patrick.net

EastCoastBubbleBoy says

Also, I’d rather put in a town or city, rather than a specific address, complete with house #, street, and zip. Often I look in generality, rather than specifics.
Just my two cents… but I do like the concept.

I missed the significance of this part of your post before. The problem with doing a more general calculation based on a city or town is that there is too much variability (in incomes and property types) in larger geographic areas. You couldn't say for example "Oh, the MSV in this town is $100,000 and there is a house for sale there for $50000. What a great investment!" You would probably find that the house is not in the best neighborhood.

Your breakdown of median incomes is based on the 2000 census? It might be hard to convince people to use numbers that are almost 10 years old. I assume there will be updates based on the 2010 census.

It is a useful tool for buyers. I don't see the same problem as Just me and Nomograph. In a working free market, buyers and sellers negotiate price. Sellers like to use comparables to set their asking prices. Buyers should be using either area incomes and/or sales history of a given house to determine their offering price. Makes perfect sense for a buyer to tell a seller, "Wait a minute, you're asking price is out of touch with local incomes. Come down or no deal."

@ Ryan
Yeah, I kind of expected the age of the income data to be the biggest objection. The 'good' news is, incomes for almost every group has remained basically flat (except for high level executives of course - but they can take care of themselves).
Assuming I can get the site to take off, I will be updating the data with the 2010 census as soon as it's available. I'm also looking at integrating other data sources in the mean time.

My biggest worry though is just making sure people understand the basic concept - especially casual users of the site that haven't made decoding the housing crisis their personal hobby.

While I am a housing bear and think housing is generally overpriced in my area, this comes up with values waaay lower than even I think they should be. I checked against a house I bought and sold between 2000 and 2005 and the calculation was extremely low relative to reality.

Also get this all the time:

Zestimate: Zillow Did Not Find a Match
Fatal error: Call to a member function getChart() on a non-object in /home/sbradt/public_html/content/plugins/php-execution/includes/class.php_execution.php(273) : eval()'d code on line 173

@ pkowen
Without knowing the specific address you searched, I can't really comment on why the numbers are so low in relation to 'reality.' It could be that this is an area where the data needs to be looked at more closely in which case you would have to run a custom MSV, or it could be that prices are just that inflated (or of course, you could just conclude that the whole concept wacko).

The errors you got mean that Zillow.com did not have the address in their database (although the second error you got shouldn't have looked like that - I've got to correct something in my code).

Are you sure that you entered the address correctly? Usually when I get an error like that from Zillow I find that I've entered the address wrong.

Hey, this is a nice tool, but one thing I believe you are failing to take into account is the difference in quality or even quantity of the square footage of a house.

So, I keyed in a house in my area that Zillow values for about $450K, and you came back with median income $81K and the house has MSV of $249K. That's about 3 times median income, so what you would expect the area to support.

What I think you are failing to take into account is that the house could be a 1 bedroom, 100 year old hole in the wall, or it could be a 5000 sq. ft. new construction McMansion. Just like median income doesn't mean everyone in the area makes $81K, MSV for every house won't be the same.

However, your tool is a good way to check and see if prices are generally in line with the median incomes in an area, which is always a good check to see if you're overpaying for the neighborhood or house you're purchasing.

I think the only way to make your tool more accurate is to come up with a model that includes price per sq. ft for 1, 2, 3, and 4+ bedrooms, and combines that with median income to determine a more accurate MSV for each individual property.

I hope this input is helpful. It's a great start, but it needs to have a more complex equation behind it to truly provide insight beyond a simple census check.

@ lyoungblood

"However, your tool is a good way to check and see if prices are generally in line with the median incomes in an area, which is always a good check to see if you’re overpaying for the neighborhood or house you’re purchasing."

You've got it! The MSV is not the market value. It's a 'check' on the market value.

One of the assumptions is that most of the properties in the area will be similar (in square footage etc). If that's not the case, then you have to break down the data further - that's what the income distribution and property type tables are for - and run a Custom MSV.

Hopefully, this can explain of this a little more clearly ... Evaluating MSV Accuracy

I'll be away from the computer for a couple of hours.

Added a zip code look-up link to the results page

That is a pretty cool site. I like the explation of income leveraging by lenders. I've never seen it broken down like that before.
Once I got my head around the value tool, I liked that too. I think getting people to understand it will be your biggest problem though.

@ will

Thanks.
As far as getting people to understand it, I agree with you. I don't think the concept is difficult to understand, I just think it's not what people expect. People assume that it's telling them the market value is $x, when it's just telling them a value more than $x can't be sustained given the income for the area.