I’ve compiled a table of home price changes over the last two large downturns (the post S&L crises, and Great Recession), and the current cycle, that should be of interest to people trading home price indices today. I’ve also posed some questions (below) that people trading CME futures, or academics looking for ideas on research papers, might consider. Now, I’m not saying that home prices are headed lower (although several contracts are inverted, i.e. with lower forward prices than spot), but if the market does turn lower, past home price declines might shed some light on how the next downturn might play out.
The table shows the Case Shiller SA (seasonally-adjusted) numbers for three cycles: 1987-1993, 2000-2012, and 2011 to today.^1 (Note that the dates are the end of the reference period, not the month the index was released. As such, the most recent month is October 2018. Also, note-I realize that the font size in the table is small. A copy of the table has been posted in the Reports section, which can be accessed here.).
The table is divided into two sections: at the top are the eleven areas that have contracts that are traded on the CME (the 10-city index and the ten component regions), and then, below them, the ten other areas that make up the Case Shiller 20-city index (that would have to be traded OTC). I’ve shown the minimum, maximum and subsequent minimum value for each index over each cycle. I’ve then calculated the percentage gain during the rally, and the subsequent percentage loss on the declines.
I’ve also shown the length of each up and down cycle denominated in months. Note that several CS indices (in the “other 10”) were not introduced until after Feb. 1987, so the percentage gains, as well as the length of the rally, may not reflect home price moves over a cycle comparable to other indices. I’ve italicized those results as the CS time series didn’t all start in Feb 1987. Note also, that that there wasn’t a CS index for Dallas for the first cycle.
I’ve further highlighted the top four largest percentage price gains and declines after the peak of each cycle (with the biggest gains in green and the biggest declines in red). (Note that only 3 of the 21 indices have a peak date other than the most recent coverage period, i.e. Oct 2018).
I’ve included a column (in grey) that shows the % difference between the peak value in the current cycle, with the highest value in the 2000-2012 series.
Finally, I’ve added (in yellow to the right) the mid-market value of the CME X20 (Nov 2020) contract vs. the spot index value. Here I’ve highlighted the two contracts that are quoted at the highest (percentage) value versus spot (so, BOS, LAV, with DEN close behind) and the two contracts with the lowest value versus spot (so SDG and SFR, with CHI and LAX just below). Note that the SDG and SFR contracts are down even more from the peak in prices in this cycle, as spot levels are already below peak prices.)
What to make of all of this? I’ll offer some points here, and more in follow-up blogs, but would be thrilled if others weigh in with comments to my email address (which I might incorporate into future blogs).
- The 1987-1993 home price downturn did not impact every region. That is, many regions (at least 9 of 21) saw prices increase right up through Jan. 1993 (the end of my measurement period). That prompts the question as to whether a future downturn will be large enough to hit all areas (as it did in 2006-2012), or small enough to only hit selected areas that may suffer from local issues. Incumbent in that question is whether there are large forces in place that will drag all markets down (e.g. a spike in interest rates), or only certain markets for selected reasons (e.g. reduction in deductibility of SALT), population shifts, decline in local wealth (e.g. Silicon Valley), and/or affordability. Will fear of seeing price declines in one area, prompt users to hedge in their area? Net, there should be lots of opportunities to debate relative forward outlooks via inter-city spread trades.
- The peak in prices in the first cycle (of those that saw a decline by Jan 1993) varied from May 1988 (in New York) to Sept 1991 (in Las Vegas). The same was true for the second cycle where BOS peaked in Nov 2005, but Charlotte didn’t top out until Aug 2007. As such, one question (related to the above) is whether the declines that we’ve started to observe could be unique to those areas, or precursors to drops in other areas .
- For those areas that did see downturns, the period until prices hit bottom was drawn out (in both cycles). Bottoms were even more spread out across regions with Denver hitting bottom in Nov 2009 (after selling off <12%) while 3 areas waiting until March 2012. Downturns averaged about 2/3rds of the time measured (since Jan 2000) but the rallies would be even longer (and bigger) from the absolute bottom in prices. (I may go back to start middle series at past low in next iteration). These points might impact shapes of forward curves on CME futures. That is, if there are price declines in 2019 (and if they are relatively small) when might we see prices later pick up. This may play out in calendar spreads for 2021-23.
- Those areas that saw the greatest increase in prices (in 2006-2012), also saw the greatest collapse.^2
- Those that saw the greatest decline in the 2000-2012 period have been some of the best performers this cycle (going from just being oversold, or something else?). Examples include Las Vegas (+109.6%), Miami (71.8%), Phoenix (87.3%). What factors drove those markets higher in the last cycle, and how likely will those factors be replicated in this cycle (even assuming that a decline has begun).
- While some areas are up sharply off the prior lows, they have not gone far past the prior peak For example, Tampa and Los Angeles are up ~70-75% from the lows, but below prior peaks. By contrast, Chicago, Las Vegas, Miami and Phoenix are still below past highs, while Denver and Dallas are up about 50%. How much should gains over past highs factor into forecasting possible future selloffs? Will homeowners in the area with large gains be more prone to hedge (and less concerned about execution price) than those who’ve seen much smaller gains this cycle (e.g. New York, Cleveland).
- The West Coast includes three cities that have already come off their highs in this cycle: San Diego, San Francisco, and Seattle, as well as the regions with some of the largest gains (e.g. Los Angeles +~75%, San Diego +~70%, San Francisco +~109%, Portland +77.3%, and Seattle +91.8%. Was there a China/Asian effect in play (particularly once Vancouver started taxing foreigners?) Will these areas be disproportionately impacted if the Chinese economy slows? What other factors may have driven California home prices to such highs (and are those reversible)? What should be the impact on volatility assumptions for options -both outright and relative to other regions?
- Some areas that came through the 1987-1993 cycle unscathed (possibly due to population inflows?) -e.g. Miami, Las Vegas, Phoenix – were the worst performers in the Great Recession cycle. Some have argued that these were areas where sub-prime lending had a large effect. Without subprime lending today (but not ignoring FHA’s role) are these areas likely to perform better this time?
All of the above topics are of interest to me in quoting CME futures and OTC trades.
Trading in CME futures has historically picked up when markets turn (e.g. 2006-2007 and 2012) so I’m hopeful that we’ll see it this cycle. I can’t think of a better public, pure-play for financially expressing a view on forward home prices, than the CME futures and options contracts. That said, given the curve inversions, and volatility of certain regions (most notably on the West Coast) I expect current traders (including me) to be more defensive as the current situation evolves. This market will need more involvement – particularly from the long side – to avoid becoming overly one-sided (e.g. all hedgers looking to sell at the same time). CME prices are currently consistent with HPA gains several percentage points below those expressed in surveys and home price forecasts. There may be an opportunity for those with such views (and their readers) to add home price exposure, at what they might argue, are attractive levels.
Please feel free to contact me (firstname.lastname@example.org) if you’d like to discuss any aspect of this blog, or have any questions on hedging home price indices.
^1 I don’t often used the SA series, but I wanted to pick out highs and lows, and was worried that seasonal factors might hide a key intra-year value.
^2 Note issues with calculating percentages, e.g. LAV went from 100.4 to 235.76, for a 135% gain, but then fell through the starting point to 90.52 for “only” a 61.6% loss.