A few notes on the jump in housing starts

  • Share
  • Read Later

This morning the Department of Commerce reported (PDF) that housing starts jumped 22% in February to a seasonally adjusted annual rate of 583,000. January’s figure, 477,000, had been a record low. Some commentators are taking the uptick as a sign that the housing market may be bottoming, though other MSM regulars—a couple of whom are my bosses—are asking if this isn’t actually bad news. With so many existing homes sitting around unsold, do we really need to be building that many more?

A good question. But before we go too far down that road, I would point out that housing starts data are notoriously volatile. Another, steadier measure of new-home activity—building permits—is also up, though much less fantastically. Permits rose 3% in February, and for single-family homes they were up 11%. Overall, permits are still down 44% from where they were a year ago.

So what’s going on with starts? A big part of the jump came from condos, apartments and townhomes. And a fair amount of that activity flowed from warmer-than-expected weather. If that leads you to think maybe we should give the data another month or two before we start drawing trend lines, I’d be likely to agree with you. As one analyst cautioned, we might be looking at a “weather-related fluke.”

It could also be a data-related fluke. Journalists almost never report on statistics precisely, which I’m sure drives all manners of social scientists batty. Here’s the actual wording from the press release: “Privately-owned housing starts in February were at a seasonally adjusted annual rate of 583,000. This is 22.2 percent (±13.8%) above the revised January estimate of 477,000, but is 47.3 percent (±5.3%) below the revised February 2008 rate of 1,107,000.”

In other words, housing starts could be up 36%, or they could be up 8.4%. It’s a range. And, technically, we’re only 90% sure the real figure is in that range.

Which is probably why there’s also this line: “In interpreting changes in the statistics in this release, note that month-to-month changes in seasonally adjusted statistics often show movements which may be irregular.  It may take 3 months to establish an underlying trend for building permit authorizations, 4 months for total starts, and 5 months for total completions.”