The housing boom (cont.)

Matt Yglesias responded, sort of, to my comments (“Was there a housing boom? Yes.”) from yesterday, countering his curious assertion that there was no building boom in the mid-2000s, by conceding that there was a boom in construction employment after all. But he refuses to give up on the argument that there was no building boom. A few more words on this topic before laying it to rest.

Yglesias graphs construction employment as a percentage of the civilian labor force with a line marking the average:

Before proceeding, a little overview to the employment stats. (No doubt CAP has a generous research budget, so if they want to send along a tuition check on their fellow’s behalf, that would be dandy.) Yglesias seems not to know that there are two surveys behind the monthly employment figures—one of about 60,000 households and one of about 300,000 employers (known as the establishment or payroll survey). He’s dividing construction employment from the establishment survey by the civilian labor force from the household survey. That’s unusual. The reason that the two surveys are rarely combined this way is that the concepts underlying the two are significantly different. The definition of employment in the household survey includes agricultural workers and the self-employed, who are not counted in the payroll survey.

Even odder is the use of the labor force as the denominator, since it includes the unemployed as well as the employed. (The labor force consists of those who are working or actively looking for work. The first set are the employed; the second, the officially unemployed. The unemployment rate is the number of unemployed divided by the labor force.) It would make more sense to use total employment as the denominator, but not the household version. My analysis yesterday took construction employment as a share of total employment from the establishment survey, which is the right way to go about this.

All that geekiness aside, Yglesias’ graph does show a marked elevation around 2005–2006. When I reproduce his numbers, I see construction employment as a share of the civilian labor force peaking at 5.4% in the spring of 2006, which is more than 20% above its long-term average. It also stayed above that average for 138 consecutive months, easily eclipsing the previous record of 75 consecutive months in the late 1960s/early 1970s. (And it was a lot further above average in the recent period than it was 40 years earlier, too.) His commentary doesn’t acknowledge the extent of the employment boom, but at least he walks back his earlier assertion, as they say in DC.

But he just won’t give up on the claim that there was no building boom, and grasps instead at remodeling. To reprise a couple of points from yesterday’s post: 1) As a share of GDP, residential investment—that is, the building of new houses and work done on older ones—hit a peak of almost 60% above its long-term trend in the mid-00s. And, 2) between 2001 and 2006, residential investment accounted for 12% of GDP growth, twice its share of the economy. If “60%” and “twice” don’t sound like big numbers, then I don’t know what does.

The Bureau of Labor Statistics also reports finer detail on construction employment (from the establishment survey), though data for most of the subcategories begins rather recently, making long-term comparisons impossible. Still, residential construction as a share of total employment peaked at 24% above its long-term average in 2006 (it’s now 30% below). Single-family contractors peaked at almost 30% above average; it’s now almost 40% below. And residential remodeling peaked at about 25% it average; it’s still slightly above that average now. A lot of the boom in mortgage debt mid-decade came from people borrowing against their inflated home equity values and using the proceeds to spiff up the kitchen or add a wing to the house. But clearly there was a lot of new building going on—much of it McMansions and other bloated structures, whose size is missed in a simple tally of starts.

Finally, the four major housing bubble states—Arizona, California, Florida, and Nevada—saw their total employment rise by 9% from the 2003 trough to the peak in 2007; the other 46 states gained just 6%. And the Bubble Four got hammered in the recession, with employment falling by almost 10%, compared with just over 5% for the other 46. The Bubble Four contributed 35% of the employment loss during the recession, nearly twice their 19% share of employment going into the downturn. The housing bubble had a lot of spillover effects, notably the use of home equity lines of credit to pay for everything from wide-screen TVs to dental work. But the direct effects of the building boom were considerable—as are those of the building bust.

5 Comments on “The housing boom (cont.)

  1. Re: original yglesias
    I fail to understand the relevance of the rate of 70s housing starts. And the labor numbers : Drywall gun, anyone? Is productivity a constant for all civilian employment?

    An ideal comparison across the boom: population; existing housing (you know, those houses they built in the 1970s, like); and new. And an idea about the vacancy rate, in both houses and rental markets.

    If there was a boom, was the oversupply commercial in nature? Speculative/private or rental? How large was the second home market?

    The original question, posed at the national rather than regional level, ultimately depends on one’s idea of social utility.

    In truth, I lean towards Yglesias (price not supply), because of depopulation and general uneven development in the US. But he fails to prove anything about his assertion!

    PS. USPS datasets on domestic addresses might be of use…See below

  2. Why is this even a debate ?

    Housing investment peaked in Q3 of 2006 (I believe) ; it’s collapse foretold the financial crisis and the recession in a very Marxian way – investment collapsed because the potential profits had been wrung out the system.

    What’s more interesting is the notion (espoused by Calculated Risk, among others) that there is now pent up demand for housing based upon a reversion to mean analysis. Except there really is no reason to expect the U.S. to go back to ‘mean’. Huge household units are the norm in the developing world after all, and this is the model the U.S. elite is planning for us.

  3. Doug the Mythbuster! Marx was also a bit of geek when it came to discussing the mismanagement of RRs because of shoddy rail quality and misunderstood rates of scrappage. Matt Y shouldn’t be so careless in his calculations, especially about employment. My take is that even relatively small changes in higher paying sectors have a big impact – especially on regional and local economies. For example, even though the insurance sector (SIC 63) is about 1% of US employment, average salaries are typically “middle class” and provide stable employment. Automation and relocation had a devastating impact in the decline of this sector in NYC. Today, the sector is a shadow of its 1970s peak with a smattering of really high paying jobs such as commercial brokers, specialty underwriters, and actuaries and some low wage jobs like a few clerks and receptionists. Nevertheless, the industry is both fairly capital intensive – in both the physical and financial sense – with OK labor productivity and very good profitability. Even road construction (vs. residential) is fairly capital intensive so that “shovel ready” projects just don’t employ a lot of “shovel weilding” human beings. Part of the reason why the rate of labor demand in many traditional “labor intensive” sectors is because of the widespread diffusion of labor-displacing technologcial changes (see the “CouponSherpa” iPhone app). While we love these innovations because they lower private prices, they also have a “social” price that doesn’t show up in any ledger.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: