One of the issues related to home prices that COVID may have produced is that users might need to look more closely on reported seasonally adjusted numbers. That is, the lock-downs in the spring, and the waves of summer and fall home sales, may have pushed volume and prices away from historic norms. Net, any seasonal adjustment process that looks to past history will not be able to find a pattern to support this year's activity (so this year's seasonally adjusted numbers might seem off), and this year's activity might be brought forward into future seasonal adjustments.
The graph below shows my calculations of the seasonal adjustment factors on the CHI and MIA Case Shiller indices over 20 years. Note that each pattern (CHI in blue, MIA in red) shows a relative increase in prices in the spring, and lower prices in the winter. This is a long-known feature as families prefer to move between school years, and houses show better (and therefore get better prices) when not buried under two feet of snow. Not surprisingly, given the last point, CHI has more seasonality than MIA.
However, seasonal adjustments have not been constant over time. I wrote long ago about the observed spike in seasonal adjustments as the Financial Crises played out. Net, while homeowners may have preferred to wait to sell until spring, banks who owned large portfolios of defaulted mortgages, seem to have decided to sell throughout the year. Thus the percent of distressed sales became an ever larger percent of overall sales in winter months a) when there were a lot of distressed sales, and b) when overall sales were slow. (Recall that getting mortgage during 2009-12 was difficult, and sellers of non-distressed homes, had to compete with distressed sellers). That is, if distressed sale were 30% of the market in the winter, and if distressed sales traded at a 20% discount (made-up numbers to keep the math easy for illustration), then the index would be pulled down by an increments 6% on such sales.
However, while that might explain the spike in seasonal adjustment factors for 2009-12 it's important to see for how many years seasonal-adjustment factors remained above levels from before the Financial Crises. Net, I believe that those calculating seasonal adjustments look back in time to see past price actions, and incorporate those observations into current adjustments, including the impact of "un-usual years" well after distressed sales abated.
What this might then mean for 2020 numbers is that the current level of summer and fall 2020 activity is being compared to years (before COVID) and so today's seasonally prices are reported as larger than historical. While that is true, my concern is that the increases in prices for July -November for 2020 will be incorporated into seasonal adjustment calculations going forward (possibly to include the most recent year -2020- being given a large weight). That might then have the impact of having next summer and fall's seasonally adjusted numbers (or those in 2022, if COVID shuts down markets next summer) look weak, even though they might be consistent with longer-term trends.
I'm focused on this as even non-seasonally adjusted home prices (to which the CME contracts refer) are a challenge to forecast when historical norms aren't in place. The February 2021 contract will become the front contract tomorrow, and since it covers Oct-Dec, the unusually busy Oct-Nov markets to date, may play havoc on historical Nov/Feb spreads.
Net, use caution, and realize that this year's market will remain atypical in ever-more ways, and may influence how you read seasonally-adjusted results in the years to come.
I'd be happy to post feedback to the blog should anyone have information on how seasonal adjustment factors might address how this COVID-year will be incorporated into futures adjustments.
Please feel free to contact me about this blog, or any aspect of hedging home price risk using home price index derivatives.
Thanks, John
#homeprices #realesate #caseshiller