By Ben Schulman and Xiaoran Li
Conventional wisdom holds that the larger the population of a city, the more successful the place must be. If the population’s growing, that city must be doing something right. If it’s withering, it must be in decline.
Seeing as so many of the feelings that are collectively construed to form a perception of a place – and inform many of the policies used to influence the design of such places – are derived from this basic fact, the question of what is actually being compared when looking at the population of cities begs to be asked.
While the health of cities is often discussed in terms of population, what’s often lost in the discussion, at least outside of certain wonky urbanist circles, is the size of cities in geographic terms. (Unless that city is Detroit, whose 139 square miles seem to be lambasted at times as one of the central causes for the city’s steady decline.) Putting density and urban design aside, simply taking a quick look at the square mileage of American cities reveals a wide disparity, from the 757.7 square miles of Jacksonville, FL to the 14.79 square miles of Jersey City, NJ.
Although cities are often judged prima facie, not to mention showered with congressional dollars via census results, based on population figures, perhaps a straight reading of the numbers isn’t a good barometer of the merits or demerits of a place given the wild variances in the geographic size of cities.
Cities are arbitrarily constructed entities with culturally loaded boundaries. So what would happen if every city shared the same geographic borders? Would population numbers reflect different realities? Would the perceptions of places change, defining which cities are viewed as declining or prospering?
To try and answer these questions, we decided to conduct a thought experiment. Using census and GIS data, we sought to standardize the size of the American city and find out the population that would reside within that uniform sq. mileage figure.
Our search began easily, using 2010 census data to confirm the population estimates and sq. mileage for the ten most populous cities in the U.S. We added up the sq. mileage of those ten cities and divided the sum by ten to arrive at an average sq. mileage for the “standard American city.”
That number comes out to be 355 sq. miles.
Rank | City | Population – 2010 |
Sq. Mileage |
Population – 2010 (Average Sq. Mileage) |
Average Sq. Mileage |
Change |
1 | New York | 8,175,133 | 302.64 | 8,020,022 | 355 | -155,111 |
2 | Los Angeles | 3,792,621 | 468.67 | 3,551,958 | 355 | -240,663 |
3 | Chicago | 2,695,598 | 227.63 | 3,197,099 | 355 | 501,501 |
4 | Houston | 2,100,263 | 599.59 | 1,700,164 | 355 | -400,099 |
5 | Philadelphia | 1,526,006 | 134.10 | 2,462,786 | 355 | 936,780 |
6 | Phoenix | 1,445,632 | 516.70 | 1,382,860 | 355 | -62,772 |
7 | San Antonio | 1,327,407 | 460.93 | 1,371,715 | 355 | 44,308 |
8 | San Diego | 1,307,402 | 325.19 | 1,433,195 | 355 | 125,793 |
9 | Dallas | 1,197,816 | 340.52 | 1,359,986 | 355 | 162,170 |
10 | San Jose | 945,942 | 176.53 | 1,416,827 | 355 | 470,885 |
14 | San Francisco | 805,607 | 46.87 | 1,277,962 | 355 | 472,355 |
18 | Detroit | 713,777 | 138.75 | 1,443,876 | 355 | 730,099 |
24 | Boston | 617,594 | 48.28 | 1,676,393 | 355 | 1,058,799 |
40 | Atlanta | 420,003 | 133.15 | 1,105,194 | 355 | 685,191 |
48 | Cleveland | 396,815 | 77.70 | 1,047,376 | 355 | 650,561 |
58 | St. Louis | 319,294 | 61.91 | 999,421 | 355 | 680,127 |
62 | Pittsburgh | 305,704 | 55.37 | 1,052,105 | 355 | 746,401 |
65 | Cincinnati | 296,943 | 77.94 | 982,924 | 355 | 695,981 |
73 | Buffalo | 261,310 | 40.38 | 810,524 | 355 | 549,214 |
Average Sq Mileage: 355.25 sq mileage | ||||||
[Sq Mileage of Top Ten Most Populous Cities Added/10] |
Using our average sq. mileage figure, we then used GIS software to graft a 355 sq. mile radius in the shape of a box from the downtown core of select cities and metropolitan areas across the country. The GIS software compiled the population within the box to arrive at a population figure if all cities were sized to the average. For example, we set our zero coordinates at the intersection of State and Madison Streets in the heart of downtown Chicago and drew out a 355 sq. mile tract from there, compiling the population figures within the newly drawn perimeter of the standardized-sized city.
Here are our results:
City | Population – 2010 (Actual boundaries) |
Population – 2010 |
New York | 8,175,133 | 8,020,022 |
Los Angeles | 3,792,621 | 3,551,958 |
Chicago | 2,695,598 | 3,197,099 |
Houston | 2,100,263 | 1,700,164 |
Philadelphia | 1,526,006 | 2,462,786 |
Phoenix | 1,445,632 | 1,382,860 |
San Antonio | 1,327,407 | 1,371,715 |
San Diego | 1,307,402 | 1,433,195 |
Dallas | 1,197,816 | 1,359,986 |
San Jose | 945,942 | 1,416,827 |
San Francisco | 805,607 | 1,277,962 |
Detroit | 713,777 | 1,443,876 |
Boston | 617,594 | 1,676,393 |
Atlanta | 420,003 | 1,105,194 |
Cleveland | 396,815 | 1,047,376 |
St. Louis | 319,294 | 999,421 |
Pittsburgh | 305,704 | 1,052,105 |
Cincinnati | 296,943 | 982,924 |
Buffalo | 261,310 | 810,524 |
A few interesting findings immediately pop out. Redrawn at 355 sq. miles, Chicago’s population jumps over 3 million, a number last crossed in the 1980 census. Some quirks of geography and the presence of independent enclaves reveal that some cities can see their footprints shrink and their populations grow, such as San Antonio. Conversely – and perhaps tellingly, revealing how arbitrary the process of drawing borders is – New York’s population would shrink by over 150,000 even as its sq. mileage increased. [Note: Obviously, drawing cities at a 355 sq. mile box is just as capricious as present practices. It does however offer a framework in which to view cities as comparable units of analysis.]
Where the altered numbers are most revealing though is for cities constrained by the legacy of their geographic confines. Enter the Rust Belt City.
Geography | Population – 2010 (Actual boundaries) | Population – 2010 (Based on 355 sq m average) |
Detroit | 713,777 | 1,443,876 |
Cleveland | 396,815 | 1,047,376 |
St. Louis | 319,294 | 999,421 |
Pittsburgh | 305,704 | 1,052,105 |
Cincinnati | 296,943 | 982,924 |
Buffalo | 261,310 | 810,524 |
Buffalo, a city of 261,310 as of 2010 over 40.4 sq. miles, is recast as a place of roughly 810,000 at the 355 sq. mile standard size. Cleveland, that punching bag on the lake, is transformed from a shadow of 396,815 across 77.58 sq. miles to a player of 1,047,376. St. Louis, the city that percentage-wise has hemorrhaged population more than any other American metropolis, sees its population rise to 999,421 at 355 sq. miles from 319,294 at 61.2 sq. miles. The pattern pretty much holds for Pittsburgh, Cincinnati and Detroit, other cities often lumped in the great fallow foundries of the interior of the country. (Other geographically constrained, yet non-Rust Belt cities express similar patterns: Boston grows to 1,676,393; Atlanta to 1,105,194; San Francisco to 1,277,962)
For Rust Belt cities who have experienced massive population loss and whose recent histories read as exemplars of post-industrial urban failure, the numbers seem to reveal a story of dispersal, rather than absolute decline. (A few years ago, the Buffalo-based planner Chuck Banas delved into this very issue.) Rust Belt cities taken at the standardized size no longer look universally like dying vines, but rather large agglomerations of people positioned at the same relative scale as more “successful” urban areas.
These points do not negate the economic insolvency and vacancy that plague certain neighborhoods that reside within the current, or imagined, bounds of Rust Belt (or otherwise) cities. And it should be noted that these findings and discussion offer nothing in terms of concrete policy prescriptions, nor do they posit that any particular form or design typology serves cities in a more beneficial fashion. Nor, although the numbers would seem to imply it, do we state that a certain density per sq. mile is essential for success. The authors may have opinions to share on such matters, but all elements of those discussions are topics for other debates.
What this exercise illustrates is that popular opinion puts a lot of faith into a metric that in actuality reveals an inaccurate depiction of the places condemned as shrinking and failing, and for that matter, as growing and successful. These assumptions tend to have a disproportionate effect on Rust Belt locales, as population trends that in reality point towards transference are read as absolute indicators of a place’s worth, or lack thereof.
What would it mean if cities were redrawn at a standardized scale?
A lot more goes into economic and human development than just geography, but if the city were standardized at the same geographic scale, it may affect the perception of place in terms of what economic geographer Jim Russell has deemed “the mesofacts.” Mesofacts relegate a place to the constraints of its own history, even if the reality of the present paints a different picture than the overarching popular narrative that defines it.
A simple exercise such as this, altering the lens in which population is measured, could help challenge the outdated narratives that reinforce negatively-charged imaginaries of what we think places to be. Doing so may help governing bodies on the federal, state, regional and municipal levels design policies that are attuned to the realities of what places are, and therefore, comport themselves to more realistic notions of what should be done to address their attendant problems.
Ben Schulman is the Communications Director for the American Institute of Architects Chicago (AIA Chicago) and the co-creator of the Contraphonic Sound Series, a project that documents cities through sound.
Xiaoran Li is an urban planning intern for Skidmore, Owings and Merrill and a master’s student in the Department of Urban Planning and Policy at the University of Illinois-Chicago.
Do you like what we do here at Belt? Consider becoming a member, so we can keep delivering the stories that matter to you. Our supporters get discounts on our books and merch, and access to exclusive deals with our partners. Belt is a locally-owned small business, and relies on the support of people like you. Thanks for reading!
Fascinating – this is another great illustration of the arbitrary nature of municipal boundaries. I wonder though – it appears a good portion of Cleveland’s 355 mi square is Lake Erie: an uninhabitable place (for humans, at least). I know this was an exercise of standardization but is there a way to account for places like lakes and rivers? Thanks for the great work!
It’s true that Lake Erie eats into Cleveland’s 355 sq mile radius, a hurdle we faced when drawing the maps for Buffalo, Chicago and other water-adjacent locales. For the methodology used in this exercise here, we simply went from downtown urban core outwards in a box shape, using that average sq mile #. A standard 355 sq mile form doesn’t necessarily have to reflect the shape of a box, and if such a standard were to ever become a reality, of course, Lake Erie wouldn’t be part of that area. Cleveland’s population would seemingly continue to “grow” then. Your comment of “the arbitrary nature of municipal boundaries” is spot-on. Discounting a place b/c of its relative population isn’t valid if the units within which the comparisons rest are entirely different.
Was Buffalo’s or Detroit’s population also including the section in Canada?
These numbers are fascinating, given that so much fed and state funding, etc. is based on jurisdiction size. I lived in Fort Worth TX for many years, and down there they have the issue of, “extraterritorial jursidiction’ (ETJ) going on. What that means is that Fort Worth can bascially annex five miles out from its boundary into unincorporated county areas (sort of like townships in OH) and keep doing so as much as it wants (the people who live in those areas have no real say). So Fort Worth’s size is now 350 sq mi. and growing. That’s why the Texas Motor Speedway is 22 miles from downtown Fort Worth and still in ther city limits even though there is ag land all around it. Based on the same distance, it would be like Avon Lake being in the City of Cleveland. That is one reason why these numbers make so much more sense (and more than the MSA numbers the census is always trotting out).
This is a fascinating piece, especially for someone like me that became an urban planner, because I loved maps and loved immersing myself in this type of data and formulating these types of analyses [gets far-off look in eyes and pines away for the days of being a graduate student].
But, I would like to offer the counter-point that ascertaining the “true size” of a city is really dependent upon what you are interested in doing with the information. If you are interested in ascertaining the buying power of a place, or the size of a media market, then, yes, by all means ignore those pesky governmental boundaries.
On the other hand, if you are looking at myriad categories of public policy, infrastructure, land use patterns, social service delivery; or if you are looking at social and economic disparities, then ignore those “arbitrary” boundaries at your peril, because they are anything but arbitrary in these contexts, and the framework presented here is little more than an extremely interesting academic exercise. In the socioeconomic and public policy context, local government boundaries are just as concrete and real as the Berlin Wall was – especially in Ohio.
I wrote about this recently:
Is the size of our central cities even important? Aren’t city boundaries arbitrary and meaningless? Isn’t it the surrounding metropolitan region that really counts?
Well, it’s a complicated story. For years, pundits, prognosticators, and policy wonks have been telling us that the age of the central city is over; that it is the region that is important. Economies are based on regional job markets, they say, and improvements in transportation and communications are making local places (even large ones) increasingly irrelevant.
The fact that economies are regional is true – as far as it goes. But like anything viewed through one lens only, it does not tell the whole story.
Are regions important? Of course. But so are places. Like so many other things in the realm of urban public policy, this is not a binary, either/or, choice.
Indeed, at the same time that we are being told by one set of pundits about the irrelevance of our cities, we have another set of pundits telling us that this is, in fact, a new golden age for our cities.
Cities entered a long cyclical downturn following World War II, they tell us, but they are now on the rebound, and are experiencing an unparalleled renaissance. Property values are increasing, Millennials are moving to our downtowns, and previously declining neighborhoods are coming back to life, replete with upscale shops, bistros, and pubs.
But this doesn’t tell the whole story, either. For every gentrifying formerly shrinking city like New York, Washington, and San Francisco, and for every sprawling boom town like San Jose, Charlotte, or Columbus; there is a St. Louis, a Cleveland, and a Detroit; and there is a Gary, a Flint, and a Youngstown.
What does the future hold for these cities? What about the giant places full of the mind-boggling, post-apocalyptic decay and dysfunction that comes with literally losing one million residents, like Detroit?
And what about the mid-sized places, like Flint, that may not have the assets or the resources to ever turn the corner. Will they continue to die a slow, agonizing death, and literally disappear? Or will they continue on in a shadow-form, serving as a cautionary tale, and inhabiting some type of uniquely American, urban equivalent of purgatory?
Or can they be restored – if not, perhaps, to their former glory, to at least something that is stable, equitable, and workable for those that remain?
Municipal boundaries are not irrelevant, whatever the regionalists may tell you. Economies may be regional, but in most of the nation’s fastest declining cities, government is not. Municipal boundaries affect taxation, land use policy, public safety, education, public infrastructure, and the delivery of social services.
http://thestile1972.tumblr.com/post/82706606733/a-tale-of-273-cities
Good point – there is no doubt municipal boundaries matter for municipal level policy. However, when one needs to analyze population, poverty, and economics at a “City” level; municipal boundaries are nearly useless as none of those attributes adhere to “imaginary” lines. The importance of municipal boundaries therefore depends on what it is your studying.
The part about NYC’s population going DOWN is unusual, as the geographical size goes up from 302.64 sq. mi. to 355….
Interesting but I would refine in a few ways:
1) 355 square miles is a nice “average” for a “top 10 city” but seems like a pretty extreme size for a Buffalo/Cleveland size city- indeed, for those cities, it includes a big chunk of a metro area. I suspect if you looked at the average for cities of 500,000-1 million people you might get a lower number.
2) because in every city but the three or four largest, 355 sq. miles includes a huge chunk of the metro area and includes places that are suburbs in form, a number that large doesn’t really set much of a city/suburb boundary. So…
3) what would be really interesting is the same project using numbers that really reflect the city/suburb distinction (e.g. extending 6 or 7 miles from downtown): that is, distances that make sense for nondrivers. For example, what would city sizes be for 60 square miles (the size of Washington or Cleveland) or 40 (the size of Boston or Buffalo, I think). Or to put it another way, if Sun Belt cities were stuck within the city limits of a Rust Belt city how small would they be?
4) a more theoretical point: I don’t think that if a Rust Belt city had 355 square miles it would have the population suggested above. Being within urban boundaries is a huge negative for parents stuck in urban school districts, so I suspect all the Rust Belt cities would be much smaller than suggested above if they had 355 square miles.
This study is excellent work, but it’s another way of stating something I’ve believed for some time. Rust Belt cities are the one group of cities in the nation where the distinction between urban and suburban, city and suburb, are starkest. Large area cities whose built environment is similar to the suburban areas that surround it lose population at the adjusted level, whether old (New York) or new (Phoenix). But cities, large or small in area, whose built environment does not mirror its suburbs gain population at the adjusted level. That contrast is consistently biggest in the Rust Belt, where old = city and new = suburbs there in ways that exist nowhere else in the country.
@pete-rock
This is an absolutely fantastic, under-appreciated, often un-made, point!
Interesting thought experiment. To me, it shows the value of looking at REGIONS centered on a city, and not just the city itself. Regions are better units of economic analysis. Lots of people live in 1 part of a region (perhaps not in the major city in the region) and work in another, as a simple example. Drawing regional boundaries is not easy, but it seems worthwhile to figure out. There should be a mathematical way to figure out each region’s borders, instead of just picking an average square miles. For example, a neighboring county is part of the region surrounding a city if X% of the county residents work in the city. The math would need to be a lot more complicated that than, however. I think your analysis is an example that points out the value of regional thinking, not city-limits-thinking.
Quick comment on the St. Louis map – the square should be drawn with the existing city boundaries on the far right if there’s any desire to measure the city (urbanized area). Including the near east side – floor plains largely – would be like centering the Chicago map on the waterfront. The functional St. Louis City is much more nearly centered at I-170 and I-64. The population center of the region is actually further west than that.
Excellent work. We know rankings of cities defined by “city limits” are deeply flawed because of those wildly varying political boundaries within various metro areas. So St. Louis, whose city limits encircle only the inner 1/10 oldest inner core of the metro area and no suburbs, ranks poorly on CQ Press city crime lists compared to, say San Antonio where limits include 70% of its metro area & suburbs. But on CP Press Metro Statistical Area (MSA) crime rankings, the boundaries are consistent, similar to yours, metro to metro – defined by the federal government. On those rankings, St. Louis metro ranks 100 places lower and safer than San Antonio and most other large metro areas. (For San Antonio, their city and metro rankings are the other way, forcing the rankers to assume that SA distant suburbs are more dangerous than the core I guess, rather that admit that their city limit ranking methodology may be badly flawed, which it is.) So if one assumes that the metro rankings are fair, and that your rankings are also fair, and that most metro area crime is higher in the core and declines with distance as you look toward the suburbs for most metro areas, then I would expect that your rankings for crime (and other things) might roughly mirror the full metro statistical area rankings, such as the CQ Press Metro Area Crime Rankings. Do you find this to be the cases? Thanks.
One more thing. This is a good start at normalizing rankings by trying to factor out big drivers of rankings that have nothing to do with the item purportedly being ranked. But it doesn’t quite work for old rust belt cities whose inner cores have been somewhat hollowed out and their residents have moved out BEYOND the 350M square, such as St. Louis. San Antonio and newer such cities have not gone through that yet, so their 350M square still l includes much of their population, newest housing, and low crime suburbs. But it is an improvement. I think you just have to take the next step of normalization, and make sure they match core to core and suburbs to suburbs, no matter how far sprawl has taken the suburbs from the original core, in order to get a fair comparison of crime safety. They could do this with zip code data. When visitors visit, the don’t know where city limits are, or 350 mile boxes. But they do know roughly where the metro area begins and ends. So, so far MSA rankings are still the most accurate rankings.
Love this story and the thoughtful comments it inspired. When I was at the PD, the US Census Bureau declared Cleveland America’s poorest big city, circa 2003. Dave Davis, the paper’s census data expert, was quick to explain to me that the label was more a reflection of geography than the wealth of the region. To illustrate, we ran a story that transposed the borders of Cleveland and Columbus, Cleveland and Charlotte. Cleveland became dramatically richer as those cities became much poorer within the new boundaries. Still, labels matter. Everyone a poverty problem. Cleveland’s is made worse by its map.
Bob Smith
OK, one more thought. Do the same thing, but instead of standardizing a box area, standardize by inner core 25% of population for the metro area. Then inner 50% from core. And inner 90% from core. And finally all of the metro area. You could do this by going down to zip code level data, which is usually more work than CQ Press and other statisticians are willing to do. But I think it would give visitors and residents a better idea of crime differences between metros, and crime differences within their own metro.
Great exercise, even if it massively screws places like Detroit and Cleveland! It would be even cooler if you could do a second comparison of populations against a smaller geographic area more representative of the core… e.g. what would Phoenix or Los Angeles look like if they were only 62 sq. miles like Saint Louis or with Boston’s 48?