More Color on Cancellations
More Color on Cancellations
Now for all readers
Since this is a novel business cycle in many ways, there are many ways to profit from understanding those novelties in real time. It can even give you an advantage over insiders, because insiders will face a lot of pressure to do conventional analysis. The role of cancellations in new home sales might be one source of advantage.
Here is a table outlining the differences between various ways of measuring new home sales. A plus means that number is included in the measure. A minus means that it is subtracted.
The Census number is the only number reported explicitly each month. Over time, cancelled sales and the resale of those units reverts to a mean of zero, so over the long term, builder net sales and census reported new home sales are the same, but there can be differences in the short term. Builder gross sales is inflated over the long term because it double counts units that were sold, cancelled, and resold again, but in the short term, it is probably the best cyclical indicator of purchase activity trends, because current sales are a coincident indicator of housing demand while cancellations are more of a correction to previously reported recent demand, though cancellations certainly do provide some indication of current frictions in demand.
I will walk through some of this data for the 2004-2012 period and for the more recent period, below, for subscribers. There may be potential for thinking through this in a way that gives you a competitive insight into coming trends.
As I mentioned, the Census number is the only number that is reported explicitly each month. In the analysis below, I use manual estimates for cancellation rates, inferred from builder quarterly reports (I roughly follow estimates from Calculated Risk for the earlier period). Be somewhat careful with the absolute numbers I use here because I also need an estimate for the number of resold units, which I have to pull out of thin air. I tried my best to use an accurate guess based on trends in builder inventory, but it is purely a guess.
With those 2 estimates and the Census sales number, I can estimate all three sales numbers. Figure 1 shows the numbers from 2004 to 2012.
From late 2005 to the beginning of 2007, the spike in cancellations and the buildup of unsold cancelled units meant that the Census estimate of new sales overstated the net sales of homebuilders.
Builders severely cut back on new starts, and even though cancellations remained high, a high proportion of new sales were made from the stock of shadow inventory from previously cancelled sales, so that, by the end of 2007, the Census number probably understated net sales because it doesn’t catch those. That was the case until around 2010, when the inventory had largely been worked off, and the Census reported sales and homebuilder net sales started moving in lockstep again.
I’ll note a few things here. First, while I think the issue of counting cancellations is interesting, it pales in comparison to the shear level of collapse during that period. What matters most is that activity just kept grinding downward, quarter after quarter.
Those cancellations in late 2007 and 2008 were from buyers who were countercyclically providing buying pressure, swimming against the currents of cyclical herding behavior, and instead of rewarding that and reinforcing a sense of courage and resiliency among American savers and investors, we screwed them and then pointed fingers at them for being “dumb money”.
It is hard to overstate the damage that was done when the Fed didn’t target a resurgence of new home sales by 2007. Not only did it worsen the long-term housing shortage and lead eventually to the deep recession and crisis; it caused Americans to internalize a pro-cyclical fear mentality as if it was a fact of nature rather than a quite explicit policy choice.
By then, the decline in homeownership rates was the steepest it had been since quarterly records have been kept. Yet, homebuyers were still treated as reckless know-nothings. Homebuilders had quickly cut off housing starts within months of the change in trend. Yet, when they were stuck with hundreds of thousands of units from cancelled orders, Fed members debated how much “discipline” to impose and builders were treated as if they had been building speculatively and recklessly and were victims of their own mismanagement. These characterizations are still the norm to this day.
Finally, if you’ll allow me a bit of self-serving just-so story telling, I can’t help but notice that the trend shift down to a second round of decline in gross sales at the beginning of 2008 coincides with the sudden shift in credit standards, as shown in the trend of credit scores on approved mortgages, which the “Credit” component of the Erdmann Housing Tracker estimates. In general, I would say there were three distinct periods.
2006-2007: Fed tightening leads to a potentially manageable contraction.
2008-2009: The Fed starts to reverse to a stabilizing posture, but not enough to overcome the deep shift in lending norms (which led to declining construction employment, various wealth shocks from collapsing home values, etc.)
2010-: A stable economy hampered by the permanency of the new lending norms which prevented recovery in wealth, construction employment, etc.
What a different world we would be living in if those standards hadn’t shifted, and if the Fed had aimed for a recovery in housing starts so that gross sales had started to increase in 2008.
The crisis is commonly described as the result of the 2004-2005 period, but really, it is purely the result of those trends after the end of 2007, which we had plenty of control to reverse, had we not been under the spell of a moral panic. In every case in post-WW II history, excess homebuilder inventories were worked off because sales recovered. In this case, federal policymakers intentionally drove sales lower, forcing builders to work off excess inventory in that horrible condition - a process that took more than a year.
That bump in net sales in 2009 is probably partly due to a first-time home buyer credit, which did help work off the last of the builder inventory. (There were much better policy options than a first-time buyer credit.) But, the stimulative effect of that credit was greatly limited since most of the buyers who might have depended on it were blocked from the market by federal underwriters, as I noted in the previous post.
Recent Trends
Now, let’s move to the recent period. In 2006, sales started to decline from sustainable levels, so that each month of decline was a further shift away from normalcy. Now, we have a strange market where demand is still low because of lending policy. Then, in 2020, there was a boost in demand from Covid migration shocks, which pushed above supply capacity (which itself was limited by Covid production shocks). And now we have reverted back to the pre-Covid limited level of demand.
Since the initial drop in sales was the result of metering sales to match constrained supply chains, there was not a concurrent increase in cancellations as there had been in 2006 and 2007. Finally, cancellations increased in the 2nd and 3rd quarters of 2022. My best guess is that some of the units from cancelled orders remain unsold, but that a decent portion are being resold relatively quickly. Builder inventories have increased, but are still low. So, the Census sales number from the past 2 quarters has overstated net sales because the Census number didn’t subtract cancelled sales. But the trend in gross sales is closer to the trend in Census sales.
The defining feature of the 2005-2010 period was a 75% decline in gross sales that took more than 2 years to develop. I frequently push back against over-focusing on interest rates. Mortgage rates were lower in 2010 at the end of that spectacular decline than they had been in 2005. The decline in the last 2 quarters is plausibly related to an unprecedented spike in mortgage rates. That has been associated with a 20% decline in gross sales since the end of 2021. That is likely the extent of it. Note it. Consider it as one of a number of factors creating marginal shifts in housing activity. But, don’t make it the centerpiece of your housing analysis.
The rise in cancellations is also likely mostly due to frictions caused by changing mortgage rates. Some buyers who couldn’t lock in rates were unable to maintain viable financing when their new homes were completed. It is possible that cancellations will remain high, but we shouldn’t conflate the past cancellations with future cancellations. The rate-induced cancellation will subside because the homes being completed this quarter, and especially next quarter and beyond, were sold after rates had risen and so the shock to buyer financing will have passed. It is possible that a deepening economic contraction could lead to continued high cancellations, but that will be largely a separate issue, related more to price expectations, potential unemployment, shifts in migration, etc.
Census numbers for October and November have been reported, and suggest that 4th quarter sales were higher than 3rd quarter sales. If that trend persists, there is potential for some positive surprises in housing data. If sales tracked by the Census Bureau remain moderate, this suggests that there won’t be a new round of cancellations triggered by cyclical decline. As it stands today, even with relatively stable Census sales over the next 2 quarters, a decline in the temporary spike of cancellations could lead homebuilder net sales (shown in the dotted lines in Figure 2) to rise at a significantly faster clip than Census numbers have.
That is just a first-order positive surprise. If builders also are able to continue selling the completed inventory at a high pace (which they already are), they will be reassured to start raising housing starts again. And, if supply chain constraints ease, and the pace of completion of the large number of homes under construction can rise, revenues of the homebuilders could really come in with positive surprises. Considering that they are generally trading at single-digit forward price/earnings ratios, there is a lot of room for a trend reversal, especially among builders with more financial leverage, etc. whose valuations have been more volatile with changing sentiment.
A loosening of supply constraints would keep input costs low, so that bullish expansion in the market would likely coincide with moderate price trends rather than a return to high rates of price appreciation. That provides another potential source of analytic advantage. Stable builder margins on rising revenues will not depend on rising prices.
And, to reiterate my cautions against over-reliance on interest rate signals, that bullish scenario will likely happen with stable interest rates while a bearish scenario with high future cancellations would likely coincide with declining mortgage rates, as it did from 2007 to 2010.
So, as numbers come in over the next few months, watch builder reports on cancellations and inventories. If those numbers improve, the news is even better than it might first appear.
(If you find this analysis helpful, please pass the word to others who might find it helpful too. The more subscriptions I can get, the more I can focus on making this analysis more thorough.)