Opinion

Felix Salmon

Why America’s population density is falling

By Felix Salmon
May 4, 2013

I’m not sure why it took me until yesterday to find Paul Krugman’s post from April 16 about population density, where he found a very odd fact buried in a new Census report. We know that the population of the US is rising, and we know that the population of the US is also becoming increasingly urban. As a result, urban density — the number of people per square mile — has to be going up.

And yet, if you calculate density the right way, weighting by population rather than by land area, you find something very odd: density is actually going down.

041613krugman2-blog480.png

In the US as a whole, population-weighted population density fell by 16 people per square mile between 2000 and 2010, while in metropolitan areas it fell by an enormous 405 people per square mile. What could be going on? The best answer, I think, comes from David Schleicher, a George Mason professor who’s an expert on the political economy of urban areas.

If you look at property and land prices in America’s cities, they rose impressively between 2000 and 2010, the property bubble and crash nothwithstanding. Cities are increasingly attractive and expensive places to live; that’s a trend which isn’t going away any time soon. And historically, when urban property values rise, it doesn’t take long for property developers to pounce on the trend. New buildings rise; whole neighborhoods get rezoned. With billions of dollars at stake, politically-connected developers normally find a way to get what they want somehow.

That’s exactly what has happened in, say, Miami, where shiny new condos rise in lockstep with property values. But note something important about Miami: those condos are being bought largely by foreigners, who have little if any political clout in the city. In most US cities, by contrast, rising property values in recent years have meant something different: a rise in the number of politically-powerful groups and individuals moving back into the city from the suburbs.

These rich and powerful have two important effects on urban density. Firstly, they decrease density just by moving to the city: they do that by dint of the fact that they live in larger homes with smaller families. My apartment in New York’s East Village, for instance, is in a 1920s tenement building, which was converted into condos in 1984. During the condo conversion, the old layout, of four apartments per floor, was scrapped in favor of a new layout with only two apartments per floor. But the number of people per apartment didn’t go up. And if the conversion were to take place today, the building would almost certainly be converted into “full-floor luxury residences”, with a keyed elevator opening directly into monster spaces. Again, without any discernible increase in the number of people per apartment.

Rich people like to maximize the amount of space they live in, whether they’re buying suburban McMansions or downtown lofts. As a result, higher property prices in dense urban areas are prone to making those areas less dense — at least until the developers come along.

This is where the second important effect of the rich-and-powerful comes into play. These people tend to fall on the spectrum somewhere between NIMBY (Not In My Back Yard) and BANANA (Build Absolutely Nothing Anywhere Near Anything). Just look at the vitriol hurled by carless Soho residents, for instance, at New York’s new bike-sharing stations. As urban areas become increasingly affluent, filled with wealthy politicians and their wealthier donors, it becomes harder and harder for developers to procure the zoning changes and construction permits they need in order to keep on producing new residential inventory.

The result is that the normal state of affairs — where powerful individuals get trumped by even more powerful construction-industry inevitabilities — is turned on its head, to the point at which new construction can no longer keep up with the de-densification endemic to gentrification. Bloggers may rail against this state of affairs — both Ryan Avent and Matt Yglesias have written at great length about how important it is to allow new buildings to rise within urban areas — but ultimately the natural conservatism of the rich is winning out, across the nation. If you want to move to a city where density is going up rather than down, you might just have to move to Miami. Or China.

Comments
10 comments so far | RSS Comments RSS

What nonsense. Time series data for desirable urban/suburban areas (that would exclude Detroit) shows a relentless increase in density over the past 70 years. 11
housing units per square mile in 1940; 36 in 2010.

Posted by ErikKengaard | Report as abusive
 

Or maybe it’s just sprawl? If the metro areas get larger, but the population doesn’t increase in the expanded areas with the same density as the core, the density of the metro area will decrease. I think that will have a bigger impact on density than foreigners buying homes they rarely spend time in.

Posted by KenG_CA | Report as abusive
 

I’m not sure if effects described by Felix are real (at least where he lives). Census reports that the population within 2 miles of New York City Hall increased by 9.3% between 2010 and 2010.

There is also at least one effect that is likely stronger than those described above. What did happen, really, between 2000 and 2010? We had a housing bubble, and it led to a construction boom. Some of it took place in city cores, but we saw significant growth of suburbs and exurbs, built out with low density housing. As suburbs grew, they soaked up population growth and attracted some people to move from city cores into suburbs.

According to the census bureau, between 2000 and 2010, the number of apartments in massively multi-unit structures (50 or more units per building, city core style) increased by 5%, from 6.13 million to 6.47 million.

At the same time, the number of single-family detached houses in the country increased by 16%, from 69.9 to 81.1 million.

Posted by Nameless | Report as abusive
 

It’s interesting to look at the raw numbers to figure out what exactly is going on:

http://www.census.gov/population/metro/f iles/CBSA%20Report%20Chapter%203%20Data. xls

First of all, metropolitan area population weighted density is in fact down 6% in 10 years.

Next, a mathematical curiosity. Population weighted densities of the two densest metros (New York and Los Angeles) are down 1.4% and 2.6%. Population weighted density of all other metros combined is down 4.9%. So how is it possible that the density of all of them together is down 6%? The answer is that NY and LA have been growing slower than the rest of the country – their populations grew only 3-4%, and the total population of all other metros grew 12%. When low density metros grow faster than high density metros, overall density goes down. It’s a form of Simpson’s paradox.

Once we exclude NY and LA, the rest of the list is pretty noisy, but one can discern two groups. In MSA’s with initial weighted population densities above 1200 people per square mile, median density change is -3.5% (with no apparent trend). In MSA’s below 1200 people per square mile, median density change is +0.8%. Census has a very broad definition of “MSA” (metropolitan statistical area), which, for example, includes Glens Falls, NY (population 129K). Among “real” MSA’s (populations of 1 million or more), median density change is -3.6%, again with no apparent trend.

Posted by Nameless | Report as abusive
 

+1 to Nameless, and I think that Felix misses the point that a decrease in population-weighted density need not mean that density declines in any particular neighborhood. It can occur simply because population grows faster in less dense metros and neighborhoods.

Posted by realist50 | Report as abusive
 

Well sort of. Urban core densities have been falling since at least 1920. Back to the 1800s for Paris and London, for pretty much the same reasons you put.

Posted by EliRabett1 | Report as abusive
 

Nope. Felix and Krugman are both making the same error regarding weighted density. If the rich people condo anecdote were true, that would make actual population density fall in addition to weighted population density. High rise residential development in the US, with the exception of Manhattan, is basically irrelevant to population density trends. What’s happening is that low-density areas are adding population faster than high-density areas.

Consider a metro with three areas: a 100 sq mile core with 500,000 people, 300 sq miles of old suburbs with 900,000 people, and 1000 sq miles of sprawlsville with 1,000,000 people. Density is 1714/sm and weighted density is 2583/sm.

By the next census, the core has increased to 101 sq mile (because zoning/permitting makes it hard to expand) and 515,100 people. The old suburbs have increased to 315 sq miles and 976,500 people. Sprawlsville has increased to 1200 sq miles and 1,260,000 people. Density is 1965/sm and weighted density is 2536/sm.

Note that every zone (core, suburb, sprawlsville) got denser and overall city density went up, but weighted density still went down.

Posted by LetsGoLA | Report as abusive
 

We can find abundant examples of where density is rising (that’s what apt and condo development does)… and also where low-density sprawl continues.
In most major metros you’ll find both these dynamics. Which is the more prominent — affecting a larger population — determines whether a given metro’s “population-weighted density” is ebbing or growing.

But recognizing coincident contra-trends does not seem to be the point of this blog piece. The author seems more interested in pushing a storyline about inner-city gentrification.

Finally, I’m not impressed by the glossed-over assertion that “population-weighted density” is the one “right way” to discuss land use. I much prefer our usual approach: We clip the urbanized area; then calculate land-weighted density within the urbanized area. If someone thinks “population-weighted density” is preferable, that needs to be explained.

Posted by TGraham | Report as abusive
 

@LetsGoLA, that’s only an issue with cities with large dense cores though. Calgary grew in the way you described, with all parts of the metro getting denser but most growth occuring in the suburbs, however, outside the downtown, the core is no denser than the suburbs, so the weighted density increased significantly. The same could have happened in many sunbelt cities but it seems most have larger areas of the core that experienced significant population loss and lower density and more leapfrog suburban growth.

In addition to the fact that a larger share of growth occured in lower density metros and lower density parts of metros (outer suburbs), there is another factor:
http://www.city-data.com/forum/26347869- post123.html

Posted by Memph | Report as abusive
 

Numerical measurements are bound to seem paradoxical if they are misnamed. “Population-weighted population density” is no doubt a useful statistic, but it is not a population density in the ordinary sense. This measure could very well be called the “crowding index.”

To illustrate this, consider a sparsely-populated rural region that has only one big shopping mall. During the day, some of the regional residents work in the mall, and many others shop there. People living outside the region seldom either shop or work there.

The result? During working hours, the “population-weighted population density” of the entire region shoots far above its value when the mall is closed.

You might say the whole region becomes more crowded Monday through Friday, but I, for one, would not like to say it becomes more dense.

Posted by Ralphooo | Report as abusive
 

Post Your Comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/
  •