There is a danger in citing correlations when there are outliers in the data, such as with the recent moves in home prices. That the power of one or two outliers, can dwarf any small differences in the rest of the sample. While this might be obvious to someone three weeks into a statistics class, it took some marketing feedback, and analysis of recent Freddie Mac home price updates, to remind me of the math behind this truism.
Importantly, the notion of "be careful with correlations" reinforces the need for users to apply both some of the macro and regional tools I've written about in past blogs.
For example, the top scatter diagram shows the YOY returns of the Case Shiller 10-city NSA index from 2014-2021, mapped against the YOY returns for the Freddie Mac NSA Rochester, NY index. The R-squared is enviably high (~84%) suggesting that the Case Shiller 10-city index might have been a good hedge. As such, someone looking to add or reduce exposure to Rochester, might have been able to hedge "large" moves, using the CME Case-Shiller 10-city index contracts. Such a hedge might have protected the user from such moves in the Rochester index, as there was a proportional move by the Case Shiller 10-city index.
Thus, if someone is worried about another +/- 15% move in Rochester home prices, the historical performance of the Case Shiller 10-city index might argue that prices have moved together (although there are no guarantees for the future), and as such, the user might consider the CME 10-city index futures as a hedge.
(BTW- This seems to be the case for the vast majority of regional indices. That is, the presence of outlier prices (for both the Freddie Mac and Case Shiller 10-ctiy index) during 2020 and 2021 has resulted in high R^2 across many comparisons of regional vs. 10-city index moves.)
Of course the large recent price changes (i.e. 2019-2021) came during an unusual timeframe. Covid has turned homeowner preferences upside down with buyers looking for more space, a platform for working from home, and (with price rises) a search for more affordable areas. inventory shortages have also incented people to grab what's available, contributing to large HPA gains. If 2019-21 is not the norm, what might correlation look like when you exclude the boom years.
For instance, a glance at the correlation from 2014-2019, shows a much weaker relationship, during a period where home prices did not move much for either Rochester or the 10-city index. As such, for this timeframe, hedging with the 10-city index futures would not have been as accurate a hedge. Given the talk of home price gains slowing in 2023, is this the more relevant diagram to decide on hedging strategies? If so relative performance agreements (structured as HPHF Ratio Agreements) might be a better hedge.
Since one doesn't know how home prices might play out over the next few years, I'd argue that a combination of the two hedges might be most appropriate. That's because, if one is "short" a Ratio Agreement (i.e. long 10-city index/ short regional) and then short the 10-ity index, the two 10-city index exposures cancel each other out (depending on the notional amounts and boundary conditions on the Ratio Agreement, short the regional index (over a modest range). (See recent blogs "Why a regional overlay might be important in hedging home prices post-Covid" and the HPHF page on this website, for more details on HPHF Ratio Agreements.
Net, be cautious when relying on correlations over a sample with outlier moves, and consider all of the tools at your disposal, when looking to hedge a regional exposure to home prices. There are few ways to hedge home price risk in many MSAs, but Ratio Agreements might be the tool to more completely hedge local exposures.
Feel free to contact me if you'd like to discuss this blog, have particular exposures that you're looking to hedge, or just want to learn more about the use of home price index derivatives in hedging strategies.
Thanks, John