Saturday, May 09, 2015

Lawler: Analyzing/Projecting Household Formations: It’s Not Just “Demographics”

by Bill McBride on 5/09/2015 02:31:00 PM

From housing economist Tom Lawler: Analyzing/Projecting Household Formations: It’s Not Just “Demographics” (With a Special Focus on the 60’s and 70’s)

When talking about the outlook for housing, economists, analysts, and even home builders themselves often focus on “demographics.” Probably the most common discussion relates to recent trends in and future projections of the US population by age, with special emphasis on the age distribution of adults.

One of the simplest “models” used to analyze and project household growth is one that breaks out the actual and projected population by age buckets (or cohorts); looks at “recent” trends in “headship” rates, and then make what is often a “subjective” assumption of headship rates by age cohort based both on recent trends and historical “averages.”

There are, of course, several “challenges” related to this approach: first, of course, there are no timely reliable estimates of US households; rather, there are multiple and conflicting estimates based on different Census surveys. The only relatively reliable estimates come from the decennial Census, which is (1) seldom timely; and (2) provides analysts with only very infrequent “snapshots.”

The second challenge is that both headship and homeownership rates were decidedly unstable in the latter half of the last century, with many of the “swings” related not just to “economic” conditions but also to massive social and cultural changes.

Consider this simple example. Suppose that at the beginning of each decade one had perfect information on what the US population – both total and by age – would be a decade later, and suppose further that one assumed that “headship” rates (or the number of householders as a percent of the population) would be the same at the end of a decade as they were at the beginning of a decade for each age cohort. Here is what the resulting forecast for the increase in the number of households would have been compared to the “actual” increase in the number of households.

  Actual Projected Assuming
Previous Census
Headship Rates
Difference
1960-197010,425,8126,781,8553,643,957
1970-198016,939,92612,747,5854,192,341
1980-199011,557,73713,267,479-1,709,742
1990-200013,532,69114,292,026-759,335
2000-201011,236,19113,546,146-2,309,955

As the table indicates, a “constant for a decade headship-rate” model for household growth produced household projections that were massively to low during the 60’s and 70’s, but were significantly too high over the subsequent 30 years.

As I hope most folks know, the 60’s and 70’s were a time of huge social and cultural upheaval and change, and I won’t discuss these. One result, however, was a massive surge in the number of people living alone – from 1960 to 1980 the number of people living alone increased to 18.2 million from 7.1 million. The 158% jump in the number on one-person households from 1960 to 1980 obviously vastly exceeded the 26.3% increase in the overall population, and the share of people living alone increased for each “adult” age group – with especially large increases in “young” adults living alone. This “explosion” in the number of people living alone reflected substantial declines in marriage rates, sizable increases in the divorce rate (from 1970 to 1980 the number of divorced people living alone double), and significant increases in the number of widows living alone. It did not, interestingly, reflect a decline in the number of young adults living at home. Rather, it reflected a surge in the number of young adults who left home to live alone.

Share of Population Living Alone by Age Group
Age 198019701960
15-647.4%4.9%3.9%
15-244.0%1.7%
25-349.0%3.8%
35-445.9%3.3%
45-6410.3%9.3%
65+27.7%24.6%17.5%

The surge in the number of people living alone, which was driven as much by and probably more so by social and cultural changes as opposed to “economic” changes, produced sized increases in “headship” rates (and big declines in average household sizes), and as a result household growth massively outpaced population growth in a fashion not readily explainable by “demographics.”

Given the surge in the number of people living along, especially among “younger adults,” from 1970 to 1980, one might have expected the overall US homeownership rate to have declined over this period (the homeownership rate for one-person households has traditionally been way below that for married-couple families). In fact, however, the overall homeownership rate increased, mainly reflecting sizable increases in the homeownership rates for married-couple families in all age groups.

Homeowership Rates
  19801970
Total64.4%62.9%
Married Couple Households77.8%70.7%
  15-2436.5%26.0%
  25-3466.9%56.9%
  35-4482.6%76.7%
  45-6487.1%80.8%
  65+83.0%78.4%
Other 2+ Households43.8%49.5%
One-Person Households43.5%42.4%

During most of the 1970’s the “typical” home buyer purchased a home that was much more modest than was the case in subsequent decades, both in terms of size, price relative to income, and housing expense relative to income, as during most of the decade buyers typically purchased a home mainly as a place to live rather than as an investment. In addition, manufactured housing was a much more important source of “affordable” housing back then – in 1980 16% of owner-occupied housing units built between 1970 and 1980 were manufactured homes/trailers.

As inflation accelerated during the decade, however, an increasing number of buyers, looking for a “hedge” against inflation, began to focus more on housing as an investment as well. That trend, which was temporarily halted by the surge in interest rates and the exceptionally severe recession of the early 1980’s, accelerated in the latter half of the 1980’s for a variety of reasons, and continued to accelerate (with a brief recession-related respite in the early 90’s) through much of last decade.

But that discussion will be in a subsequent edition.