Saturday, April 29, 2006

Today's junk stat: Realtors make you $31,000

Today's junk stat comes to you from the National Association of Realtors blog http://narblog1.realtors.org/mvtype/narinthenews/ , which reports that "the average seller who uses a real estate professional makes 16 percent more on the sale of their home than do sellers who go it alone. That's an average of $31,800 per home."

Of course, I don't usually hang out at the NAR site. The Freakonomics blog has a blurb about this http://www.freakonomics.com/blog/2006/04/27/realtors-get-a-blog/ . The bloggers note that this figure is mysteriously high.

So where does it come from? "Leonard", A realtor who talked to the author of the study noted "I talked to the author of this study at the National Association of Realtors and it was based on a survey return from 155,000 mailings. The returns totalled 7400. The criteria was a sold, single famiy,detached home in the suburbs." 

So there's massive selection bias likely at work: the survey return is about 5%, people with very expensive homes might find it less than exciting to sell the home themselves and be more likely to hire a realtor, etc. But selection bias is expensive to eliminate, and almost certainly would lead to a less impressive headline than "The Cost of Selling without a REALTOR(trade mark registered): $31,800". So, if you are the NAR, why spend money on good statistics when there's better bang from cheaper bad statistics?

My own comment is that a good study would seem easy to do:

Wouldn’t a real study be stunningly easy to do? (1) there is usually an assessed value on the tax rolls, which are public information so people can compare assessments. Do this within an appropriate governmental unit so systematic variations in assessment don’t apply (2) house is sold either by a realto or not, which may be part of the publically available closing info (or may not, but would be easy to get by phone survey). One covariance analysis later and you’re done, and published!

Replicate this for 3115 US counties, publish each one, and sit back and enjoy that tenure—maybe even a chair endowed by a grateful real estate industry![*]

In fact, such a design is SO simple that one suspects that the results don’t provide much support to this.

[*]Some poetic license taken on this.

No comments: