Is the U.S. Housing Market Turning Around?

Two additional details are making these numbers even sweeter. The American Recovery and Reinvestment Act of 2009 allows first-time homebuyers (someone who has not owned a home within the last three years) to take a tax credit worth up to $8,000 on a home purchase. And, the 30-year fixed mortgage rate that Freddie Mac tracks dropped to an all-time low of 4.78% in the latest survey week (data start in 1971).

Sweeter still, more good news is in the pipeline. February's 16% jump in single-family permits points to a possible double-digit gain in single-family starts during March. And the NAHB/Wells Fargo Housing Market Index, which measures builder sentiment, is likely to climb in reaction to the latest string of good news.

So, has housing hit bottom? Perhaps—and if it has, February 2009 will be the month the market started to recover. But we suggest waiting at least a couple of months before making a call, because of two impediments that make many of these recent numbers hard to decipher: weather and statistical significance.

February 2009 was the eighth-driest and 27th-warmest February in the 1895–2009 record, according to the National Climatic Data Center. December and January, on the other hand, were colder than normal in the Northeast and Midwest, and February was unseasonably warmer in the Northeast, and about normal in the Midwest. Swings in the weather, especially in the winter, will swing the housing numbers, and the seasonal adjustment factors will augment these swings. This winter's weather patterns probably moved the housing starts more than the sales numbers, as houses that builders planned to start in December, January, and March were instead started in February. More than likely, however, weather distorted the new and existing home sales numbers as well.

Not only did weather distort the numbers, but many of February's estimates failed the test of "statistical significance." The Census Bureau's estimates are based on samples that in many cases are too small to produce a statistically significant estimate. For example, February's gains in single-family housing starts and new home sales fail the significance test. The 4.7% estimated increase in new home sales is particularly egregious, since the confidence interval, plus or minus 18.3%, is many times larger than the monthly increase. January's 1.8% increase in the FHFA house price index should also be discounted because, as the press release noted, sales volume was low, making the estimate imprecise.

The housing starts and new home sales press releases contain explanatory notes warning users that month-to-month changes may be "irregular," and that "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." For new home sales, the waiting period is five months. In other words, a turning point cannot be established until months after the fact.

Still, one key statistic pointing to "better times ahead" bears watching. February's 16.1% jump in single-family housing permits was broad based, with most states improving over January levels. Single-family permits are good barometers for monitoring housing because they are not influenced by weather as much as other housing numbers, are well measured, and are not subject to large data revisions. In addition, they are leading indicators of housing activity, and are available at the state and metropolitan level at a monthly frequency (although the state and local numbers are not seasonally adjusted). While most of the housing numbers that have come out recently are hard to decipher, February's increase in single-family permits is a solid number that may eventually mark the turnaround in the worst housing market in the nation's history. But we need two more months of additional data to make this call.