The new home builders index (DJUSHB) opened trading today with close to a 3% jump after the Census Bureau released its housing starts report (pdf link), which showed a big jump in both starts and permits for November. This follows yesterday's National Association of Homebuilders (NAHB) builder sentiment report (pdf link), which spiked up to its highest level in 8 years. Once again, as I've demonstrated in many previous articles, the headline reports for housing starts use seasonally adjusted annualized numbers. But if you dig into the report, as I will show, the unadjusted actual sequential month to month numbers show a completely different story. I will present several reasons why the headline housing data released today and yesterday is not representative of the actual condition of the housing market and why today's bounce in the DJUSHB should be used to sell/short homebuilder stocks.
From the housing starts report link above, while the year over year seasonally adjusted annualized comparison for housing permits and starts provided headline fireworks, looking at the actual non-adjusted data for November vs. October, sequentially, shows a different story. The annualized number shows a little over 1 million permits issued. But if you look at the unadjusted number for November, it shows just 70.9k (about 850k using simple annualization). Moreover, the November number was a drop of 19.2k from October - a 21% month to month plunge. This was not part of the headline broadcasts.
Similarly with starts, the year over year seasonally adjusted annualized rate shows close to 1.1 million starts, nearly 30% higher than for 2012. But the actual unadjusted number for November was just 82.8k, 7% above October's number but 6% below the peak monthly number for 2013 reported in May. What's most interesting to me about the housing starts report is that while the homebuilders appear to be starting on over 1 million new housing units, as I demonstrated in my last new home sales report the seasonally adjusted annualized sales rate through October was just 444k. As I further demonstrated, if you consider the actual number of homes sold in October using the unadjusted number reported and factor in a general cancellation rate, the annualized run rate I derive is somewhere between 333k and 312k (I run through the calculation in the article).
Why the discrepancy between permits, starts and actual sales? There's a big difference between the cost of paying a small county fee to file a building permit and the cost of actually building the home and letting it sit in inventory. Please note that, as I've shown in previous articles, homebuilder inventories are ballooning. Furthermore, a "housing start" is defined as "when excavations begin for the footings or foundation of a building." From a cost standpoint, a homebuilder can file a permit application or can start digging a hole in the ground and then decide to stop without incurring much expense if current sales do not justify plans. Finally, based on the actual current rate of sales, I bet we will see some significant downward revisions to the housing starts numbers as they are released in future reports.
A point of clarification on the seasonally adjusted annualized data. No one outside of the Census Bureau statisticians knows how the numbers are "seasonally adjusted" other than that a complicated linear regression model is modified and used by the Census Bureau. What we do know is that it uses mathematical relationships of previous data and growth rates and has been shown to significantly overstate the forward projected growth rate after an extended period of growth is followed by downturn in the data. This would explain why the new and existing home sales rates reported over the summer have been subsequently revised lower, in some cases significantly, in the most recent reports as I've detailed in previous articles. For point of illustration, as this graph linked on housing starts going back to 2005 shows, the last time the seasonally adjusted annualized rate jumped as much as it did in today's report was twice in 2005, which was when the bear market in housing began.
Finally, the housing permits and starts data do not in any way correlate with the mortgage purchase application data. You'll see from this mortgage purchase applications index link that applications being filed and used to purchase homes has been declining since mid-2013. While buyers who use a mortgage do not capture the entire homebuyer market, most buyers who purchase a new home do so with a mortgage. If mortgage applications are declining - and today's MBA mortgage index report shows a 6% drop in purchase applications from last week, with 8 of the last 12 weeks registering weekly declines - it means there were will be a drop going forward in actual new home purchases.
Touching briefly on the NAHB builder sentiment index, it showed a big spike higher in yesterday's report (linked above) and was well in excess of expectations. The term that comes to mind with this report is "irrational exuberance." I say this because the last time this particular report's index was at 58 or higher was in 2005. The high reading in this home builder confidence index in 2005 exactly correlated with the peak of the housing bubble and the start of a precipitous decline in housing prices and sales volume. This graph linked illustrates this point.
As I've thus laid out above, the actual economic condition of the housing market is significantly different than the impression one might get from just looking at the headline reports and media broadcasts. In fact, as I've shown, if you look at the actual non-adjusted monthly data and use month to month sequential comparisons rather than year over year seasonally adjusted annualized comps, the market appears to be headed south.
On this basis I continue to recommend using any big bounce in the DJUSHB new homebuilder index - like today - as an opportunity to sell positions and establish shorts in select builder stocks. At the very least any portfolio manager who might be overweighted in the sector has a fiduciary duty to look closely at the data and at take profits and pare down position sizes. My favorite builder shorts continue to be DR Horton (DHI) and KB Homes (KBH). I also like Beazer (BZH), Pulte Homes (PHM), Lennar (LEN) and Ryland (RYL). You can also take leveraged short position by buying the ProShares Ultra Short ETF (SRS), which is 2x short the Dow Jones US Real Estate Index.
As I have stated in the past, leave yourself room to average "up" on your cost if you establish a short and the market spikes up against you after you dive in. The sector is volatile and already heavily shorted. Remember, you are taking a long term view of the difference between the current market value of these stocks and where they will be a year from now when the housing bear is in force.