort and not very many households need (or want*) more than one. So what I want to look at is how many new households have been created versus how many housing units we've built.

Between 2002 and 2007, the number of households in the US increased by 7.8 million (see table HH-1). During the same time period we completed 10.6 million housing units. For comparison, between 1995 and 2000 the number of households increased by 7.6 million while we completed 8.8 million housing units. The difference between the two periods is 0.2 million households but 1.8 million new units. That means we have about 1.6 million more new housing units that are looking for a new household. Eyeballing the completion data, that's about a year's worth of new housing units.

At about 1.5 million, 2007 was a more normal year for housing completions. So far for 2008, we're running about 25% behind 2007 (Table 5.) At that rate it will take four years to work through the excess housing inventory. If the completion rate slows further, we will be out of the housing slump that much sooner.

While there are probably other ways to consume the excess housing inventory I don't see how simply refinancing them causes that to happen.

* Suppose someone offered you a second house with a few minor conditions. You had to pay the taxes and the cost of maintaining it. In addition, you couldn't use it to house another household. Finally, you had to keep and maintain your existing home. The house is probably located in an outer suburb of a major growing city in California, Nevada, Florida or Arizona. (No it's not in a prime vacation spot.) What would you be willing to pay for the house?

Note: You may have noticed that I didn't use the 2001 data. The Census folks reset their baseline in 2001 so the household data appears to contain a discontinuity.