Happy Case-Shiller day! The last Tuesday of the month is always an important one for the housing market: two of the key price indices, the S&P CoreLogic Case-Shiller home price index and the FHFA home price index for January were released this morning¹. These are the gold-standard measures of home prices because they are based on repeat sales.

So what did today's data show us?

This morning's data showed a continued rebound in home prices. The National Case-Shiller index rose by 6.0% from a year earlier, while the FHFA index rose by 6.3% (both non-seasonally-adjusted). This is even faster than the 5% average rate of annual home price appreciation we have seen over the past 50 years.

…but this is a statistical quirk

In fact, the y/y figure disguises much softer home price appreciation than usual.

Here's why: after starting to turn down in June 2022, home prices bottomed in January 2023—exactly one year prior to the data released this morning. As a result, the year-over-year comparison looks unrealistically solid.

I prefer to look at price changes in the non-seasonally adjusted data over the last 3 months, compared to prior years over the same three months². By this metric, home prices are going into the spring with much weaker momentum than usual.

I prefer the non-seasonally adjusted data because the seasonal corrections are questionable.

“…seasonal adjustment adopted by statistical agencies is sometimes seen as a potentially dangerous procedure that may compromise the intrinsic properties of the original series.” —IMF³

This is not the conventional way of looking at the data, but I believe it’s the right way.

3 reasons I use 3-month changes vs history as a gauge of momentum

First, this measurement solves a big problem with y/y changes: a year is a long time. Home prices may be higher than back then, but that tells us little about what's happened in the interim, and in which direction, and how quickly, things are moving these days. The "comps" from 3 months ago are less flattering, and give us a closer to real-time gauge on the direction of prices.

Second, the 3m change vs history measure allows us to gauge home price appreciation in a timely way without picking up too much noise. Prices bounce around from month to month, so by looking at them over a slightly longer horizon, we smooth these bumps out.

Third, it solves the seasonality problem. Seasonal series cannot be measured on a monthly basis. But measuring prices on a 3m basis, and then comparing them to history over that same period, neutralizes the seasonality: we're making a like-for-like comparison.

What about the seasonally-adjusted data?

For those who are skeptical about my aversion to seasonally-adjusted data, the chart below tallies the frequency of various month-over-month changes in the seasonally-adjusted FHFA index.

A 0.1% month-over-month fall in prices is pretty unusual, and corroborates the analysis above: the housing market is entering the busy spring season on a weaker footing than usual.