Measuring home value appreciation: the method matters

• When Tomo Economics reports on housing markets, we use a blended measure of the Zillow Home Value Index and the FHFA Home Price Index to balance the strength and weaknesses of the methodologies behind two of the most popular and readily available home price indices.
• Year-over-year home price growth in September 2021 varies from 19.5% (Case-Shiller HPI) to 17.7% (FHFA) depending on home price index used.
• There are some interested timing differences in the AVM powered ZHVI versus repeat sale home price indices. The repeat sales indices reached peak home value appreciation earlier and so started to slow down earlier. The FHFA HPI peaked at an annualized monthly appreciation of 25.2% in April, the S&P/Case-Shiller Index at 23.9% in May, and Zillow’s ZHVI at 27% two months after than in July.
• The S&P/Case-Shiller index, intended to capture the evolving value of the housing market taken as a whole like a portfolio, experienced by far the most aggressive growth in the beginning of the pandemic due to its greater focus on higher end homes.
• The FHFA uses stock-weighing to better incorporate and focus on lower valued housing stock.
• The Zillow Home Value Index, powered by AVMs can pick up housing price signals in areas with newer stock and lower transaction volume by not needing two sales on every property in the usable sample, but value weighted obscures its use for the typical homeowner.

Median sales prices will jump around depending on the changing mix of homes — big, little, fancy, modest — that sell in one period versus another. So they can tell you what they are — the typical price of homes that sold — but changes to median sales prices don’t mean very much for any one home. So you have to use an index to isolate the market pressure (or slack) on prices from the influences of an ever-changing inventory mix.

When I report on housing markets I use a blended measure of the Zillow Home Value Index and the FHFA Home Price Index (that I first dollar denominate¹) to balance the strength and weaknesses of the methodologies behind the two most popular and readily available home price indices.

Indices can be designed to capture the price of the market taken as a whole by using value-weighting to focus on higher end homes that take up more of the real estate asset portfolio. Others use stock-weighting to include lower end homes and better capture the experience of middle-class Americans, rather than wealthy.

The FHFA index is stock-weighted, and so better for my focus on consumer experience, but repeat sales indices have some disadvantages.

Indices built using AVMs use rich data sourcing to elicit signals without requiring homes to sell twice, a boon to capturing price changes among newer homes and where transaction volume is low. It’s also handy if the AVM can pick up on quality changes a repeat sales pair would erroneously pick up as market dynamics.

Unfortunately the computation load is huge for AVM histories, so the ZHVI, previously an elegant median AVM across a steady set of homes, is now a value-weighted repeat AVM index.

The fix is fairly elegant too: whereas you once had to re-estimate 20+ years of AVM history with the new code to make sure that changes from month-to-month are market price dynamics and not your tweaking, a repeat AVM means you just need to re-estimate the previous month and use the change to chain it forward into the future.

This should work out great to no real interpretation change from the moving median except for the switch to value weighting at the same time. To better focus on homes for the typical buyer in their reporting, they narrow the sample to the middle third.

There are some interested timing differences in the AVM powered ZHVI versus repeat sale home price indices. The repeat sales indices reached peak home value appreciation earlier and so started to slow down earlier. The FHFA HPI peaked at an annualized monthly appreciation of 25.2% in April, the S&P/Case-Shiller Index at 23.9% in May, and Zillow’s ZHVI at 27% two months after than in July.

[1] I dollar denominate the FHFA HPI by pegging the value on 2017–06–30 to the median reported home value from the 2019 5-year American Community Survey (ACS).

S&P/Case-Shiller Home Price Index, released at a 7–8 week lag, captures the evolving value of the housing stock taken as a whole portfolio. It is a value-weighted repeat sales price index that covers the full spectrum of home values (excluding outliers, I’m sure). Because home prices have a long tail (the top 10% of homes are significantly more expensive than the typical home), it’s going to more heavily reflect the experience of higher end homes.

FHFA Home Price Index, released at a 7–8 week lag, is an equal or stocked-weighted repeat sale price and refinance appraisal index relying on Fannie and Freddie data on conforming mortgages rather than transaction records from county offices. These differences, and the history back to the 1970s, make the FHFA HPI an attractive alternative to the popular ZHVI.

Zillow Home Value Index, released at a 3–4 week lag, is a value-weighted repeat AVM index — it uses pairs of home value estimates from an AVM from one month to another rather than pairs of sales. This helps with sample size and captures more of the housing stock (anywhere with quality AVMs). They try to capture the value of a “typical” home by limiting their sample to just the middle third of each metro by home value, but value-weighting remains. Released at a 3–4 week lag.