How many excess houses are there?

"Sometimes, indeed, the reduction of stocks [of capital goods] may have to be virtually completed before any measurable degree of recovery can be detected. "

- John Maynard Keynes, The General Theory of Employment, Interest, and Money, Chapter 22.

One of John Maynard Keynes theories was that recessions occur when producers cut back on the production of durable goods because there is a real or perceived oversupply of durable goods.  See Chapter 22 "Notes on the Trade Cycle" of the General Theory of Employment, Interest, and Money.   I see this effect at work in the current downturn in the US housing market.  Housing starts have plummeted as the inventory of houses for sale swelled and prices declined (a sure sign of oversupply), and the resulting decline in construction spending has contributed to the recession.  Not to mention that declining home values related to the oversupply of housing have helped trigger a financial and credit crises.

What I think is fascinating about Keynes's theory is that he postulates that before a recovery can begin the excess stocks of durable goods must be eliminated one way or the other.  That got me wondering: just how many excess houses are there in the US right now and how long will it take for population growth to soak up this oversupply?

To explore this I put together some rough calculations on this Google Documents spreadsheet.  All these numbers were pulled from Census bureau data at the links cited below.

The calculations on the spreadsheet seem to indicate that there were more housing units built in the period 2004 to 2008 than could be absorbed by population growth.  From 2000 to 2004 the ratio between population growth and housing stock growth was 4.19 (4.19 new people for every new house).  In the period 2004 to 2008 that ratio was 1.29. Given the average people per house over the years (see spreadsheet) it looks like more housing got build in 2004-2008 than could be justified by population growth.

But how to measure how much excess supply there is in the housing stock?  The absolute number of vacant houses is not very useful because it doesn't take into account changes in population.  So the first thing I did was calculate what population increase it would take to return to the year 2000 ratio of people per housing units, and that number turned out to be 4.5 million more people to get us back down to the people per housing unit ratio of the year 2000.  Since our average annual population growth rate is 2.7 million it would take about a year and a half for population growth to consume the excess housing stock assuming that all new housing construction stops immediately.

I also calculated the number of excess housing units assuming the year 2000 ratio for people per housing unit and got 1.9 million extra housing units.

Of course these are really rough calculations.  I haven't gone back to look at how stable the ratio of people per housing unit is over time, and presumably preferences change over time for how many people live together in one house. However, even if there was a trend towards smaller households on average earlier this decade, it seems unlikely that that trend would continue during the recession.  There are a lot of reasons for people to move in together when the economy is shrinking.  About the only thing that would make smaller households attractive at this point would be truly spectacularly low prices on housing.

To me these calculations indicate that:
  • Housing prices will probably continue to fall until either:
    • (1) population growth has absorbed the excess housing stock (another 1.5 years at least?) or,
    • (2) housing gets so cheap that people prefer even less crowded living arrangements than they preferred in the year 2000.  Intuitively, it would seem that that price point would have to be significantly lower than the housing prices in 2000, especially since the future was looking bright in 2000 and its looking dark right now.  The Case-Shiller home price index was at 107 in June 2000, and it was at 154 in Dec 2008.  107/154 = 1.47, i.e. maybe another 50%+ drop in prices to absorb the excess supply without waiting for population growth?
  • Any new home building this year and next year will prolong the period of falling house prices.
  • To instantly stop the housing market decline we need to either bulldoze about 2,000,000 housing units, or get an extra 4,500,000 people to immigrate to the US.
One final comment.  I find it really strange that I have not come across any discussion of this issue in the media.  Prices are all about supply and demand, yet when people talk about the prospects for the housing market they never seem to talk about the real supply of housing as it relates to population.  To me that is the crucial issue. It seems to me that as long as there is more housing than our current population needs to be sheltered there are going to be strong downward pressures on housing prices.


Census Bureau: Rental and Homeowner Vacancy Rates for the United States: 1960 and 1965 to 2008

Estimates of the Total Housing Inventory for the United States: Fourth Quarter 2007 and 2008

Housing Vacancies and Homeownership historical tables.

No one home: 1 in 9 housing units vacant - USA Today 2009-02-12

Housing bubble peaked in early 2005.