In what is probably the most unusual new home sales report I've ever seen released by the government's Census Bureau, new home sales for both September and October were released today. Although the media headlines are reporting a massive gain in new home sales for October, a careful analysis of today's report reveals serious weakness in new homes sales, beginning in July, and it shows fundamental problems with how the Census Bureau ((NYSE:CB)) estimates monthly new home sales. Based on my careful review of the numbers, my thesis that the housing market is heading off of a cliff continues to be reinforced.
Because of the Government shutdown, the CB released new home sales for both September and October. For starters, the number published for August - originally said to be 421k and a 14.1% sequential increase over July - was revised down massively to 379k. With this revision, the original report that showed an increase from July to August has now become a 2.9% decline from the original July revision. But July's revised number was once again revised lower from 390k to 373k. While this revision yields a slight gain from July to August, August will likely be revised lower once again next month. You can read my analysis of the original report for August here: Housing Data Further Confirms My Bearish View. As you can see just from the big downward revisions to both July and August, it's dangerous to chase spikes in the homebuilder stocks thinking that initial report is bullish. It also demonstrates the problematic nature of the Census Bureau estimates, especially when the market is in a state of decline.
For as problematic as was the original report and the huge downward revision for August, so too is the current report for both September and October (Census Bureau data link). The September new home sales number is estimated to be 354k. Not only is this a 6.6% decline from August, but it is also highly likely that both August and September will once again be revised lower. For October, while the initial estimate shows a 25% headline increase over September and 21% headline increase over October 2012, given that the three previous months have been revised lower by a significant amount, with July subsequently revised lower twice now, it is highly likely that this number for October will also incur substantial future downward revisions. I say this because it would be consistent with the reported big decline in mortgage purchase applications that began in the spring: Mortgage Purchase Applications Graph (Zerohedge). It would furthermore be consistent with the increase in cancellation rates being reported by new homebuilders. You can review my last few articles on Seeking Alpha which document and cite the cancellation rate data.
Although I have shown the difference in the seasonally adjusted, annualized monthly number vs. the estimated actual monthly number of sales in past reports on new home sales, I think it is worthwhile to look at the numbers for October in detail (data linked above). If you scroll down to the last page you'll find the not adjusted data for October broken out by "not started, under construction and completed." As you can see, of the 35k estimate, nearly 69% are either not started or under construction, with only 11k completed. If we take a simple mathematical annualization of the 35k (35 x 12 months), that yields 420k annualized, which is lower than the annualized estimate. This is not a bad assumption given that trend in the revised numbers is showing a significant decline in new home sales.
However, if we apply a cancellation rate of 25%, which is about the average among several new homebuilders right now (see my past articles for data and sources), and apply it to the 35k, that would mean 26k homes would actually be delivered. This results in a 312k annualized rate. Although there would be some seasonal bias using just a simple annualization calculation, given the trend in new home sales over the summer through September and given that interest rates are rising again, my bet is that 312k may actually be on the high side. As we saw in the last housing bubble, the cancellation rates in general were in the low 30s, with companies like Beazer (NYSE:BZH) reporting cancellation rates in the 40s. If take a 25% cancellation rate going forward and apply it to the 444k seasonally adjusted annualized number reported, that yields 333k annualized, which is consistent with my estimate of the monthly-unadjusted data.
Another bearish trend in today's report is new home pricing. If you scroll to the second page of the report linked above, you'll find the data for median prices. The sequential decline in price from September to October was $257.4k to $245.8k, a 4.5% decline. But if you calculate the price decline from the high-tick in April of $279.3k, it's a 12% decline in price. If homebuilders have had to discount prices 12% in order to generate sales, with the decline in mortgage purchase applications and rise in interest rates, how much more will prices have to be lowered in order to move new homes? Furthermore, price cuts like this will severely impact cash flow and profitability.
Based on the data and my analysis of the latest CB new home sales report, I thus believe that the decline in the housing market that began this summer is likely starting to accelerate. Certainly the degree of downward revisions to the numbers since July would confirm this. It would also seem the stock market agrees with my analysis and view, as the DJUSHB home construction index is down .8% and down 1.3% from its initial spike higher as I write this. As such, I recommend that anyone still long homebuilder stocks should at least take profits to the point at which you don't have any investment basis in the stocks. However, I believe the DJUSHB will eventually fall to at least its low of 143 on November 21, 2008 (it's currently at 437). I continue to like DR Horton (NYSE:DHI) and KB Home (NYSE:KBH) as my favorite short-sale names. Other names to consider and for which I'll write reports in the future include Beazer, Pulte (NYSE:PHM), Centex (CTX) and Lennar (NYSE:LEN). The caveat in shorting the homebuilders is that they have very high short interest, which creates high volatility. I like to use market bounces to short them and I always take some profits on declines.