Saturday, August 20, 2016

Two Maps I Would Like to See

Suppose you are planning to move–across town or across the country. One consideration in deciding exactly where to move to is the price of housing. With a little effort, you can probably find average house prices in different cities you are thinking of moving to, but that isn't quite the information you want. Low house prices might mean inexpensive houses, but they might also mean small houses in poor condition. What you want is an apples to apples comparison, relative prices for the sort of house you would want to buy.

The data to produce that information almost certainly exist online, since there are extensive databases of webbed real estate listings. Run some regressions on that data and you can use the results to estimate how much the same house costs in different places. The results will also tell you how the price of a house depends on its area, lot size, age, etc. Do it right and the potential buyer can input a description of the house he wants and get estimates of how much it would cost in any of the places he is considering moving to. He can input different house descriptions, compare prices, and use the information to help him decide just how much house he wants to buy and where. The same approach could be used for rental prices. And it could be done not city by city–prices within a single city can vary a lot–but neighborhood by neighborhood. 

What I am imagining is a webbed map. Put in the relevant information about the house or apartment you want, click anywhere on the map, and get a price.

Housing prices are not the only thing you want to know. Another consideration is the crime rate–relevant not only in deciding where to live but where and when to take a walk. The map for that information lets you set a category of crime (burglary, mugging, assault), a time of day, and see a map of the relevant area with crime rates shown by color, running from bright red for the highest to dark blue for the lowest.

I don't know how much of this exists already, but perhaps some of my readers do.


Loquitur Veritatem said...

Although list prices are helpful and plentiful, I would prefer a look-up tool that gives selling price as a function of the variables of interest to me. Two problems: (1) There don't seem to be a lot of databases that include selling price. Homesnap has selling prices for some areas (including the part of Texas in which I live), but it's not nationwide. (2) Even with a fairly large data set for my ZIP code (currently 618 observations), the best linear regression model, with $/SF as dependent variable, includes only 5 rather broad explanatory variables: total square footage, lot size, location east or west of a major highway that divides the city, contract date, and whether the house was built in the 1990s or later. The r-squared of the regression is 0.6 and the SE is $33/SF, with a sample mean of $242/SF.

That leaves out a lot of information of interest to the potential buyer, though it would give a buyer a basis for estimating the actual selling price of houses (in my ZIP code) that he'a already interested in. The dataset would be a lot larger if expanded to include the entire city or county, or adjoining other for which Homesnap has selling prices. But the number of explanatory variables would also have to expand, in order to help a house-hunter search on specific areas, number of bedrooms, distance from downtown, distance to work, and the many other factor that affect one's search for a house and decision whether to make an offer on it. One of those factors is condition (e.g., age of roof, age of paint job, age of HVAC, etc., etc., etc.) -- and I'd be surprised if there's any database (other than a very small, private one) that includes such variables.

Karie, the Regular Guy's Extraordinary Wife said...

Not to mention school information; ratings, programs, title information, etc.

But I'm not sure it would take all of the legwork out of the search. Sometimes you just have to move and rent something until you are sure what you are looking for and what you are willing to negotiate.

Unknown said...

Insurance companies insure homes based on replacement cost (cost to rebuild) so I would research in that direction.

Power Child said... and both feature maps of houses with prices. You can filter by year constructed, square footage, number of bedrooms, lot size, amenities, etc. Even keywords.

These sites also show you school district scores, which are a decent proxy for things like crime rate. (But you can't filter by it, so far as I know.)

I'm also not sure how you'd search by "condition" without proxies. Realty listings don't tend to be very honest about condition, though maybe you could theoretically filter out listings with words like "opportunity".

AbsoluteZero said...


It's not what you're looking for, but for moving/relocation, particularly internationally, these are useful:

Also somewhat off-topic, but for most of the world, "cost of housing" does not mean "house prices".

Unknown said...

A) Sounds like you could start with a 'heat map' in $/sq-ft for an area. I believe zillow has one. Unfortunately you can't limit data is $/sq-ft for 3 and 4 bedroom houses without septic systems.

B) You can also try . It gives heat maps on various metrics for towns like school performance, Housing costs, Crime rates, Taxes, etc. Because it takes time to compile this data and run a web server it is a pay service but useful when you get down to actually house-hunting. Depending on the area you can also get bell-curves for age of housing, cost of housing, racial makeup, religious makeup, average income and many others (all from the latest census.)

Anonymous said...

Hello David! Please check out - they are trying to solve the exact same problem you are talking about. They are backed by a top VC firm Andreesen Horowitz, so I really hope they might be able to solve this problem..

Anonymous said...

Somewhat related: I'd like to see when and where real estate markets are historically "compressed" and where they're "stretched". That is, given an expensive house A and and cheap house B in the same market, is the ratio between A's price and B's price historically small (a good market for "buying up") or historically large (a good market for downsizing)?

Tibor said...
This comment has been removed by the author.
Tibor said...

This probably contains a part of what you'd like to see
It has a lot of other useful information like a comparison of local purchasing power, crime rate, air pollution etc.

Unknown said...

For the other side of the pond you can get a lot of what you want

House prices for last sale are available from the Land Registry copied by real estate sites but often only give house/ flat and number of bedrooms

Ken Arromdee said...

You are not going to get this, because the first time anyone creates such a service, and it is used by customers to avoid high crime areas or areas with poor condition houses, and it so happens that those areas have many minorities living in them, the person providing the service will at least be heavily fined, maybe jailed, for illegal redlining.

Kitty_T said...

I was going to suggest Zillow as well - it allows you to search by a fair number of parameters (price, square footage, # of beds/baths, type of housing), and has other metrics like walkability, school ratings, & price comp to zip code averages (possibly a proxy for "neighborhood crudiness compared to what you want"), though those aren't search parameters it offers. It gives purchase and listing history going back 10 years. It can also show recent sale prices, which pop up on the map next to current asking prices and "zestimates" of the market value of houses not on the market. (The zestimates are ... pretty worthless, actually, other than the entertainment value of wondering what the devil is in the algorithm.)

I found Zillow very useful when looking for houses long distance back in 2010. Particularly helpful in that market was the recent sales data. You could see what other houses in the neighborhood matching your criteria recently sold for, how long they sat on the market, when the asking price dropped and what discount was finally accepted off the asking price. That, coupled with purchase history, was enough that I could guess at what expectations the sellers in an area might have, and also to some extent who was unlikely to accept a lower offer (based on how recently they had radically overpaid, you got a good sense of who was badly underwater). We ended up getting a house at about 15% below the asking price (which was fairly reasonable to begin with), in part because of how that information enabled us to target our search.

Anonymous said...
This comment has been removed by a blog administrator.
Anonymous said...

Another interesting consideration is wage-adjusted COL; if homes are cheap but there are no jobs or wages are low, fat lot of good that is.

Fixr has a simple and pretty interesting tool for this, but it only covers some Texas metro areas and is a few years out of date:

albatross said...


I don't think you're right. Discrimination law applies to sellers and renters and real estate agencies, but I don't see how it would apply to someone offering a website.

Years ago, when I was in the market for a house, my real estate agent told me she was not allowed to discuss the quality of schools or the crime rates of any neighborhood. She then told me exactly how to find that information myself, should I want to see it.