Making markets seem thicker

Geoffrey Manne —  9 February 2006

zillowThe Internet (read: inexpensive information dissemination) comes to the notoriously informationally-challenged housing market. The WSJ reports on a new website,, which, as the WSJ says,

uses data such as tax records, sales history and the actual prices of “comparables” — homes in your area that are similar to yours — to come up with an estimate, which it calls a “Zestimate” [of a house’s market value]. It backs up the estimate with lavish data — aerial photos and maps showing prices in a neighborhood; loads of charts and graphs displaying historical data and price movements, as well as details on the size and room totals of a home. It even allows you to enter information, like the types and prices of recent renovations, that might change an estimate.

I know the site works and works well because it values my house at a level far higher than I would have expected.

Prof. B worries about the privacy implications. I guess. Just like it’s worrisome that my neighbors know exactly how much I paid for that bag of apples I brought home from the store the other day by going to the store and looking at the price. (I know, I know: “house” does not equal “bag of apples”).

Anyway, I think whatever the costs to privacy, the benefits are potentially substantial.  Housing markets are extremely thin and information is scarce.  The ratio of price to purchaser income is relatively large, repeat players are few and far between (except realtors, and we all know how helpful they are), and values are idiosyncratic. This site, if reliable, will provide some corrective and inject more information than one would expect from such illiquid markets.

Geoffrey Manne


President & Founder, International Center for Law & Economics

12 responses to Making markets seem thicker


    There’s some fun stuff on zillow. Enter an address. Go to update the zestimator. Change the number of bedrooms (add one) and said house is often worth less! Change the date the house was built (from, say 1960 to 2000) and the house loses 200,000 or so in value! Oh, must be that “classic construction.” I don’t mind internet hucksters, but the current zillow is the Segway of the net. Information has no value if it is not accurate.


    I found it quite inaccurate. Zillow valued my house at about 60% of what I was offered (unsolicited) last month.

    I also notice that the houses sold most recently had the highest values, and those that hadn’t sold in the last ten or so years had the lowest.

    For what it’s worth.


    There’s another use of Zillow’s approach which hasn’t been discussed here–I don’t know if they promote it or not.

    When you are house shopping, the realtor typically asks you how much you want to spend. That’s a sensible question for the realtor to ask, given that he wants to maximize your expenditure and so his commission, but not a sensible one for you to answer, since how much you want to spend depends on what you can get for additional money. And, if you tell the realtor the maximum price you can pay, he has an incentive to preferentially show you houses at near that price–even if a less expensive house would be, on net, a better deal.

    Suppose you ask Zillow how much a three bedroom two bathroom house with 2400 square feet will cost in the area you are interested in. Then you ask how much it costs with three bathrooms. With 2800 square feet. With four bedrooms. You could estimate the marginal cost of an additional bathroom or bedroom, decide if it’s worth it, and thus get a sketch of what sort of house you are looking for.


    Thanks Josh – unfortunately, I don’t want to move and am too conservative and lack the finance sophistication to try and utilize it in more agressive ways. After all, I typed logorithm when I meant algorithm.

    I presume that when many people use terms like “inaccurate” and “overvalued” they are simply failing to articulate the methodological flaws that have led them to that conclusion. They probably have a fairly detailed knowledge of the various characteristics of homes (and their comparability – including railroad proximity in my case) that have recently cleared the market (for instance, the value given by Zillow for my own home last June compared to the amount at which it actually cleared as well as homes not even listed in their computing) and just use less precise language. I’m also skeptical of the hockey-stick look of the graph and assigning an increase over the last week.

    Geoff also points to a different (I take it) question – will Zillow have a shaping effect on the market for homes, i.e., encouraging people to pay more (or less) than they would have without the Zestimate.


    In case anyone missed it, Josh makes a great point that a few of the comments here and at Prof B.’s overlook: Terms like “inaccurate,” “overvalued,” etc. that aren’t grounded on possible methodological flaws (as Mike suggests may be a problem here) are pretty meaningless. Why is your sense of the “value” of your house any more likely “correct” than Zillow’s? By which I mean, if we’re trying to price a house for sale, your idiosyncratic valuation of it may not be any more “accurate” than a (well-functioning) algorithmic estimation. So, as Josh says — enjoy your newfound wealth! You may think you don’t deserve it, but if enough potential buyers look up your house on Zillow, you may get it anyway.


    Randy, congrats on your newfound wealth!

    Michael Guttentag 10 February 2006 at 10:14 am

    You can see posts on Steve Bainbridge’s site that many of us found the algorithm they are using to find comparables pretty unimpressive.


    In my town, it was wildly inaccurate. I just typed in my own address (and I bought last May) and it was completely overvalued. In fact, the value they assigned it last May was approximately 20% more than I paid for it. In addition, the value graph had skyrocketed in the last 2 months (almost 30% over the November value with a gain of $34,507 in the last week – sweet). I think the new ‘window treatments’ in the master look great and am quite pleased that my wife has wrapped up painting my son’s room, but not $30,000 worth. Granted I know that my home has an important ‘incurable defect’ that may not be revealed in their logarithm, but they still missed by a mile.


    GMUSL2L: I gave exactly that admonition! These descriptors are usually frequently misused by lawyers and law and econ types, i.e. calling a market “thin” (without much thought as to why this is or what this means) when there is some “sense” that the market is not operating in the way one might expect is sufficient to conclude that markets wont clear, and some sort of fix is necessary. You will be happy to know that my co-blogger Geoff knows what the term means when he uses it.

    I found this site extremely interesting! The site was not tremendously accurate with prices in neighborhoods I know well in Arlington and San Diego. An accurate site performing this function is an obviously good thing. I’m not quite sure I understand the privacy objection.


    I’m wondering how accurate this thing is. I, too, experimented with it a bit yesterday. It valued my old Chicago condo fairly accurately, but my neighbors’ homes in Columbia, Missouri were greatly undervalued. (I know because I’ve been house-shopping in the neighborhood.) I’d be interested in what others think. In general, though, I wholly agree with Geoff that this is (or at least has potential to be) a pretty great development.

    Given the growth of discount brokerages and the development of sites like this and, I wouldn’t want to be a realtor these days. But maybe if technology renders realtors obsolete, the government will sponsor another one of its successful job-training programs.


    I’m sorry, I’m just amused by one of Prof. Wright’s co-bloggers speaking of “thin” and “thick” markets after he admonished in Contracts II last year to NEVER EVER do such a thing.

Trackbacks and Pingbacks:

  1. TRUTH ON THE MARKET » Please Don’t Regulate Zillowtalk - October 31, 2006

    […] Several months ago, Geoff posted about, a website purporting to provide “Free, Instant Valuations and Data for 67,000,000+ Homes (…and you don’t have to enter any personal info and no one will contact you).” Several of us played around on it a bit and concluded that it’s not all that accurate at estimating willingness-to-pay, but it’s fun and at least somewhat helpful. […]