I wish I had caught this a few months ago when it came out, but the Federal Reserve went ahead and re-interviewed folks studied in the 2007 Survey of Consumer Finances. The SCF does not usually follow specific people through time—the last panel SCF was in 1989—so this study offers an interesting opportunity to see what happened to family finances following the collapse of the housing bubble.
Of course, we have produced several papers based on SCF data in which we estimate the impact on various groups of people based on housing and stock prices. But a follow-up study of the 2007-vintage respondents would in theory be a better approach.
According to the Fed study, median family net worth fell 23 percent from $125,400 in 2007 to only $96,000 in 2009. Largely, this change reflected falling assets rather than increased debt. In particular, the median value of primary residences fell 15 percent from $207,100 to $176,000.
However, the numbers reported by the Fed do show something surprising. According to the panel results, the percent of families owning their primary residences in 2007 stood at 68.9 percent. This figure is slightly higher than the quarterly numbers reported by Census over the survey period (67.8-68.2 percent.) Over the same period two years later, Census reported homeownership rates had fallen to between 67.1-67.4 percent. The Fed, on the other hand, estimates that in 2009 some 70.3 percent of the 2007 panel families owned their primary residence.
Importantly, much of this reported surge in homeownership is heavily concentrated in the least-wealthy families. For every 30,000,000 families in the bottom quartile of 2007 net worth, the SCF implies an additional 1,650,000 homeowners and another 600,000 in the second quartile. By contrast, at least 630,000 homeowners in the top half of 2007 net worth no longer owned their homes by 2009.
On one hand, falling home prices have made homeownership much more affordable for the non-homeowners in 2007, and these renters were heavily concentrated in the bottom half of the wealth distribution. Thus, it is not surprising that many bought homes in this time. However, the 2009 SCF implies at least 7 percent of the less wealthy half of families became new homeowners within two years of the previous survey.
Some (half or more) of the differences between the SCF and Census may be explained by the fact homeownership rates increase with age and that the potential 2009 SCF population is two years older than the 2007 SCF population. For example, homeownership among those aged 35-44 in 2007 rose from 66.6 to 68.9 percent in the SCF data—a 2.3 percentage point change. Simply adjusting the 2007 Census homeownership rates for ages 35-44 to ages 37-46 would result in an increase of 2 percentage points. Thus, even accounting for aging, the SCF data show an increase in homeownership rates that requires some explanation.
There is also likely some upward bias in homeownership due to survey dropouts. If the homeowners who dropped out of the survey did so because they wound up deepest underwater in their mortgages, then this would bias upward the changes in ownership. If the renters who dropped out of the survey did so because they were in tough situations, this would bias upward both the initial ownership rates and (because they would be less likely to own by 2009) bias upward the increase in ownership.
The Fed authors do refer to the effect of aging and explore the possible character of the dropouts, but it is not clear that the dropouts are predictable based on 2007 data. It seems more likely that a change in the household situation rather than initial net worth would lead to nonresponse. Unfortunately, it is the changes themselves that are of interest.
In sum, we are pleased that the Federal Reserve authors have undertaken this interesting panel study but it may well understate the deterioration of household net worth since the financial crisis. The results should be taken with a grain of salt.