Over the last few months headlines have simultaneously touted rising AND falling home prices (or at least indices). As seen in the table below, nominal, non-seasonally adjusted (NSA) home prices have continued to grind higher (albeit at ever slower implied HPA) while seasonally-adjusted (SA) home prices index values have declined for several regions. For example, a reporter could either look at the non-seasonally adjusted numbers (in green) and write that price levels are rising, while someone “looking under the hood” might rather focus on the seasonally adjusted numbers (in orange) that show a decline over the last three months.
While both can be correct, the dilemma that I’d like to focus on is how can there be falling SA prices, while one-year forward prices on the CME Case Shiller futures are higher? After all, home prices tend to trend, and if SA prices are falling shouldn’t that imply that CME market prices one-year forward (to avoid seasonality issues) should also be lower?!?
(This is, of course, critically important to me as the market maker, but also useful to reconcile any inconsistencies between headlines and reality).
The lower portion of the table to the right shows mid-market levels for the CME contracts on the same Case Shiller indices that have declining SA index values. Year-on-year prices (so Nov ’15 index vs. Nov ’14) are showing gains of 4.6-6.7% despite each index declining over the last three months. Unless one argues that the recent decline is temporary and index levels are about to reverse, or unless one argues that the CME prices are not real (even though they are consistent with the Pulsenomics survey on a Zillow index) there’s an inconsistency that needs to be addressed.
My sense (and the theory that I want to propose) is that is that since the NSA numbers are calculated while the SA are derived that the inconsistency between reported falling SA index levels and higher CME prices may be function of issues surrounding seasonal adjustment factors.
The graph to the right shows the difference (in percentages) between the NSA and SA values for each of the ten CUS components (and the HCI -10 city index) since 2000. For purposes of this blog I’m labeling the difference as the “seasonal adjustment factor” that is used to translate NSA #’s in SA #’s.
One can easily observe that these seasonal adjustment factors have risen sharply over the last few years when compared to the relative stability from 200-2007. The spike in seasonality factors coincided with the crash, and subsequent rebound in home prices.
Recall that during 2007-2013, the percent of homes that sold as distressed rose dramatically. These sales tended to pull down CS index values in the selloff, while the reduction in %-distressed gave rise to some of the bigger gains in CS indices over the last few years (particularly in places like Phoenix, Las Vegas). My sense is that as distressed properties tended to be sold when they could (as opposed to when the original homeowner might have wanted) they came to account for an ever larger component of sales during seasonal lows (e.g. during the winter). As such, that higher amount of distressed sales tended to exacerbate existing seasonal patterns. For some poorer-performing northern cities (e.g. CHI, showing the largest seasonal-adjustment factor in the graph) those seasonal-adjustment factors expanded dramatically.
The role of distressed sales would seem to be a theory that supports the changing pattern observed in the graph.
The issue for today’s dilemma (falling SA index values versus higher CME prices) is that seasonal adjustment factors that may have been appropriate when distressed sales dominated index values, are now being applied to a market that has somewhat rebounded and stabilized.
In effect, I would argue that seasonal adjustment factors that may be too large for today’s markets can explain the inconsistency of falling SA values and higher forward prices. Net, SA prices would not be falling as much, or might even be positive, if lower, long-term trend seasonal adjustment factors were being used.
Again, this is only my two cents, but an angle that I need to have resolved to address the inconsistency I’ve described above. If true (that seasonal adjustment factors are today, too large) then some of the negative headlines about falling home prices also needs to be taken with caution.
So which is it? Are seasonal adjustment factors too large? Are the falling SA prices only a temporary dip? Are CME prices too high given declining SA prices? Trading in CME futures tends to increase when there’s a change in sentiment and it seems that we have two diametrically opposing sentiments at play -home prices will be higher vs. home prices are falling. There should be more trades as people take sides.
I’d love to hear some thoughts on any of the questions in the last paragraph, or to turn any of these discussions into trading ideas. Feel free to contact me (firstname.lastname@example.org) if you want to help me stir the pot.