The Case-Shiller index is the most accurate measure of aggregate home price changes, for reasons long-since described here. But it’s been an ongoing gripe of mine (and everyone’s) that the index lags so badly. The data that was just released a couple days ago on October 27, for instance, only tracks home prices through August.

So lately I’ve taken to using the median price per square foot data to guess, for lack of a better word, what the Case-Shiller index values for more recent months might be. An example of this estimation can be found in the graph below, which appeared in the writeup of the most recent median price data:

Here’s how I arrive at these estimates. (Non-nerds may wish to fall asleep for the remainder of this paragraph). The Case-Shiller index only uses single-family home prices, and in order to increase its accuracy it is calculated based on three months’ worth of data. So I just take a three-month average of the median price per square foot for single family homes. The median-based indicators are less accurate and more volatile than the Case-Shiller index, but it seems like they can at least provide a rough idea of what’s happened price-wise in the last couple of months.

I wanted to get a better read on just how rough an idea we were getting. So I went back to the beginning of 2008 and plotted two numbers for each month: the estimated price change calculated as described above, and the actual price change per that month’s Case-Shiller data.

Now I’m wondering whether I should have invited the non-nerds to remain asleep for the rest of the article.

Anyway, this was the result:

It looks like the proxy did a pretty good job of predicting the Case-Shiller value for each month, except for the six-month period from November 2008 through April 2009, when it did a pretty awful job. What happened there?

It turns out that was right around the time that composition of home sales started tilting overwhelmingly towards low-priced homes. In a November 2008 article I noted that in October, home sales had decreased by 11 percent on an annual basis for the most expensive San Diego zip codes while at the same time they’d increased by a blazing 186 percent in the least expensive zip codes. It makes sense that such a huge shift in preference towards lower-priced properties would result in a lower median price per square foot of homes sold.

This effect couldn’t last forever, though, and it started to recede in 2009. As of an update on the July 2009 data, the gap in year-over-year sales growth had collapsed: sales in the most expensive zip codes were up 8 percent for the year compared to an increase of 21 percent for the cheap zip codes. Sure enough, that is near the time that the estimation technique started to work decently again.

So for the time being, since it’s been behaving itself in recent months and since we have a culprit for past episodes of innacuracy, I am going to give my little Case-Shiller proxy the benefit of the doubt and assume that it’s providing a decent preview of the what the actual Case-Shiller data will bring when it is finally released. Future disparities will be noted and their origins hopefully surmised.


Leave a comment

We expect all commenters to be constructive and civil. We reserve the right to delete comments without explanation. You are welcome to flag comments to us. You are welcome to submit an opinion piece for our editors to review.

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.