Home Envy?

In a survey Robert Frank finds that people say they would rather live in a 3000 square foot home when their neighbors have 2,000-square-foot bungalows than live in a 4000 square foot home in a neighborhood of McMansions.  Greg Mankiw asks:

Do people really behave as reflected in this survey? I bet the 4,000 square foot
house surrounded by McMansions would sell for more than the 3,000 square foot house
surrounded by bungalows.

Let’s take it to the data.   I regressed the sales price of about 12,000 houses in Northern Virginia on a bunch of housing characteristics including number of bathrooms, bedrooms, levels, age of the house and so forth.  I also included the average sales price of homes in the same neighborhood.  The result?  Houses in neighborhoods with high average prices sell for more than similar houses in neighborhoods with lower average prices.  Thus the prima facie evidence is that the same house is worth more if it is surrounded by more expensive houses – the opposite of Frank’s hypothesis.

Now it could be that the high average price of other homes in the neighborhood is controlling for unobservables of "your" home so here is a more precise test.  I defined a variable (sqft-avgsqft) where sqft is the lot size of your house and avgsqft is the neighborhood average.  I then split this into a positive difference, when your house is bigger than your average neighbor’s house and a negative difference when it is smaller.  Bigger houses ought to sell for more everywhere but if Frank is right then square footage is more valuable when other people’s square footage is low (lording it over your neighbors).  Similarly, if Frank is right people should be especially averse to living in small houses in big neighborhoods thus a negative difference should result in much lower prices.

Below you can find the relevant part of the regression.  The bottom line is that houses with bigger lots sell for more (the positive coefficient on lotsqft) but the increase in price is less when your lot size is bigger than the average lot size.  In other words, people do not want to own the biggest house in the neighborhood.

What about when your house is smaller than average?  Here there is no penalty.  Contra Frank, people do not mind having a small house in a neighborhood of McMansions.

SalesPrice Coef. Robust Std. Err. t P>t [95% Conf. Interval]
pos dif -.4012873 .0992195 -4.04 0.000 -.5957739 -.2068006
neg dif .0352269 .0474728 0.74 0.458 -.0578278 .1282815
lotsqft .760224 .0527708 14.41 0.000 .6567846 .8636635

Although the data are inconsistent with Frank’s argument that the rich make the middle class worse off they are consistent with an alternative status effect in which people dislike lording it over their neighbors or in an alternative interpretation, the poor make the rich worse off. 


Contra Frank's view is the old real estate investor's advice to buy the smallest house on the street, which will offer better returns to rehabbing or expansion.

Perhaps not directly on your point, since the investor probably doesn't much care what the neighbors think. On the other hand, the presence of investors to arbitrage envy-based price differences (if any) might confuse the data.

If I'm interpreting your 'rich' and 'poor' groups correctly, I'd argue that McMansion owners are not rich. In fact, they may be worse off than their bungalow neighbors because of strangulation levels of debt. Certainly they’re more at risk should something happen to any part of the economy.

Additionally, McMansion owners in my neighborhood are mountainously unhappy/tense/hostile, perhaps at the prospect of life-long commitment to a job where they have little if any control and are bound by the choices they’ve made.

No numbers or studies, just experience from living in a rural-turned-prestigious neighborhood.

Frank's and Alex's data can be reconciled if the smallest house in a big-house neighborhood has lower intrinsic value but higher option value, due to the possibility of upgrading it to a McMansion. As Mike-FP says, those houses are often preferred by real estate investors, who may bid up the price of these houses above their value to people who live in the homes.

An easy test for this would be to repeat Alex's experiment, using property tax data that separates land value from building value, since the option value will be reflected in the land value more than in the building value.

OTH, I would think that living in a rich zip code or a prestigious street would have more signalling value than living in a big house on a poor quality street, because people are more likely to learn your street address than your square footage.

The root of the problem may be Frank using survey data on people's opinions, when data are avaiable on people's behaviour.

As I understand it, economists are usually more interested by what people do rather than what they say.

Talk is cheap, actions speak louder than words.

This is interesting but did you control for proximity to downtown? We are more likely to see heterogeniety in housing size in older neighborhoods close to downtown (compare Arlington to Centerville). Since being close downtown is all-important in determining house value, we might see a spurious correlation between small homes next to larger homes and larger than average home prices.

Conchis' point above is critical to the interpretation of the results. A substantial body evidence from behavioral economists indicates that people do behave myopically. For instance, when households migrate to a new region they choose commuting distances similar to those of the regions they depart, and subsequently move to be more like the average commuting distance in their new region.

If households behave myopically, we wouldn't expect Frank's effect to be found in housing prices.

Bruce Charlton is almost certainly correct- pay less attention to what people say, and more attention to what people do.


My hypothesis for why some impute such motives is their desire to justify action to cut the wealthier down to size.

There's an old saw in real estate that is relevant here: always buy the worst house on the best block rather than the best house on the worst block. Not only is this useful investment advice but in my experience living in the best house on the block (as you say, lording it over the neighbors) is not so great. I had this situation for awhile in LA (priv home on a block of mostly rentals) and there are many unpleasant assymetries. There are good reasons the value of such homes is constrained by their surroundings, which brings us to another old saying: location, location, location.

"Conrol for school quality? Apparently, who your child's classmates are makes a pretty big difference in how they behave."

I was going to point out something similar: it's possible that, although people in the abstract prefer the bigger house surrounded by smaller ones, the smaller house in the big neighborhood might have other things going for it, such as better schools or a safer feel. If you could hold all other factors constant, perhaps people would behave as they told Frank they would, but in the real world this never happens.

Not to say that Alex's hypothesis isn't correct, as it might be, but other issues might also be involved.

The median household income is about $50,000 a year so almost half the people in the survey can not afford to buy a house and few could afford a 6000 square foot house which would probably take 10 times their median income. It has long been know from empirical studies that people prefer neighborhoods with a median income slightly above their own but there is no data about how they value neighborhoods where the median income is greater by a factor of 10 or more. They are not in the market in such neighborhoods any more that they are in the market for $100,000 diamonds or first class airlines tickets. You can believe the survey or not but it is all there is.

If it were school quality we would expect that small houses in neigborhoods with big homes would sell for more than expected. That is not what we are finding.

Note also that I control for zip code.


Controlling for zip code should effectively eliminate the proximity to downtown effect and the school district effect. However, I would recommend as a test that you run separate regressions in each zip code just to verify that your model is robust to regional effects. If you get basically the same results in each zip code then your model is probably fine. If the coefficients are radically different from zip code to zip code then something is misspecified.

The effect noted by Frank is well known in the psychology literature since the 70's. People have cognitive biases associated with groups and form group associations very easily. Tajfel showed people would exhibit this behavior when assigned to groups that they know are assigned at random (i.e. coin flips), and it is much stronger when there is a reason such as ethnicity or religious beliefs. I think that all the survey shows is that people don't like their neighbors and don't consider them part of their "in-group".

John Dewey of Dallas:

I believe your experience is a Texan cultural influence, where southern manners enjoy a priority over in-your-face truth. I have a degree from UT-Austin and now live in the northeast.

The amount of debt vs. income for McMansion owners vs. bungalow owners would be thoroughly interesting to see. Maybe by region and for the whole country.

My comments were based on my own experience, and the way I know my neighbors are tense and overwhelmed is by the increase in hostile driving, and screaming emanating from several houses following the rural-to-prestigious development.

Also, there's another McMansion development not far away containing inhabitants with which others I've talked with have had horrendous experiences. It's a pattern. Also, these houses sell for >1 million.

jane2: "I believe your experience is a Texan cultural influence, where southern manners enjoy a priority over in-your-face truth."

Jane, I think the southern manners explanation is somewhat of a caricature. In the first place, I've also lived in Philadelphia, St. Louis, and Sacramento. I worked one summer on Wall Street in New York. People are pretty much the same everywhere.

New York folks seemed a bit more rushed when commuting. But in the workplace and in social settings they were very nice, decent people. I'd hardly describe them as tense.

Folks on my block in my Dallas suburb moved here directly from California, from New York, from Michigan, and from Kansas as well as from other parts of the South. There is very little difference in their manners or civility.

The idea that 2000 sq.ft home is a bungalow says a lot about the discussion. That notion adds to the angst of those who "feel" left out. Maybe Frank could use the tax increase to subsidize therapy sessions for those who covet thy neighbor's Mcmansion.

I tend to believe that people actually do act according to this survey, at least most of them. Your social status is dictated by your means of comparing. If you are the only one with a 3000 square feet house and a pool filter b in your neighborhood, it's there where your social status is the highest.

Having the best is something that many people want nowadays, but the truth is that the best of 4000 compared to the best of 3000 is not that different. It is just a matter of mentality. I was just browsing the Gainesville Homes For Sale website and was asking myself if I need house with a swimming pool or not. Some people ask themselves if they want a 4000 or 3000 squared foot property.

Comments for this post are closed