When I read the newspapers from the 1930s I’d occasionally see hopeful articles about how consumer spending would have to pick up soon, because of all the “pent-up” demand. If people hadn’t been buying cars for a while then presumably their cars would wear out, and this would trigger new demand for replacements. Of course I knew that there actually was no light at the end of the tunnel, which made these articles seem slightly pathetic—as if they were grasping for straws. They overlooked the fact that the depression made America much poorer, and that low consumer demand reflected that poverty. For similar reasons, there isn’t much “pent up demand” for cars in Somalia, despite low sales in recent years.
Sometimes I see this argument applied to the housing slump. Housing construction is down 70%, to levels far lower than at any time in post-war history (relative to population.) And this slump has been going on for a number of years. Surely we’ll soon need to build more houses, to meet our growing population. If only that were true. Unfortunately, as sharply as housing construction has fallen, household formation has fallen even faster.
Jim Glass sent me some very interesting data on household formation, which casts a very different light on the recent housing crash.
By my simple measure recent housing starts peaked with an inventory of 40% of an average year’s worth of starts above the trend line in 2007. That’s a cyclical high but a typical one. About the same or a little higher was reached in three other cycles since 1960.
But the plunge in starts since 2007 is unprecedented — by the end of 2010 cumulative starts were 72% of an average year’s worth of starts below trend. The next-lowest figure was 46% below trend back in 1970. If things were “normal” this would predict a huge boom in housing starts soon.
But housing starts are *following* household formation, which is plunging even faster, like an ICBM heading straight to its target.
In 2007 household formation was 1627k (average 1998-2007: 1499k) and housing starts were 1355k (average 1998-2007: 1716k). In 2010 household formation was all of 357k, down 78% from 2007 and down 76% from the prior ten year average. Housing starts were 587k, down 57% from 2007 and down 66% from the prior ten years. That’s a big fall, but it is still *well behind* the fall in household formation.
If I still had my blog I’d post the graphs — the line for household formation is heading straight down like to the bottom of the sea, it’s three times the fastest-deepest decline of the last 40 years. The line for housing starts looks like it is just striving to not fall too far behind.
I hate to be the bearer of bad news, but that light at the end of the tunnel is an onrushing train called falling household formation. It’s caused by three factors:
1. Less immigration due to the post-2006 crackdown.
2. Less immigration due to the severe recession and high unemployment
3. 20-somethings who can’t get jobs are living with their parents.
The problem is not that we built too many houses and need to work off the excess. Yes, we did, but we worked off that excess long ago. No the current problem is crashing demand for homes due to an unprecedented plunge in household formation. Call it a demographic depression. And the root cause? I know I’m going to sound like a broken record, but the biggest cause in low NGDP (although obviously other factors are also at work here–including immigration crackdown, minimum wage increase, extended unemployment insurance, etc.)
PS. I can’t wait for some smart alec commenter to write in and tell me the minimum wage increase can’t possible affect household formation, as no one can afford to live by themselves on the minimum wage. I already look forward to slapping you down. So go ahead and make my day.
PPS. This website shows that the Census had forecast household growth of about 1.1% per year. In this website you can find the data Jim used; only a 0.3% growth in household formation last year. And it’s possible the Census is a bit behind the curve, as for instance they didn’t catch up to the 1990s immigration surge until the 2000 census. Thus it might be even worse than Jim’s figures show.