Zillow's recent monthly update to their 12-month HPA forecasts provides a great example to a theme that I've been articulating for the past few years -that longer-dated home price futures tend to clear at a discount to expectations.
That is, while futures and expectations (on Case Shiller indices) need to converge as contracts near expiration, for longer-dated contracts there are (today, given low volume) more Natural Shorts (e.g. hedgers of individual homes) willing to sell at the "expected" level of future home prices, than there are Natural Longs (e.g. pension plans, renters, future buyers) willing to buy at that level. Speculators, who make up the difference between the two groups prefer a favorable risk/reward and thus bid below expectations.^1
There are only a small number of forecasts of forward Case Shiller levels (thank you CoreLogic) and most of those are at the national level. However, the Zillow forecasts are public, and cover 100's of metros. (PlacesPlatform - with whom I have an affiliation -also projects forward home prices, out to five years and at a very granular geographic level.)
The graph below shows the year-on-year ("YOY") percent differences for the mid-market levels for the CME Aug '23 and Aug '24 contracts (on the Y axis) versus Zillow's most recent forecasts on the X-axis.^2 I've also inserted a 45 degree line (in red) to highlight the difference between the two sets of numbers.
Note that the mid-market level for the YOY % changes in the CME Case Shiller futures are several percentage points lower than for Zillow's forecasts (with LAV having the biggest difference of 7.0% (4.2% Zillow forecasts minus -2.8% discount of Aug '24 mid-market to Aug '23 mid-market)).
While there are reasons that the actual Zillow index (not their forecasts) might run higher than the corresponding regional Case Shiller index (e.g. Zillow including new homes, Case Shiller reducing the weight of repeat sales that took place decades apart), the premium of Zillow over Case Shiller in HPA, for LAV, has averaged 42 basis points per year for the last ten years (see below).^3
Even so, while CME LAV might appear attractive to those who embrace Zillow's views, what I see in my market-making capacity is mostly sellers (as talk about reduced affordability and a pending recession concerns current homeowners).
Now, I concede that there differences between any two approaches to creating home price indices to include: approach (Zillow uses hedonic while CS is based on repeat sales), universe of homes (Zillow includes all homes, while Case Shiller doesn't include new home sales), timeframe (Zillow numbers are published for the preceding month, while Case Shiller has a two-month delay on a three-month moving average), and geographic boundaries of an index (e.g. the Case Shiller NYM index includes parts of New Jersey and Connecticut).
Nevertheless, the two sets of indices have done a reasonable job tracking each other over the last ten years. With the exception of the NYM index (see color above) the regional spot Zillow and Case Shiller indices all have correlations above 85%. (The correlation can be further improved by running a weighted average of three Zillow indices that mimic the Case Shiller measurement period, but for purposes of the above illustration I'm limiting the analysis here to spot vs spot.)
Not only are the indices highly correlated but the average of the YOY percent changes for many regions (except NYM, CHI and MIA) are close.
Net, if it's true that longer-dated CME futures tend to clear at a discount to expectations, than there seems to be favorable risk/rewards on the long side of the CME regional contracts. I leave to the reader as to how to hedge that position.
Feel free to contact me to discuss this post, or if you'd like to learn more about how home price index derivatives might be used in hedging strategies.
Thanks, John
^1 - This discount might become smaller should volumes increase, making institutional participation more likely. Also, any programs that allow renters and future home buyers better access to trading futures, might also narrow this gap.
^2 - Note that I used the difference between the mid-market of the Aug '23 and Aug '24 contracts (instead of spot vs. Aug '24) to compare two 12-month periods. However, given the lookback on the Case Shiller index, these are not measuring home price activity for the same 12 months. That said, the impact of adjusting for slight differences in the measurement periods is likely overwhelmed by the difference between the CME markets and the Zillow forecasts.
^3 - Convergence is not assured between the two indices, but let's revisit this blog in a year to see if any occured.