This morning’s miserable new home sales report gives us an excuse to write about a theme we’ve been distantly following for some time: the possibility that the housing downturn has overshot the pre-crisis boom and will eventually provide an economic boost when it corrects.
Whether this turns out to be the case depends on whether demographics can trump the recent adjustments in household behaviour, as a recent paper from RBC explains:
Some investors have expected that population growth would drive new home sales back to prior levels. We believe that this view intuitively makes sense and agree that demographic trends will ultimately drive demand in the long-term.
In our opinion, however, other factors including vacancy levels, declining homeownership rates and an increase in household size are temporarily exerting a disproportionate influence on housing demand. …
Historical Trends in New Home Sales
Median new home sales have dropped precipitously since the recession began in 2007. The United States experienced a median of ~675,000 new home sales from 1980 to 2006. Since 2006, however, median new home sales have been 430,000 annually.
And here’s a graph from Calculated Risk updated to include this morning’s figures (click to expand):
The bullish case — the argument that eventually we’ll see a burst of construction activity after a reversion to previous levels of housing demand — is partly based on straightforward population growth. Eventually more households will have to form and we’ll start to see prospective buyers of new homes emerge.
Here’s more detail from RBC…
The United States population has increased by ~3 million people on average since 1990. It is currently projected by the US Census to increase from ~310 million people today to ~340 million by 2020. This assumes a growth rate of ~1% or ~3 million people annually.
… and in graphical form:
Of course, what also matters here is whether or not there’s a lot of excess capacity already in the market. This is difficult to say, but if you look at the pace of homes built in the last few years relative to the population growth, we can at least start to get a sense of whether or not there’s currently a housing shortage in the making.
Helpfully, Dan Indiviglio did exactly this last month:
As you can see, the rate of population change per homes created declined over the last ten years from the previous decade. It’s higher than the decade before that, but not by much — suggesting that, as Indiviglio writes, “if there’s still some excess capacity in the market, there doesn’t appear to be much.”
Now back to RBC and the factors likely acting as countervailing forces to demographic trends: an increase in household size, declining homeownership rates, and vacancy levels.
To take each of these in order…
First, after decades of a declining population per household (one impetus for new home starts and sales during that time), we’ve seen an uptick in the last few years:
The trend driving this is the number of jobless college graduates and young adults moving in with their folks. RBC: “The Harvard Center for Joint Studies on Housing estimates that the number of young adults (age 20-29) living at home rose by 1.6 million from 2005 to 2010.” Newly jobless adults of all ages moving in with relatives and friends are also likely contributors.
Second, the national homeownership rate has dropped from a peak of about 69 per cent in 2004 to 66.4 per cent now. Even more worrying is that the biggest decline has come in the under-35 category, which “contains the largest proportion of potential first-time homebuyers”, according to RBC:
Third, high vacancy rates matter because prospective homebuyers could opt to buy cheaper existing homes, including those foreclosed on, rather than new homes. So higher vacancy rates now would imply a longer time before demand for newly built homes returns.
The estimates on vacancy rates tend to vary considerably by economist, making this the haziest of the three factors listed above.
The census bureau tracks that number. However, its month-by-month estimate was well out of whack with the preliminary data coming in from the formal 10-year Census. Still a credible guess is that there might be in the range of 1.5 million “excess” vacant homes. That number includes empty rentals as well as homes for sale. Even in the best of times some homes are vacant, but there are roughly 1.5 million more than there were in 1990, adjusted for population changes.
That 1.5 million number is less than our estimate of the number of missing households. Somewhere around 500,000 new homes are scheduled to be completed in 2011. However, typically around 200,000 to 300,000 homes are demolished each year as they become unlivable.
This suggests our total inventory of available homes is less than our total number of “shadow households” and is not about to catch up anytime soon.
RBC, for its part, pegs the number a little higher…
… and is less optimistic:
Based on anecdotal evidence from the National Association of Realtors (NAR), we believe that remarketed foreclosures are being priced on average at a 20% discount to new homes. The large size of this discount suggests to us that many consumers will elect to purchase foreclosed properties as opposed to investing in a new home even if homeownership rates remain constant. We expect that the level of vacancies will remain stable at ~2 million homes which is meaningfully above the long-term average.
There’s a chance that vacancy rates could be high very high now and remain high if homebuyers were to simply express their choice for new rather than existing homes. But this is certainly not what’s happened lately, as the percentage of total sales that new home sales has accounted for continues to decline:
On these points, it would seem that RBC economists are right to be pessimistic — to think that too many current trends will continue to overwhelm natural demographic pressures for more housing construction. And if you really want to pile on, check out Scott Sumner’s recent post on why previous household formation trends won’t return anytime soon.
Sumner notes that along with young people moving in with their parents, another problem is something that we previously worried about in this space: less immigration, caused both by recent crackdowns and by less incentive to move here because of the stalled economic recovery.
But as ever, the whole story is actually a bit more complicated.
As both Smith and Indiviglio have noted, the pre-2006 boom in housing took place almost entirely in single-family homes, while new construction of multi-family residential buildings stayed flat (chart from Indiviglio):
Multi-family structures, of course, are those that typically rent out their units. We’ve seen a significant climb in rents inflation in the past year or so, which makes perfect sense: increased demand for rental units coincides with the lower homeownership rate. Indiviglio adds that once young people finally do decide to move out, it will likely be to rentals, given their limited credit histories and tighter standards for buying homes. The same would apply to the jobless who have stayed with friends and, especially, to those previously foreclosed on — given their damaged credit histories. So there’s a good chance this pattern will remain.
All this has obvious implications for the possibility of new construction activity. Given that there never was a boom in multi-family buildings, this is precisely where the shortage would be (Smith’s case all along) — and if demand for rentals continues, at some point you can expect to see a supply-side reaction, ie more construction. In addition, existing single-family homes might be turned into rentals, which would (helpfully) reduce the amount of shadow inventory. All of this would also be entirely consistent with a more fundamental shift among Americans in the perception of the value of owning homes versus renting, if there’s been one.
At any rate, we find the case persuasive, so we’re not ready rule out the possibility of a meaningful increase in housing activity just yet. It’s too early to know what will happen, and as Ryan Avent explains, construction cycle operates with a big lag on increased demand, so we wouldn’t see much progress for a while, maybe a year or two.
What everyone writing on the matter seems to agree on is that a higher level of economic growth would be needed to launch the construction takeoff. Growth — and jobs in particular — is needed to rejuvenate household formation, get young people to strike out on their own, reduce vacancy rates, and attract immigrants. Once this happens, the demographic trends will do the rest of the work.
But that’s an awfully big “once”, as the prospects for higher growth in the next few quarters remain uncertain. So although we tend to agree with the idea that there’s a burgeoning housing shortage, we remain agnostic on when the eventual correction will begin.