In promoting potential home price index agreements (#HPHF) for cities not referenced by CME Case Shiller contracts, I've stumbled back into the topic of revisions. Since I (and many other academic works) rely on past home price trends to extrapolate into forecasts of future index levels, changes to recent index levels can generate meaningful changes to HPA, and therefore have a potentially big impact on forecasts. While this may be the "nature of the beast" for forecasting (where no money is involved) it wrecks havoc on home price index agreements as even relatively short-dated clearing levels may change 1-2%. In some cases that translates into 2-3 point price changes.
I ask (and not rhetorically) how can one post 1-1.5 point bid/ask spreads on multiple short-dated HPHF regional index agreements, if the price might (should!?!) change by >2 points on an index revision.
An example of recent revision is the history of the Pittsburgh index. (Note that I picked Pittsburgh as I'd quoted it a few months ago. While the revisions to the Pittsburgh index are meaningful, they are dwarfed by some of the changes in the smaller Freddie Mac geographic regions.)
The table below shows the Pittsburgh SA index values (since Dec. 2017) that had been released as of May 1 (the March data) and Aug 1 (the June data). Through May 1, the YOY price change was 3.37% and the 2019 YTD annualized gain was 2.72%. Based (heavily) on those numbers, I posted a Pittsburgh home price agreement for the Dec data (to be released near Feb 1), of 173.4/175.1 consistent with 2019 HPA of 2.25% (on the bid side) and 3.25% (on the offered side).
However, over the next few months, the Freddie Mac index data was revised higher (by > 2.0 points in the case of the March data). Those changes had the effect of raising the YOY gains to 5.51% and the 2019 YTD numbers to 5.48%. ( I concede that these each measure different time frames, but even the 2019 YTD gains using the same three months as before leads to ~5.08% annualized gain for 2019).
Based on the higher revised historical gains, I've re-posted levels of 177.4/178.7 for Jan home price index agreements. Not only is the bid 4 points higher, but my sense is that during this period Case Shiller futures prices have fallen.
Net, it this just a reality of forecasting, something (the revisions) that can be anticipated, and/or is this an opportunity for people who can forecast revisions? Are revisions predicable, and might they be correlated, such that positive revisions in one month are more likely to be following by future position revisions?
I'm going to be studying this, but am also (very) open to taking suggestions from others -all while continuing to post agreement levels for smaller nominal exposures.
For all the reasons that I support the notion of having robust home price index markets on the CME, I believe that home price index agreements on other cities are a potentially valuable hedging tool for those looking to change the exposure to home price index movements.
Feel free to contact me if you have any question (or advice) about this blog, or on any aspect of hedging home price indices.
Thanks, John