U.S. Household Formation Is at a Record Low

The sharp drop in household formation largely explains why the housing glut remains stubbornly high, despite the plunge in housing starts in recent years. The gross vacancy rate—the proportion of housing units that are vacant—stood at 14.5% at the end of the second quarter of 2010, just below the 14.6% record high set in the first quarter of 2009. The rate has not budged, despite housing starts running under a 1.5 million rate since December 2006, and under a 1.0 million rate since June 2008.

During normal times, builders need to add about 1.5–1.7 million homes a year to meet underlying demand. Underlying demand consists of replacement demand, demand for second homes and miscellaneous factors (such as conversions from nonresidential to residential use), and increases in the number of households. Roughly, about 250,000 new homes are needed to replace homes destroyed by fires and natural disasters or that wear out from neglect or old age. Demand for second homes and miscellaneous factors account for another 50,000–100,000 new homes a year. Finally, the growing number of households normally requires another 1.3–1.4 million units.

Why is household formation down so much? One can cite several reasons, most related to the intensity of the jobs recession. The divorce rate and birth rate slipped in 2008 and 2009—as they usually do in recessions, according to the National Center for Health Statistics. Homelessness is also probably up. But the key reasons are a drop in immigration and an increase in "doubling up."

Hard data on immigration and "doubling up" do not exist. One can infer from the data, however, that a drop in immigration may have played a greater role than "doubling up" in the early stages of the jobs recession, since the number of households headed by foreign-born households under the age of 35 declined by 338,000 in 2009. One can also infer from the newly released data that "doubling up" played a greater role in 2009 than it did in 2008. For example, the number of households headed by those 15–24 years old fell by 124,000 (students moving back in with parents), while the number of households with six or more people in the home rose by 355,000, an 8% increase. The breakdown by age groups also suggests that "doubling up" increased in 2009. By 10-year age groups, the number of households headed by those in the 15–24, 25–34, and 35–44 age brackets all fell in 2009, while the number in all of the older 10-year age brackets increased. Job losses and foreclosures are concentrated in the younger age brackets.

By region, households increased in the Northeast, Midwest, and South, but fell in the West. By Census Divisions, households fell in the Pacific Division (Hawaii, Alaska, Washington, Oregon, and California), the Mountain Divisions (Montana, Idaho, Wyoming, Nevada, Utah, Colorado, Arizona, New Mexico), the East South Central Division (Kentucky, Tennessee, Mississippi, Alabama), and the West South Central Division (Oklahoma, Arkansas, Texas, Louisiana). The remaining five divisions had increases.

We should note that these estimates, which come from the 2010 Current Population Survey Annual Social and Economic Supplement (CPS ASEC), are subject to sampling errors. The top-line household number is a credible estimate, since it is based on a large sample. But many subcategories are less reliable because their sample sizes are too small.

A common misconception is that foreclosures account for the housing glut. The truth is more complicated. A foreclosure (2.5 million in 2009) or a bank taking possession of a home (about 1 million in 2009, according to Realtytrac.com) does not by itself add to the housing glut. If a household vacates a home and moves into a rental unit, the glut is unchanged. It increases, however, if the household moves in with another household ("doubling up"), or becomes homeless.

The decline in household formation suggests that once the economy gets back on track, and job growth resumes, housing markets will tighten as the formation rate increases, and housing starts will grow. But this will take time. We are not expecting housing starts to climb above 1.5 million until 2014.

by Patrick Newport