Back in the early stages of the 2006-09 housing collapse, the economy behaved as it is supposed to behave. Housing construction fell by more than 50% from January 2006 to April 2008, but there was only a tiny rise in unemployment, from 4.7% to 5.0%. Jobs shifted from housing construction to other sectors. That’s how things work when the Fed is keeping NGDP growing at an adequate rate. You move along the production possibilities frontier, not inside the line. Indeed we would have done even better if not for the modest slowdown in NGDP during late 2007 and early 2008.
Then the Fed let NGDP growth collapse in the second half of 2008, and almost all sectors started shedding jobs. The unemployment rate soared to 10% as we moved inside the PPF.
The recent Covid-19 shock is even bigger, but there are a few heartening signs that we’ll avoid the worst. The Covid-19 epidemic has caused saving rates to soar much higher, as consumers hold back on buying many services. This depresses interest rates, with 30-year mortgage rates recently falling below 3% for the first time ever. This surge in saving also tends to boost sectors less impacted by Covid-19, such as housing construction:
US homebuilder confidence back at pre-pandemic levels
US homebuilder confidence jumped in July, taking it back to levels seen before the pandemic rattled the US economy as the 30-year mortgage slipped to a record low, data on Thursday showed.
The National Association of Home Builders’ Housing Market Index jumped to 72 in July from 58 the previous month. That exceeded economists’ forecasts for a reading of 60, according to a Reuters survey, and matched its reading in March.
It’s a myth that low interest rates are good for the economy. That’s “reasoning from a price change”. Low rates are often caused by a slump in investment demand, which is bad for the economy. But higher saving rates are good for the housing sector. If monetary policy is adequate, we should see a V-shaped recovery once the worst of the epidemic is behind us.
READER COMMENTS
Alan Goldhammer
Jul 18 2020 at 3:00pm
Scott – you are more optimistic than I am. I see a lot of small and medium size businesses going under and cannot see unemployment dropping below 10% by year’s end. There is still too much disposable income that will be sitting on the sidelines because of COVID-19 concerns. The hospitality industry will be in the dumps for some time. We will see if big national meetings make a comeback in the first half of next year.
I am unsure about the housing sector given the unemployment rates and current job uncertainties. I don’t think this a ‘build it and they will come’ scenario. I hope to be proven wrong.
Scott Sumner
Jul 18 2020 at 5:26pm
I agree that many business will suffer as long as Covid-19 is a big problem, and that means elevated unemployment. So we aren’t far apart. Still, it’s good to know that there’s enough AD so that some sectors are doing OK.
Bob Murphy
Jul 18 2020 at 11:21pm
Scott wrote: “Housing construction fell by more than 50% from January 2006 to April 2008, but there was only a tiny rise in unemployment, from 4.7% to 5.0%.”
That statistic might mislead some readers. I’m pretty sure it’s referring to housing starts. The last time Scott used these statistics to pooh-pooh the idea that the housing boom had much to do with the Great Recession, I showed that the actual construction employment figures behaved just as a “real resource reallocation” story would suggest.
Scott Sumner
Jul 19 2020 at 1:32pm
Bob, Nope, housing completions also fell by roughly 50%, which means construction activity fell by roughly 50%. So my data is accurate and not misleading.
https://fred.stlouisfed.org/series/COMPUTSA
Some of the unemployed housing construction workers went into the booming non-residential construction sector–which then collapsed in late 2008 with falling NGDP.
Bob Murphy
Jul 20 2020 at 1:24pm
Thanks for replying, Scott. Just to be clear, in the article I linked to, I acknowledged your point about housing completions. I also made an argument as to why we would expect total construction employment related to a housing boom to look the way it did.
Anyway, I’m mainly just encouraging onlookers to click through to my article and see why I think these housing stats don’t make the broader point Scott intends. Specifically, I think the data are still consistent with the story of “Too many workers went into housing during the boom years, and this helps explain what happened in the Great Recession.”
marcus nunes
Jul 20 2020 at 10:27pm
Bob
The charts in the post show that employment in construction peaked in 2006, while non construction employment peaked in 2008. The same for residential & non residential investment. The effect of the Fed big mistake in mid 2008 is also clear.
https://thefaintofheart.wordpress.com/2012/02/06/weaving-a-narrative-through-the-facts/
Kevin Erdmann
Jul 19 2020 at 2:31pm
Bob, your last graph in your link comparing US construction employment with owned vacancies looked like a real zinger, so I looked up a few representative cities to see how that data lines up from place to place.
Cities: Phoenix, Los Angeles, Atlanta, Detroit
4 Columns are:
Construction Employment Growth from 2000-2007
Construction Employment Growth from 2007-2011
Owned Vacancies 2000
Owned Vacancies 2007
Owned Vacancies 2011
MSA C07 C11 V2000 V2007 V2011
PH +37% -51% 1.4% 3.7% 3.1%
LA +25% -33% 1.4% 1.6% 1.8%
AT +08% -35% 1.7% 4.7% 4.3%
DT -26% -23% 1.1% 4.1% 1.8%
Phoenix matches your graph, more or less. A lot of construction, then vacancies.
But, it doesn’t really match regionally well outside of a few cities like Phoenix. In fact, there is no correlation at the MSA or state level between rising construction and high vacancies.
So, a lot of construction workers in LA lost their jobs when there were vacancies in Atlanta and Detroit. Were “US” construction workers part of a “GDP factory” that overproduced before 2007, or were regional construction workers part of a geographic “recalculation”?
And even in Phoenix, that rise in vacancies coincided with net domestic migration that had been +2.8% and then dropped to +0.5% or less in 2008 and after. LA, on the other hand, was depopulating when their construction employment was up 25% because even then their housing production was paltry, which is why their vacancy rate never even hit 2%.
The housing boom was a process of economic recalculation and geographic patterns of sustainable trade during a time when economic opportunity was high in some important cities that have endemic housing shortages. That process was killed by a ham-fisted NGDP shock after 2006.
One might say that the US economy was best described as a recalculation story before 2007 and a GDP factory after 2007 and that you are doing the opposite.
Bob Murphy
Jul 20 2020 at 1:30pm
Kevin, thanks for this. Can you point me to your data source(s) so I can check a few things without spending an hour looking?
Also, what is your explanation for what happened up through 2007? Were the vacancies piling up in cities where new houses weren’t being built? (I.e. how could the national aggregate look like my chart, if you’re saying that pattern only holds in Phoenix but no other big housing bubble cities?)
Kevin Erdmann
Jul 20 2020 at 6:02pm
Bob,
Let me know which data, in particular, you might want. Scott and I have been working on a paper on monetary policy, and I also have been working on papers on regional prices and housing supply through this period, so I have collected some data sets of vacancies, etc. at the MSA level. It might be easiest to just to e-mail you an excel spreadsheet of data I have collected from Fred, Census Bureau sites, etc., depending on what you want.
If you click on my name, it should take you to my blog, where you can click an e-mail link.
On your question, a couple notes I would suggest are that the vacancy measure you are using there is only for owned homes. Rental vacancies peaked in 2004 and were relatively flat after that. I would argue that the rented vacancies measure is more sensitive to basic supply and demand for shelter and the owned vacancies measure is more sensitive to transactional frictions. Second, as you or Scott mentioned, the total construction employment measure lagged residential construction during that time. Housing permits and starts began to decline in 2006 and were extremely low by 2007 and 2008 when vacancies peaked. An increasing portion of construction employment in 2006 and 2007 must have been transitioned into non-residential markets. Finally, I would suggest that the y-axis is doing a lot of work there, which isn’t your fault. It’s a Fred default, and I’m not sure you can even change the scale of the axis. The change in construction employment isn’t actually large enough to create that many vacancies. This is especially true at the metro area level. In what I call the Contagion cities (cities like Phoenix) owned vacancies, collectively, went from 1.9% in 2005 to 4.5% in 2007. It simply is implausible that builders constructed excess units amounting to 2.6% of the housing stock. That’s one of the tricky things on this topic. In this case, the more damning the evidence seems, the less damning it actually is. To imagine the extreme, if vacancies suddenly jumped to 10% next year, and that was all you knew, the scale of the change would make it clear that excess supply wasn’t the cause of it. In this case, the vacancies were from whiplash shifts in migration patterns that hit the Contagion cities hard, and market frictions in homebuying and mortgage lending.
Kevin Erdmann
Jul 20 2020 at 10:21pm
I tried to post a response. Maybe it’s held up in moderation. If I don’t see it come online, I’ll try again.
Kevin Erdmann
Jul 20 2020 at 10:38pm
Oh, while I’m waiting to see if the other comment shows up, a short answer to one of your questions is that cities like Phoenix, where active building was followed by vacancies, account for maybe 15% of the country (the vacancies in those cities coincided with very sharp downshifts in population growth < and NGDP growth 🙂 >, which is mainly what caused them.) Cities like LA account for maybe 25%, and they had active building but always relatively low vacancies. And the rest of the country generally didn’t have much increase in building, but ended up with a bunch of vacancies anyway. You add them all together and you get national construction increases followed by rising vacancies, but that only happened in about 15% of the country.
Alan Goldhammer
Jul 19 2020 at 9:19am
I came across an interesting preprint this morning as I was getting my newsletter out. A group of University of Virginia did some modeling of the medical costs of keeping the US open various SARS-CoV-2 mitigation measures. I found it an interesting paper but as a non-economist I will defer to others on whether the approach is correct.
Thomas Hutcheson
Jul 19 2020 at 12:11pm
The Fed has let TIPS expectations fall less far bellow its supposed target at this stage in this recession compared to 2008-09 (centering them on the dates of maximun, deviation.) Still the expectation for inflation is still more than 1% per year over five years below the equivalent of CPE 2%.
What can explain this massive fail?
marcus nunes
Jul 20 2020 at 8:17pm
With his Thermostat Analogy, Friedman is very close to the heart of Market Monetarists!
https://thefaintofheart.wordpress.com/2020/07/20/after-covid19-inflation/
Comments are closed.