Hi! I’m Aziz. My newsletter contextualizes America’s housing market using data visualization—hit reply and let me know what you think.

For this article, I teamed up with Mike DelPrete. Mike is a global real estate tech strategist, and a scholar-in-residence at the University of Colorado Boulder.

Today’s collaboration with Mike came out of our shared frustration with the way real estate data is described by major media outlets.

To be fair, seasonality and other factors make describing the data in a simple way pretty difficult.

Take yesterday’s data, from the National Association of Realtors’ (NAR) existing home sales report—how can we describe a series that displays the following characteristics?

Seasonality

A lot of economic data is seasonal, but especially so in real estate (people tend to want to move in the summertime, when the kids are out of school). The NAR bases its seasonal adjustments on the “X-13” methodology developed by the Census Bureau.

But, especially after major shocks, accurately adjusting seasonal factors is extremely challenging. As the NAR writes, “for everyday practitioners, simple raw counts of home sales are often more meaningful than the seasonally adjusted figures.” For all but the final analysis here, we use unadjusted data.

Regional variation

There is regional variation in all economic data, but in many cases it’s not that relevant. By contrast, as they say, “all real estate is local”. There are risks of taking this too far—Bernanke underestimated the GFC for this reason (“we've never had a decline in house prices on a nationwide basis”).

Still, understanding the real estate market entails grasping both the macro drivers that push prices across the country up or down, as well as the regional idiosyncrasies that help determine price moves in each region.

Both cyclical and secular trends

Some data—like GDP growth—is cyclical. Other series rise consistently over time (price indices, for example).

Home prices are a function of both the economic cycle and long run (“secular”) trends. If we were building enough housing to meet demand, it’s not clear there would be much a secular trend to real home prices. But, at least over the last 50 years, prices have been steadily rising.

Disentangling the long run trend from the ups and downs of the economic cycle is not easy.

Shocks

These long stretches of rising prices have been interrupted by shocks—in the late ‘70s, late ‘80s, the mid ‘00s, and during the pandemic. When the market does turn down, it helps to be able to keep an eye on the bigger picture: how does this downturn compare to prior bouts of weakness?

Mike and I haven’t solved these challenges.

But, we’ve come up with a series of graphics that we hope clarify the existing home sales data by addressing its nuances in a straightforward way. As Mike wrote over at his site, “ultimately, data should tell a story and provide meaning.”

National homes sales are far below average

401,000 homes were sold across the country in August. We think the best way to contextualize the data is to compare these figures to the same month in prior years. Sales in August (the latest data) were 24% below the historical average for that month, and the lowest since August 2010.

Real estate is local: the West is the weakest

Sales in the West have declined 38% against the long-run average—more than in other regions (though the Northeast is not far behind). Transactions are holding up better in the Midwest and in the South.

Accounting for seasonality, sales are rising very slowly

Momentum in home sales over the past six months has been sluggish, and well below historical averages.

The monthly trend is not much better

Sales are consistently falling below the historical average, especially during the past five months—although August was a slight improvement over July.

Historical context: beware the “dead cat bounce”

Looking at the three major historical market downturns before this one, the average decline in sales (based on seasonally adjusted sales at an annualized pace) from peak to trough has been just over 40%—the same as this one, so far. The average duration of prior downturns has been 50 months—more than a year longer than this one.

What emerges from a visualization of prior downturns is that there are frequent “dead cat bounces”—recoveries in sales that are not sustained. The current bout of market weakness, which started in late 2020, has already had two recovery phases that quickly fizzled out.