I have been reviewing the between and the business cycle. What does this mean for the homebuilders?
I previously looked at the Philadelphia Fed’s survey of housing starts . Before the 1990s, forecasters generally expected starts to continue at their current pace, but instead production fluctuated wildly with demand. Then, after the coastal metros lost the will to grow and multi-family housing was capped by universal municipal obstructions at about 300,000 units annually, we started transitioning to a national market that was perennially short on supply.
From that point on, starts have basically been on an escalator that rises by about 100,000 units annually, which was interrupted by the massive demand shock in 2008 and the supply shock associated with Covid. Starts did cyclically decline by a few hundred thousand units in 2006 in the traditional pattern. Presumably, without the mortgage access demand shock that followed, starts would have recovered and levelled out somewhere not too far above 2 million units annually at some point. The cyclically neutral sustainable rate of construction today is probably below 2 million units, annually, but we have accumulated such a shortage of units since 2008, it’s hard to know how high active construction would need to get before it would naturally level out.
Before 2008, the construction market was generally about reaching an equilibrium number of units demanded by a given population with a given income level. Today, it’s more about frictions – how quickly can the market move from the deep disequilibrium we are currently in toward the size of housing stock a population of this size and with this income would choose.
It’s north of 2 million units annually. How far north? I’m not sure. I expect starts will return to the 100,000 units per year escalator until we get to that mystery number.
One odd thing about that first decade under shortage conditions – from about 1995 to 2005 – is that forecasters were oddly pessimistic. Instead of forecasting flat starts, they started forecasting a decline. (In Figure 1, the red line is what they expected starts to be in 6 quarters, at the time they responded to the survey. The green line is what the forecast for that given quarter had been in the survey 6 quarters prior. If the red line is below the black line, that means they expected a slowdown. If the green line is below the black line, that means they overestimated the slowdown.)
Forecasters were overly pessimistic for a decade straight, all the way back to 1994. Then, when the housing bust happened, every quarter, they said, ok, this is the bottom. This is where starts will level out. But starts kept dropping.
As I noted in the , the Fed started out that way, but became more pessimistic by the end of 2007. Then, of course, by 2010, everyone believed that they had personally called the bust by, like, 2002, which has a toe in reality. The average forecast had been pessimistic in 2002. And, by 2010, everyone had personally known all along that there had been such a glut of homes that several years of 600,000 starts had been an inevitability for some time.
Forecasters at least didn’t subsequently return to the pessimism of the 1990s. The median forecast was pretty accurate as starts got back on the 100,000 unit escalator. And, since 2018, forecasters have returned to expecting starts to be flat.
What does this mean for homebuilders?
Original Post
