Friday, November 17, 2017

Lawler: Has US Household Growth Slowed, and If So, Why?

by Calculated Risk on 11/17/2017 01:31:00 PM

From housing economist Tom Lawler: Has US Household Growth Slowed, and If So, Why?

Yesterday the Census Bureau released estimates of America’s Families and Living Arrangements (which includes household estimates) based on the Annual Social and Economic Supplement of the CPS for March 2017, and the estimates suggested that US household growth slowed considerably in the 12-month period ending this March – though by less than that shown in the Census Bureau’s tables.

According to the data released yesterday, the CPS/ASEC-based estimate of the number of US households in March 2017 was 126.224 million, up just 405,000 from the March 2016 household estimate of 125.819 million from the 2016 CPS/ASEC. As I noted in an earlier report, however, this meager yearly gain is understated, because the 2016 CPS/ASEC estimates were based on “2015 Vintage” population estimates for 2016, and last year Census (in its “2016 Vintage” release) revised down considerably its population estimates for 2016 (and for earlier years.) Annoyingly, Census does not go back and revise earlier year CPS/ASEC household estimates to reflect revisions in population estimates.

For example, CPS/ASEC tables show an increase in the US adult (18+) civilian non-institutionalized population of just 0.59%, from March 2016 to March 2017, while Census’ updated population estimates shown an increase of 0.89% over this period (again, reflecting a downward revision to 2016 estimates).

If one were to adjust the 2016 CPS/ASEC household estimate to reflect the downward revisions in 2016 population estimates, the “adjusted” CPS/ASEC household estimate would be about 125.445 million, and the “adjusted” increase in the CPS/ASEC household estimate for 2017 would be about 779,000. This “adjusted” gain still represents a marked slowed in estimated household growth.

The CPS/ASEC-based slowdown in household growth for 2017 (an “adjusted” increase of 0.62%, compared to an “adjusted” increase in the adult population of 0.89%) reflects, by arithmetic, an increase in the average household size and a decrease in the aggregate adult “headship” rate. The lower headship rates for 2017 compared to 2016 were not driven by a gain in the number of young adults living with parents (which decreased slightly from last year’s elevated levels), but instead by a decline in the number of people living alone (mainly in the 35+ group).

The other major CPS household estimate, based on the Housing Vacancy Survey (HVS) supplement to the CPS, has also showed a marked deceleration in growth. The CPS/HVS estimate of the number of US households in the third quarter of 2017 was 119.085 million, up only 407,000 from the estimate for the third quarter of 2016. (Reminder: the CPS/ASEC household estimate is “controlled” to independent population estimates, while the CPS/HVS household estimate is “controlled” to independent housing stock estimates.)

One other household estimate produced by Census comes from the American Community Survey, though only annual estimates are available. For 2016 the ACS-based US household estimate was 118.860 million, up just 652,000 from the 2016 ACS estimate. The ACS household estimate is also effectively “controlled” to independent housing stock estimates.

Here is a table showing the latest yearly increase in household estimates from various surveys compared to the average annual increases in the previous 5-year period

Recent Increases in Estimated Number of Households, Various Surveys (000's)
Latest Yearly IncreaseTime PeriodAverage Increase, Previous 5 YearsTime Period
*Adjusted by LEHC to reflect downward revisions in 2016 population estimates

One of the striking things to note is that all of these surveys suggest that US household growth in either 2016 or part of 2017 was slower that was the case in the previous 5-year period, and (not shown) well below the average annual gains of last decade. If true, then the recent declines in headship rates may either be more “structural” than “cyclical,” or may be reflective of societal shifts and other economic factors. (more on this later).

Of course, one needs to be cautious of using any of these surveys to gauge “actual” trends, as none matches up that well with Decennial Census results. However, the latest data suggest that some of the most “bullish” cases for housing demand based on “demographics” and “pent-up demand” may be “bull.”

On the latter score, Census still plans to release updated US population projections sometime before the end of this year. These projections will almost certainly show slower projected population growth than the latest available projections released in 2014, and that had unrealistically high assumptions about net international immigration.