Looking for answers: Has new home construction peaked in a short term, business cycle sense? Has it peaked in a longer term, secular sense (borrowing Mauldin's use of the term "secular" in describing the obvious 12-20 year bull and bear phases the stock markets have gone through since the beginning of the 20th century)?
Seasonally adjusted, housing starts took a big dip in November, so its worth asking those questions now. Today I dug through Federal Reserve, Census Bureau and BLS data to try and give the best answers I could.
To see if we've reached a peak relative to the business cycle I compared the current data during this Fed tightening cycle to the last two (1994-95, and 1999-2000).
When the Fed started tightening with two hikes during Q1 1994,
The Fed Funds rate had bottomed at 3% from September 1992 to February 1994.
10-year rates had bottomed at 5.55% in January 1994.
Housing Starts went from 1266 in January to a peak of 1536 in November.
It took 2.5% worth of hikes in the Fed Funds rate an almost 2.5% rise in the 10-year and about 11 months before housing starts turned down.
When the fed made their final, 50 basis point hike in February 1995,
The Fed funds rate peaked at 6%.
10 year rates had dropped back down to 7.64%.
Housing starts bottomed in March 1995 at an adjusted annualized pace of 1241.
When the Fed started tightening with a 25 basis point hike during Q2 1999,
The Fed Funds rate had bottomed at 4.75% from November 1998 to June 1999.
10-year rates had bottomed at 4.42% in 1998 during the Russian debt panic.
10-year rates were at about 5.81% at the time of the first hike.
Housing Starts were already at a short term bottom of 1562 in June.
But rose with interest rates to a peak of 1822 in February of 2000.
It took 1.5% worth of hikes in the Fed Funds rate about a 0.6% rise in the 10-year and almost 11 months before housing starts turned down.
When the fed made their final, 50 basis point hike in May 2000,
The Fed Funds rate peaked at 6.5%.
10 year rates inverted the curve to 6.4%.
Housing starts bottomed in August 2000 at an adjusted annualized pace of 1519.
When the Fed started tightening with a 25 basis point hike at the end of Q2 2004,
The Fed Funds rate had bottomed at 1.00% from June 03 to June 04.
The Fed Funds rate had bottomed at 3.10% during June of 2003.
10-year rates were at about 4.62% at the time of the first hike.
Housing Starts were at a short term bottom top in June if 2004 at 1817.
But rose with to a second top at 2039 in October of 2000.
As of November, 1.0% worth of hikes in the Fed Funds rate, but a 0.6% rise in the 10-year and almost 8 months before housing starts turned down.
When the fed made their final, 50 basis point hike in May 2000,
The Fed Funds rate peaked at 6.5%.
10 year rates inverted to 6.4%.
Housing starts bottomed in August 2000 at an adjusted annualized pace of 1519.
Looking at housing starts relative to the Fed's role in these 3 recent business cycles, this latest dip is unique in the severity, in promptness, and in relation to interest rate activity. At 13.1%, it is more than 2/3rd the size of the 4 month drop in 1994/95, and more than 80% of the 6 month drop in 2000. Starts didn't finally peak until 11 months into the 1994 cycle and 8 months into the 1999 cycle, but if the October high holds this one started just 4 months in. Finally, it took much larger rate hikes at both the short and long end to bring down housing in the earlier cycles. To sum it up, it has gotten progressively easier for the Fed to slow down the housing sector as the underlying strength of the economy and housing has weakened over time.
While building permits still remain high, it appears that homebuilders are choosing not to begin construction. More than 204 thousand units were approved but had not begun construction in November, which may be a record and is certainly a high for the recent past. Builders themselves may be sensing a change in the market, even if this isn't what they are telling investors.
If housing has turned down, and this month isn't just a statistical fluke, housing may have been much more robust in the earlier cycles because real estate was in a long term, secular bull market, while it may be entering a secular bear market now. Just as stocks benefitted over the past 20+ years from falling interest rates, credit expansion, decreasing but constant inflation, an expanding pool of investors, mass ignorance of past cycles and self fulfilling prophesies, the market for real estate appears to have benefited from:
A fall in short term interest rates from over 20% to 1%
Falling mortgage rates
Loosened loan standards
Increased GSE activity
Credit and money supply expansion
Falling prices for oil (making urban sprall and suburban living possible)
Rising percentages of homeowners
Increased real estate investment and speculation
Other factors...
All of the above operate on their own schedules. It's hard to tell in real time when each one has reached its secular peak. The advantage from cheap oil almost certainly peaked a few years ago and rising oil prices make new home construction for long distance commuters less attractive. Short term rates most likely bottomed in June of this year and longer term rates likely bottomed in June of last. We may or may not have seen the fantastic end to loosening loan standards. Fannie Mae appears to be cutting back on activity. While the Fed might not be done pumping money into the system, credit expansion in the money supply seems to have slowed.
Another interesting trend is that an increasing percentage of new home construction is being done by large builders for sale, rather than by contractors for individual land owners. This trend may speak to the conversion of farm land and the expansion on the outer edges of communities, a dynamic that could slow as prices for food and fuel increase.
There are now more than 15.5 million vacant homes in the US (11.9 million if you exclude seasonally vacant homes). When the tightening cycle began in 1994, 7.5% of rental homes were vacant. At the start of the 1999 cycle, the number was 8.1%. At the start of the 2004 cycle it was up to 10.2%. Clearly there is a much greater supply of unused homes on the market now, decreasing overall demand for new homes. The percentage of homeownership has also risen, from 63.8% in 1994 to 67.2% in 1999, to 69.2% this year. The growth in the pool of potential homebuyers has likely slowed its pace. Both factors suggest oversuppy at the end of a long cycle.
Taken as a whole, the factors that contributed to the, long secular bull market for real estate have likely reversed course. In particular, the recent rise in interest rates (if they continue on their new course) will likely mark a critical shift in market forces. November's data may be something of a fluke, especially given the size of the decline, and December may recover some, but to me it appears likely that the end of 2003 marked the secular top, that October of 2004 was the recent business cycle top, and that 2004 will be the peak year for home construction for many years. |
