Do We Overvalue Our Desire to Live Among People of Our Own Race?


A new working paper from a quartet of economists proposes a new method of estimating how we value the “non-marketed amenities” of neighborhood choices such as avoiding pollution and living among people of our own race.  The old static method, they say, underestimates our willingness to pay to avoid air pollution and crime, but overstates how much we value living near neighbors of our own race. Here’s the abstract:

We develop a tractable model of neighborhood choice in a dynamic setting along with a computationally straightforward estimation approach. This approach uses information about neighborhood choices and the timing of moves to recover moving costs and preferences for dynamically-evolving housing and neighborhood attributes. The model and estimator are potentially applicable to the study of a wide range of dynamic phenomena in housing markets and cities. We focus here on estimating the marginal willingness to pay for non-marketed amenities – neighborhood racial composition, air pollution, and violent crime – using rich dynamic data. Consistent with the time-series properties of each amenity, we find that a static demand model understates willingness to pay to avoid pollution and crime but overstates willingness to pay to live near neighbors of one’s own race. These findings have important implications for the class of static housing demand models typically used to value urban amenities.

And here’s the upshot:

The findings from this exercise indicate that the preference estimates derived from our dynamic approach differ substantially from estimates derived from a comparable static demand model. For example, the per-year willingness to pay to avoid a 10-percent increase in the number violent crimes per 100,000 population is $586 (in 2000 dollars), which is about seventy percent higher than the $344 recovered from a comparable static estimation procedure. In the case of air pollution, the corresponding differences are even larger ($296 from the dynamic model versus $73 from the static) though still in the same direction. In contrast, the per-year marginal willingness to pay for race (in particular, the preferences of whites for living in proximity to other whites) is $1,558 whereas the estimate from a naive static model is substantially higher at $1,973. Given the time-series properties of each of these variables, the sign of the bias from ignoring dynamic considerations accords, in each case, with the intuition that the valuations of mean-reverting amenities will be understated while those of positively-persistent amenities will be overstated by a static model.

There’s one big caveat: the data come from San Francisco, which is itself an outlier when it comes to housing markets. It’s probably a safe bet that people in San Francisco don’t necessarily value the avoidance of pollution or living among their own race the same way the rest of the country does. Still, it’s an interesting exercise in understanding our willingness to pay for certain intangibles when shopping for a neighborhood.

Dave C

> For example, the per-year willingness to pay to avoid a
> 10-percent increase in the number violent crimes per
> 100,000 population is $586 (in 2000 dollars)

Would people pay more or less to avoid a 10% increase per 50,000 population.

Does it trend to zero or infinity to avoid a 10% increase per 1 population?

Joshua Northey

What does "using rich dynamic data" mean?

I may misunderstand something but isn't it "preferences for dynamically-evolving housing and PERCEIVED neighborhood attributes". I have known a several people who have purchased homes and a few realtors, none of them were as demographically/economically minded as myself and I would hazard that they collectively know almost nothing about the violent crime rate, pollution levels, or racial composition except on the most general and qualitative of terms. I was able to do some rudimentary research into these types things with a lot of work, and people who knew about it thought it extremely uncommon.

Hell even if you look for the data on those things it is not very helpful. In my major American city the police reports crime data to citizens in very large chunks, maybe 20 or 30 of them, much larger then the granularity of the actual crime.

So is this basically just finding correlations between prices and attributes in a model? Because then the model is not showing what you say it is showing. Almost no one makes real estate decisions thinking about their "he per-year willingness to pay to avoid a 10-percent increase in the number violent crimes per 100,000 population".


Debra McCawley

First let me state that I am white and let me add that I'm disheartened by these models static or dynamic. And further, while these types of studies will show pertinent and factual results, I can't believe that is always so.

Case in point; When I moved to Shreveport LA and was looking for a place to live our options were for the most part limited to all white neighborhoods, all black neighborhoods and only with looking was I able to find a mixed neighborhood. Yes there were other factors involved but, for me the important factors were middle class, single family homes and my preference was for a mixed racial community which seemed hard to find.


Using data from San Francisco would seem to bias the results in another way, as it selects for people who've chosen to live in a dense urban environment.


I am probably mis-interpreting the data... but is that report indirectly stating that people will pay 3 times as much to avoid an equivalent increase in other neighbouring races than they will to avoid an equally proportional violent crime?

That, is, bonkers.

Eric M. Jones.

As a whitebread who was looking for a house in L.A., it came as a shock to discover all-black neighborhoods that were far too pricey for me. I somehow thought that this was impossible.

It further shocked me to discover poor, bullet-marked drug and gang-infested multiracial neighborhoods that I could not possibly afford...even if I had wanted to live in a hell-hole.

This turned my reality upside down.



It would be interesting to see the same theories applied in a red state, or more right-wing area, like, say Texas, or Arizona even, with its anti-immigration stance.

As for me, I grew up in an all white, small New England town. When I first moved to my city home, I lived in an Irish section. Then I lived for 6 years in a Latino neighborhood. Now I live in an area that is mostly black, but becoming gentrified. I really do appreciate the chance to live in a diverse neighborhood. I hear different kinds of music, see different perspectives, learn new cultural words, and basically get to stop living in the "white person" box. It's nice to see what everyone else has to offer, and I've always had really nice neighbors.


oh man....this showed up as "Do we overvalue our desire to live" in my reader. That would have been a way cooler post.


Is there a difference in gender?

Mike B

I think that many people assume that selecting a neighborhood based on race gets the crime and environmental amenities included as part of a combination package. In fact by heaping economic disadvantage, pollution and all sorts of other negative factors on select minority communities, race can become a proxy for everything that people would naturally select for in a neighborhood.

purple black

Preach it my brother.