Are Zestimates Really Accurate?

Zillow.com has made a splash with their Zestimate feature, which tells homebuyers and owners what a property is worth, based on data from the local area. Anyone can search for an address and, if it’s their own home, they can “claim” it and refine the data Zillow uses. (I wonder if the Obamas were at all perturbed to find that the White House’s Zestimate declined over 5% last year?)

We all love finding out what a home is worth, and it’s thrilling to watch the number move up and down (we like up much more). It’s like Antiques Roadshow, come home.

An aerial view of some Zestimates in San Francisco. Clicking on each home reveals a few stats, but it's hard to tell how the numbers really compare.

But the big question is, are Zestimates and similar online valuations accurate? Do they serve any real purpose other than sheer entertainment? This is a big question, as buyers sometimes make purchasing decisions based on Zestimates, and even banks have been known to evaluate foreclosures and short sales based on this number.

In seeking the answer, the Appraisal Institute did a study of Arlingon, TX and found that the Zestimates overvalued 40% of homes by more than 10%. Their conclusion (and let’s keep in mind that their livelihood is doing appraisals) is that Zestimates are no more accurate than homeowner’s own estimates. Click through here to read the full study.

Another, smaller, study came to a similar conclusion. The personal finance blogger behind MyMoneyBlog.com tracked his home’s value over a one year period, using several different online valuation tools. Here are the surprising results:

Anarchy! After seeing the graph, he stopped tracking his home’s value, deciding it was a waste of time; he was shocked to see that these tools disagreed with each other not only on value, but on price trends, in nearly every month. (Read more about the project at My Money Blog.)

I’m surprised to see that even overall trending isn’t pinned down with these tools. But I suppose I shouldn’t be surprised, since I can think of several reasons why the tools can’t possibly compare homes apples-to-apples. Some factors that skew data include:

• Short sales and foreclosures. These types of homes are priced lower than regular equity homes, which can skew data.
• Condo amenities. When comparing condos, these tools don’t account for variation like whether parking is included, what floor the condo is on, the view from the condo, whether it has a balcony, and so on. These factors hugely influence the selling price of a condo.
• Condition of the home. Does the foundation need retrofitting? Is the home newly remodeled? Is there mold in the basement? Does the house need a paint job? Is it move-in ready? These factors clearly influence price.

Over time, as Zillow’s algorithm accrues more data, the Zestimates are supposed to become more accurate. I don’t see how they could possibly control for many of the variables I listed, which can impact the price of a home by tens or hundreds of thousands of dollars. Zillow itself concedes that the estimates are not an appraisal, and simply represent a starting point for learning more. Fair enough.

So knock yourself out, but keep in mind that you’re playing a game–not finding indisputable financial truth. Zestimates represent just one of many ways to put together a sense of a home’s value, a number which, ultimately, is proven only by the amount someone is willing to pay for it.