Recall that most forecasters are counting on a housing surge to be one of the major drivers of economic growth this year, at a time when government is cutting back. If that optimism proves to be misplaced, this could turn out to be a gloomy year indeed.
So should we worry? The answer is: Not yet. And the reasons why say a lot about how to read--and, more importantly, to avoid misreading--economic data.
It is true that the slide to an annual pace of 890,000 housing units is a big step back from December. It’s also true that it is the second negative housing indicator in a row, as Business Insider points out, following a surprisingly weak survey of homebuilder confidence. But the more you look at the details of the new Census report, the more you see evidence of a sector that is holding up just fine in the new year.
First, housing starts numbers are often volatile in the winter months. A major blizzard over a large part of the country can easily stop home construction in its tracks, and conversely an unusually mild winter can mean a boom in housing construction that normally would have waited until spring instead beginning in January or February. So it can be helpful to look at averages over two or three months rather than any single month.
And by that measure, things look better. December’s housing starts number included an unusually big rise—a revised 15.7 percent bump, even stronger than the 12.1 percent gain first reported. Average December and January, and builders started on new home construction at a 931,500 annual rate, which easily surpasses the 865,000 monthly average of October-November.
If that pace of gains continues, it would be a blockbuster year for homebuilding.
And if you look at another key indicator in Wednesday’s Census release, the case for a solid—and maybe exceptional—year for housing looks even stronger. The number of permits issued for new home construction actually rose in January by 1.8 percent to a 925,000 annual rate. That was stronger than the 1 percent gain analysts had forecast.
For reasons I don’t fully understand—convention seems to be part of it—media and Wall Street analysts tend to focus more closely on the housing starts number than on housing permits. And it is true that when home construction actually begins tells you more about the near-term performance of the economy. But in many ways permits are a more reliable indicator.
First, while weather often distorts housing starts, the distortions tend to be less severe for permits. If it is an unusually cold weather, that might mean builders delay construction starts—but that doesn’t stop them from going down to the county permitting office to get their paperwork lined up for future building.
But the bigger reason to put more faith in January’s decent permits numbers than in the steep drop in starts comes down to sampling. The Census gets reports from local governments each month about how many permits they have issued for new construction that tend to be quite accurate. That’s why the Census says with 90 percent confidence that the housing permits number, 1.8 percent, truly lies somewhere within 0.9 percent of that growth rate. It is almost certain, in other words that the number of permits issued was up a little bit in January.
But the data on how many homes construction companies actually began work on is much shakier, depending on surveys that have plenty of error in them. That means that the confidence interval is much wider. The 8.5 percent drop is, within that 90 percent confidence interval, plus or minus 11.3 percent. Think about that for a second—this huge drop could have possibly been a truly massive drop of nearly 20 percent. Or it could have actually been a slight increase of around 3 percent. The survey is unreliable enough that we don’t really know for sure.
So there is no reason to radically transform one’s expectations for housing based on the January numbers—and if anything, every reason to think that a steady recovery continues apace. The report also offers a nice lesson in the limits of economic data, especially when digested into headline form.