Friday, May 17, 2013

Homeownership, Unemployment and Economic Growth: Looking at the Trends

A new study* that has been receiving attention in the media claims that rises in the homeownership rate in U.S. states are correlated with subsequent sharp rises in unemployment in those states, and advances some provocative explanations to account for the correlation, including the conjecture that homeowners are more prone to support restrictive zoning that impedes business formation. The authors also find evidence that high homeownership is associated with lower labor mobility and long commutes (a point which contrasts with Randal O'Toole's argument, which I have mentioned before, that high homeownership is a product of high labor mobility).

As sympathetic as I am to arguments about the economic downsides of homeownership, I was a bit surprised by these findings, knowing that the largest increases in the homeownership rate during the period the authors examined, 1950-2010, occurred in the southern states, and that these states, I believed, also experienced the most rapid economic growth during that same period.  The study does not examine or even mention state income, so I decided to chart income growth in 49 states (excluding Alaska) and Washington D.C.

I have posted the full results here, but below is shown the top and bottom ten performing states over the 1950-2010 period, using household median income (figures are in 2010 dollars):

This is about what one might expect -- the Southern states, starting from a very low baseline, did very well during these six decades, while the rust belt states stagnated. Since we are talking about housing, though, what happens if we take into account the change in housing values, using these values to estimate the cost of housing in each state?

Discounting incomes by the cost of housing (based on the annualized expense of a 30-year mortgage at 6% interest, with 20% down**) gives the following results:

The Southern states again dominate the best performing list, with Maryland dropping out (note that Mississippi was the only state of all 50 examined where housing grew more affordable during the period examined relative to income).  Due to tremendous increases in housing prices, however, several states have moved far down the list, including New York and California.  Oregon and Montana are unusual in that housing prices increased substantially despite poor wage growth.

Finally, I correlated all the information I compiled plus the change in homeownership rates over 1950-2010 (showing correlation coefficients):

The chart shows that increases in income are strongly associated with increases in homeownership.  This compares to the weaker correlation found in the study between unemployment and homeownership over the same period (an R squared value of .108 as compared to .280: see Figure 2 on page 17). More surprising is the relatively low correlation between change in income and change in housing value: of the top ten states for housing cost growth, only two (Maryland and Virginia) were also on the list for top income growth.  For comparison, although North Dakota, Utah and California each experienced income growth of 88%, home prices rose 189%, 294% and 483% respectively. 

I didn't plot unemployment change over this period, as I'm not sure that such a volatile statistic is all that useful when measured at only two points in time. In any event, even a casual glance at unemployment rates during the recession shows that states that had the lowest levels of homeownership experienced some of the highest unemployment levels (notably California, Oregon, Nevada and New York), and have also had some of the highest levels of domestic outmigration, perhaps due to the very high housing costs that have suppressed homeownership rates in the first place.

Although none of this alters the patterns carefully discerned by the authors, I think it does at least point to the complexity of the homeownership phenomenon, which is affected by numerous economic, demographic and political factors, and which benefits from being studied in the broadest possible context so as to avoid placing too much importance on any particular correlation.

*David G. Blanchflower and Andrew J. Oswald, Does High Home-Ownership Impair the Labor Market?
*Rates were of course lower than this in 2010, but 6% is a rough long-term average for the past two decades, and in any event mortgage rates were comparable in 1950 (4.5%) and 2010 (4.7%).


  1. Isn't using 1950 something of a problem? In 1950, 35% or so of Mississippi's population was confined to (mostly) rental ghetto housing and denied equal employment opportunities. As those barriers were dismantled, they were able to get better jobs and move out of overpriced ghettos. Naturally, both the average income and the homeownership rate of black Mississippians has increased dramatically since 1950. That doesn't tell us anything about the effect of home ownership on income, other than whatever effect it has will be completely dominated by a system of de jure and de facto racial segregation.

    1. Thanks for the comment, Chris -- this is a nice invitation to look at the specific numbers. According to the Census data, in 1950 60% of white Mississippians owned their homes, as compared to 31% of "non-whites." Black homeownership had already risen rapidly since 1940, when it was only 18% (in fact, black homeownership in Mississippi rose faster in the 1940s than white homeownership did, so this was not purely a post civil rights-era trend).

      As for unemployment, the 1950 Census would have us believe that in Mississippi it was only 3.4%! This contributes to the apparent dramatic rise in 1950-2010 unemployment that underlies the authors' findings. The black unemployment rate is reported as 4.5% in that year, which hardly seems believable to me, as compared to 18% in 2010 (I doubt the way the numbers were computed is comparable, but the authors seem to think it is). What can possibly be made of these numbers?

      In the end, homeownership for all Mississippians rose greatly over 1950-2010, but the changes in unemployment varied dramatically among subgroups. This alone calls into the question the interpretation of the correlation by the authors, but it does require drilling down deeper into the state-level data to see exactly what is going on.

  2. Rather than unemployment itself, was the study based on unemployment volatility? I understand there can be a correlation between income and ownership, and even levels of unemployment and ownership. But correlation is not causation, and while rising income is more likely to have a caustive relationship, I'd think uncertainty (volatility of employment rates) would also play a role.
    Secondarily, what role would education have? Most of the people I know who didn't get a college education typically rent rather than own.

    1. Hi Richard -- I looked at the study again, and I can't see any indication that volatility itself was measured, just what happened in various states following rises in the homeownership rate. If I had good information on volatility from 1950-2010, I'd try to plug it in as a variable myself!

      As for education, I think you are right, and that we'd see income, education and increases in homeownership all positively correlated with one another. A caveat is that homeownership is also higher in rural areas and lower in urban ones (note the generally high rates in more rural states), so that might tend to sway things in the opposite direction.

  3. it was a wonderful chance to visit this kind of site and I am happy to know. thank you so much for giving us a chance to have this opportunity.. הכל על יין אדום