Beware the household formation headlines! The current data—1.3 million households formed in 4Q14—likely overstates actual household growth. Since 2007, 6 out of 8 of the gains in the fourth quarter were given back in the subsequent first quarter by negative household formation (see graph below). Also, no other data point we found in Q4 supported household growth that strongly.
On average since 2007, roughly 76% of the Q4 gains were reversed in the following quarter. If you believe the trend will continue, the real 1Q15 number may be more like -1.0 million. While this would be extreme (as was the Q4 number), we do believe some of the overstating in 4Q14’s data will be reversed in 1Q15. In the 41-year period from 1965 through 2006, household formations only went negative in the first quarter five times, so this seasonal shift appears to be new, and the Census Bureau has not been able to explain it to us. First quarter data is available April 28.
We advise using caution regarding all data, especially the housing vacancy numbers, exactly as the Census Bureau reports. For housing vacancy, we use a rolling four-quarter average, as it smooths out the end-of-year jumps that likely occur due to second-home double-counting of snowbirds. Do not criticize the Census Bureau, however, as estimating housing vacancy is quite difficult, especially on a constrained budget. Also, the goal of the Housing Vacancy Survey is to estimate rental and homeowner vacancy rates, as well as homeownership rate, and not to estimate household formation.
In the following rolling four-quarter average chart, you see more steady growth (and also adjustments we made to be more in alignment with the Decennial Census). Households have been forming at a lower than usual pace since the Great Recession began.