I’m was trying to peek around corners a bit with this idea of looking at Zillow’s dataset of percent of homes that are increasing in value.
Zillow doesn’t publish these graphs themselves but they do publish the raw data so I took that data and made these graphs.
Zillow looks at their Zestimates for all homes in a market and then they look at what percentage of their Zestimates are increasing in value.
Obviously, the dataset is very dependent on the accuracy of the Zillow Zestimate and the Zillow Zestimate is very dependent on current home prices. So to some degree, I was trying to use Zestimates to predict prices but the Zestimates are largely based on prices. See the problem?
One good thing about the “Percent of Homes Increasing in Value” is that it moves a lot. A small change in home prices can produce a big change in the metric. I was hoping it would make it easier to see coming changes in home prices.
It’s an interesting idea but it’s not working out great.
The Potential
Looking into the past, sometimes the “Percent Homes Increasing in Value” metric will turn 6 to 9 months before it shows up in the Case-Shiller Home Price Index.
For example, the “Percent Homes Increasing in Value” tanked for about 6 to 9 months before home prices tanked in 2006 and 2007.
Just eyeballing the graphs, more often, however, it seems that home prices and “Percent of Homes Increasing in Value” tend to move at the same time.
Sharp Turns
Another problem is that a trend will reverse course on a dime. That may reflect the real market but I suspect it may be caused by something else, like a change in how the Zillow Zestimate is calculated or some other reason.
The head fakes we saw in San Francisco in late 2014 and late 2016 really show the problem.
Were those changes based on changing fundamentals?
Or, perhaps, those changes are just the standard background noise in the dataset. If that’s the case then we would need to see larger changes before thinking there might be a change in the underlying fundamentals.
Revisions
Another problem with the metric is the revisions. You’ll sometimes see pretty significant revisions to data points several months old.
Anyway, after watching this metric off and on for over a year, it’s not that great of a metric.
However, I’m thinking today that when the metric moves into or out of the neutral zone – that middle zone with the white background – that’s a pretty good sign of a change in direction in home prices but the change may – or may not – show up in home prices at the same time.
A weak market like Cleveland, for example, could have strong increases in “Percent of Homes Increasing in Value” without moving prices… at least not yet, anyway.
Future
If you like following “Percent of Homes Increasing in Value,” I’m going to keep publishing it here, at least for a while.
It’s a very interesting dataset but I haven’t figured it out yet.
# # #