I like the idea that more people have easier access to the raw data. I wouldn’t get too wound up in the conclusions Zillow draws, especially when you consider how the inclusion of outliers and non-market transactions can corrupt the data pools being used. But if you use a little judgement on which data is valid you can do your own analyses.
I think the easiest way for people to use that data is to print it out and start driving by these sales. In a given neighborhood it won’t take but half a dozen or so to give you a baseline to start forming your own opinions. I am 100% in favor of data availability.
The automated valuation models that Zillow is following are already being used by lenders to make some of their decisions, particularly for their low risk loan underwriting and automated appraisal review functions. As with any type of analysis, the quality of the result depends in large degree to the quantity and quality of the data being used. As those information systems get ahold of better quality data their accuracy rates are going to increase as well. Most appraisers figure it to be a matter of time before AVMs are used more commonly than live appraisals, and that a reduced number of appraisers will be employed doing primarily the most difficult and technical of assignments.
AVMs do have the one stumbling block that going to hold them back for a while – at this point they can only attempt to mimic the subjective judgements that people use to make some of these decisions. It’s not uncommon for a buyer to be unable to articulate exactly what it is about a certain property that is calling their name when they see it. The machine’s got a ways to go before they can capture that pixie dust in an algorithm.