The Claim

This is one I’ve heard frequently since moving to Kansas City, and it shows up all over the place. It’s on our wikipedia page, Kevin Klinkenburg has written it, it’s a frequent talking point in our local subreddit, and it’s been covered in Streetsblog and Strong Towns several times each. But is it true?

The Finding

No! There is no way I could figure to define the city’s extent such that this is true. Not for the city proper, not for the UA, not for the MSA or CSA, nothing.

…But it used to be; the claim didn’t come from nowhere.

As far as I can tell it has two independent origins:

1. The longtime anti-urbanism blogger and think tanker Wendell Cox published a list on his blog in 1999 called “Lane Miles per Capita 1999: US Urbanized Areas over 1,000,000” that listed KC #1, ahead of Fort Worth, Dallas, and St Louis.

2. The blogger behind a now-defunct site called WalkableDFW tabulated 2007 FHWA data for a post in 2012. Kansas City came out #1 followed by Houston, St Louis, and Cleveland, with the caveat that the source only estimated lane-miles. They also used a population threshold of >1mil.

Both these bloggers relied on the same dataset, a publication FHWA puts out through its Highway Statistics Series program with the eye-catching name “Urbanized Areas - Selected Characteristics”. It gives a range of stats—including lane miles—for Urbanized Areas in the US. Confusingly, FHWA doesn’t mean Census Urban Areas but rather their own in-house version that MPOs are allowed to add territory to for a range of different purposes.

I ran the numbers again and found Cox was right for the 1999 series and WalkableDFW was right for the 2007 series. Kansas City’s FHWA Urbanized Area did have the most lane-miles per capita (min pop 1mil) in both those years.

Happily, however, there’s an updated version out this year and Kansas City is no longer #1. We have fallen all the way down the list to…..#2. Nashville is now #1. Their 1.297 lane miles per 1,000 people beats out 1,267 lane miles per 1,000 people.

It isn’t for the reason you might think, though: Nashville just crossed the 1mil population mark for the first time in 2024. The full tables show that they have had more lane miles per capita than us since 2014 but were sitting just below the population threshold until FHWA recalculated Urbanized Areas with 2020 Census population figures. Both cities have added lane miles in recent years; it’s not clear how much construction has contributed.

As a partial aside, the Streetsblog writeup of the WalkableDFW list calls it a ‘who’s who of urban decay’. That strikes me as completely wrong. St Louis, Cleveland, and Cincinnati are all slow-growth Urbanized Areas, sure, but Nashville, Dallas, and Jacksonville are three of the fastest growing in the country. Like the first three, Kansas City has localized decline that the freeways contribute to, but it still consistently grows by 10% or more each decade.

Further, if you drop the population minimum, Kansas City completely falls off the list: there are fifty-six cities with <1million residents in this dataset that beat us out. The five highest have more than double our lane miles per capita. Kansas City doesn’t even have the most in its own bi-state area — Topeka, Springfield, and St Joseph are all among the top thirty overall.

For the smallest cities on here this doesn’t suggest all that much about urban morphology or transportation systems — if just one or two major freeways passes through your Urbanized Area and you have <100,000 residents, you’ll have a high ratio regardless of how the local population interacts with it. But plenty are large enough that it’s fair to say they’re more freeway-oriented than KC.

Two More Thoughts on KC Freeways

There’s another interesting stat in this dataset: daily lane utilization. The average freeway lane in a US city carries ~16,700 vehicles per day. Kansas City’s carries 11,144, a third below that, and only Pittsburgh is lower among the forty-five large urbanized areas. So we don’t just have a lot of freeway relative to our population, we have a lot of freeway relative to the number of people actually using it.

Finally, I think at this point the Kansas City freeway network is basically built out. Sure we will fix some chokepoints (widening I-70, rebuilding the I-35/I-29 split, eliminating the US-71 stoplights) but there are no plans for (and no need for) any new freeway alignments anytime soon. With our growth slowing and both commute times and lane utilization low, there simply isn’t much need for more big roads here. This was already my view before writing my last two posts but I am now more confident in it.

The Claim

When I criticize the excesses of Kansas City’s freeway network—a hobby of mine over on Twitter—I consistently get the same pushback from commenters who aren’t ready to fully jump on the urbanist bandwagon: sure the loop is a mess, they say, and the construction of the freeway system destroyed Kansas City’s emerging identity as one of America’s best planned cities, but at least it’s a useful system. For all the mistakes we may have made building it, the result is a network that keeps congestion low and commute times short. And those benefits matter far more to the average person than any aesthetic preference for distinctive urban design.

Are they right? Do we have particularly short commutes, short enough that I might begrudgingly admit our midcentury planners were onto something when they sliced this place apart with overpasses and on-ramps?

The Finding

The answer is pretty clearly yes. Kansas City is tied for the shortest average commute time of any Urban Area with at least one million residents in the U.S.

Analysis

Among Urban Areas with at least one million residents, Kansas City is tied with Milwaukee for the shortest average commute time in the country. Its 22.2 minute average is 18% shorter than the US overall and 38% shorter than the longest, New York’s. The three closest UAs are all substantially smaller than KC, too—it’s not until you get to Minneapolis (40th) and then St Louis (35th) that you find larger. We’ve managed to keep commutes short as we’ve grown to a relatively large size.

It would therefore appear that below some population threshold a built environment like Kansas City’s — near-uniform suburban densities, weak polycentricity, and a thick network of freeways — is ideally suited for getting people to jobs quickly.

The linked posts here give a flavor of this urban geometry.

The distribution of commute times is worth considering, too. Commute time inequalities manifest in various ways: some cities have wealthy cores with short commutes and poor suburbs with longer commutes; others have a segregated zone where workers commute long distances; and some have a patchwork of long-commute areas produced by several different forces, including low-cost housing at the metropolitan fringe, concentrated employment centers, transit dependence, and affluent exurban growth.

Kansas City commutes are reasonably uniform, with a long but thin right tail. A tiny share of our residents commute more than half an hour, and the majority commute 21 minutes or less (the tail drags the average up). We don’t appear to have any of the patterns described above, at least not at scale.

You can visualize the distribution geographically on the following map.

The favored quarter generally has the shortest commutes, and more isolated pockets like Independence and Riverside are somewhat longer. There’s some correlation between freeway access and commute times, but it’s weak—for example, there are a few long-commute tracts within a mile of the downtown freeway loop, and the large rectangular block of neighborhoods between I-35 and US-71 in northern OP and KC proper has some of the shortest commutes. There’s no clear pattern of disparity of any kind.

The next map checks this by looking at the tracts with the shortest and longest commutes. They’re randomly scattered.

Here’s the data sliced by jurisdiction. Suburban Jackson County comes out looking worst and Johnson County best (another W for the favored quarter). KC proper is right near the average.

There’s no convincing patterns of socioeconomic disparity visible in a simple review of the data. The following charts regress commute times by tract against tract median income and racial group shares. None has any statistical significance.

To give you a sense of how these short commutes happen, here’s a chart showing commute mode in the UA. <2% of commute trips are done by transit, walking, or biking. I’m a very satisfied member of that 2%, though I’ll acknowledge my work/home trail connection is mostly a happy accident, and not common. The vast majority of Kansas Citians drive, either alone or with someone else. 16% usually work remotely.

But!

Writing this post has basically convinced me that our freeway network creates short commutes. But of course that’s too simple a correlation, and a myriad of other factors no doubt contribute—geography, political economy, idiosyncratic historical events, etc.

This also isn’t a real cost/benefit exericse. The cost of this network is real, and significant. Even knowing we would have the shortest commutes I would not have designed the city this way. There are substantial social, economic, and environmental costs to our large freeway network.1

And to go back to my disparities analysis, commute times are not the only transportation-related issue for poor and minority communities in the United States. This data misses inter-tract variability and can’t speak to the cost burden and access issues that a car-dependent built environment places on poor households.

Close

It’s somewhat surprising to me that I’ve never heard local boosters tout the fact we have such short commutes. “Shortest average commute of any big city in the country” is a factoid you’d think KCADC, the Chamber, local columnists, politicians and sports broadcasters would all love to yap about. But perhaps because the Urban Area—while superior to all other city definitions—is a bit obscure, and this only holds with a constrained set of big cities 1mil+, no one seems to have put this together before.2 If you are one of these people and you would like to start talking about this, feel free to credit my blog for it.3

The freeway enjoyers are right, at least on the narrow question posed here. Kansas City has built a transportation system that moves workers to jobs quickly, with relatively little variation across neighborhoods. Urbanists should acknowledge this rather than pretending the system is merely wasteful or dysfunctional. The harder question, of course, is whether shaving several minutes from the average commute justified the land consumption, neighborhood destruction, infrastructure burden, and automobile dependence required to achieve it. The data here cannot answer that question but they clarify at least one element of the tradeoff.

Still. Land of the short commutes. That’s a pretty good thing to be.

Subscribe now

Share

This post makes some of those comparisons against my home UA, Kansas City.

The Kansas City Urban Area has ~1.75 million residents across 751 square miles, for a population density of 2,330 people per square mile.2 It is the US’s 29th largest UA by population, the 20th largest by size, and the 35th densest of the 44 with at least a million residents.3

Historical Comparisons

1.75 million people might only be good for 29th largest in the US today but is massive from a historical perspective. No urban area4 crossed the 1.5 million mark until London during the first wave of industrial urbanism in the middle of the 19th century, and no American UA topped KC’s size until New York somewhere around 1870. The first midwestern UA to hit 1.75 million was Chicago a decade or two later, followed by Detroit in the 1920s. In the first census that measured UAs, 1950, only eight UAs had more than 1.75 million residents.

Regional Dominance

Kansas City is relatively isolated for a city of its size, as I’ve written before. Of the 40 largest US cities, we have the 34th fewest number of people within a four hour drive of us. To get to any of those 40 cities you have to drive a good ways — to Minneapolis to the north, Denver to the west, Dallas to the south, and (closer but not close) St Louis to the east.

This isolation actually helps explain why KC grew as large as it did. As its region—the eastern edge of the central Great Plains—was peopled in the late 19th and early 20th century, Kansas City drew the rail network, the central technology of the Plains economy, around itself, an advantage that began when the Hannibal Bridge opened in 1869 as the first to bridge the Missouri and compounded from there. It became the natural hub for the trade and manufacturing of Plains commodities. Central place theory predicts that a city’s market area expands to fill the space around it; relatively low population density coupled with KC’s dominance of transportation kept that void very large.

Area

Our 1.75 million residents collectively consume 751 square miles of urban land. How big is that?

One way to think about it is to compare with historical cities. One of the most interesting findings in urban economics is that the length of time people are willing to commute to work on average has been relatively consistent throughout history—about 30 minutes. Before mass transit and automobiles, this limited the size of a city to radii of ~ that walking time from any major job center. And job centers clustered near the center, both for transportation infrastructure access and to maximize their access to labor.

When London surpassed the modern KC UA in population in the 1830s you could have walked its entire length north to south in a couple hours (~6 miles). To walk today’s KC UA north to south would take—per google maps—15 hours.

Today, of course, commuter rail, cars and freeways have allowed Urban Areas to sprawl across vast regions.

Of contemporary American Urban Areas, Kansas City is the 20th largest, behind Pittsburg, San Juan, and Cincinnati, and ahead of Indianapolis, Cleveland, and San Diego (population 3 million). The largest, NYC, is 3200 square miles. San Jose is the smallest with a population larger than ours (1.85 million), consuming just 285 square miles.

Here are a few non-UA comparisons with some well-known geographies. We’re about twice the size of New York City proper.

We’re a fifth larger than the Hawaiian island of Oahu, and 3x the size of Singapore’s main island (population ~6 million).

Density

It’s also interesting to compare the density of the Kansas City UA with its peers.

Density comparisons at the scale of a modern Urban Area can be misleading, as industrial districts, airports, rivers, and other places where few people live are often included in the UA boundary.

A better metric is population-weighted density, a measure that captures the density of the neighborhood the average resident actually lives in by weighting each area’s density by its population so that empty land doesn’t drag the number down.

Of the 45 UAs with at least a million residents, Kansas City ranks 37th by this measure (3303 ppl/mi^2). Its closest neighbors are St Louis, Pittsburg, and Cincinnati (8%, 4%, and 1% denser) and Richmond, Indianapolis, and Atlanta (3%, 5%, and 6% less dense). 11 UAs (including Las Vegas!) are at least twice as dense; the densest—NYC—is more than 8x.

In a previous post I look at weighed density another way—by the share of people living at each density. It gives a nice view of how much the UA clusters at its average density.

Close: Beyond the Imageable City

It’s my view that the Urban Area is the best representation of what a city is and how it is bounded, because it is the most legible (easy to tell where the edges are) and corresponds with how we live in cities today (we freely cross jurisdictional borders but few go beyond the built up era day-to-day).

Still, Kansas City’s 751 square miles are hard to comprehend as a single place. I love to walk, run, and drive around the city just to look at things, and still I have only been to a fraction of the Kansas City UA. I could live here decades and not see large swaths. And even if I did see them — if my ‘run every street’ project gets truly out of hand — it would be hard to keep a mental map of the whole UA in my head.

The great urbanist writer Kevin Lynch proposed a concept of ‘imageability’ to explain how people understand urban space. Imageability, or the ease with which a city’s parts can be recognized and organized into a coherent pattern, shapes people’s perception of the city and their ability to understand it as a whole and their place within it, physically and otherwise. Writing in the late 1950s, Lynch was concerned this was something we were losing as cities scaled up and sprawled out into the vast regions they consume today:

“Nowhere in the world is there a metropolitan area with any strong visual character, any evident structure. The famous cities all suffer from the same faceless sprawl at the periphery. One may reasonably ask, then, if a consistently imageable metropolis, is in fact possible; and whether it would be appreciated if it did exist.”5

This is the challenge I see with defining the city today, and with how I perceive the Kansas City Urban Area: the UA is the smallest unit at which I think it is reasonably bound the city, and yet it is far too large to really know, or to see, it. Kansas City is tremendously large.

Subscribe now

Last year I told a friend I moved to Kansas City and he said “Oh that’s not bad, you’re smack dab in the middle of the country, actually kind of close to everywhere.” This of course is true—if you look at a map, KC is pretty close to the geographic center of the mainland US, which falls just northwest of Lebanon, Kansas. We’re decently close to the population center too, which, in a win for nominative determinism, is just north of a tiny Missouri town called Hartville. I’m no more than a few hours’ flight time from friends and family across the country.

From the perspective of mainland US geography, Kansas City is the most central of any major city. This gives us great access by air—the longest domestic flight out of MCI is less than four hours. There’s a lot more that factors into it, but I imagine this is also why small regional airports at Branson, MO and Fort Collins, CO have the cheapest average domestic airfare of any airports in the country. We’re very conveniently located for long distance domestic travel.

We’re centrally located in other ways, too. Freight traffic (in many ways our raison d’être) flows through Kansas City at a rate wildly disproportionate to the population or industrial base.

This is particularly true for railroad freight. Kansas City is the 31st largest metro in the country, but we’re the second largest railroad hub by freight volume and car traffic (the two most common measures). Only Chicago, a city four times our size, sees more trains than we do.

Most freight now moves by truck and we punch above our weight there too. Four primary interstate highways interchange in the Kansas City loop, connecting us North-South on a line through the population nodes that form the edge of the Great Plains (the divide between the densely populated east and sparsely populated west) and East-West along the most direct route between the country’s second largest city and its capital. This ends up mattering less for truck freight than rail freight, but we are still a top 15 trucking node. And that’s despite being far from major ports!

And we have only recently started to explore what more inland waterway traffic could look like on the Missouri again.

Centrality? No, Isolation

It’s when you look at geographies smaller than the country as a whole that this centrality starts to break down. You can get anywhere pretty quickly by air, and ground freight is fairly insensitive to long distance travel. But human beings primarily travel by car for work and leisure, and it’s here that our centrality completely reverses.

To demonstrate this, here’s a map of the area within 500km of Kansas City, a decent proxy for what or who you can reach on a weekend driving trip. You can do this more precisely with isochrone mapping—and I might eventually; I have a good R tutorial saved—but this is good enough for a rough pass.

There are a number of midsized and small cities in this loop: St Louis, OKC, Des Moines, Omaha, Tulsa, Lincoln, the Quad Cities, Fayetteville, Wichita, Topeka. The Ozarks, particularly at the farther end of that range, give us access to great scenery and outdoor rec. The cultural diversity of this drivable area is underrated, too: Indian Country in Oklahoma, the Great Plains, the Ozarks, the Missouri Rhineland, midwestern farm towns, deep French influence along the major rivers, and more. It’s not like we’re in the middle of a desert.

But think about what’s missing. We have no oceans, no top-tier cities, no world-class universities, no spectacular mountains, no great beaches, no national parks (except the Arch NP but that hardly counts), and no defining industry cluster (like tech in Austin/SF, medical/biomedical in Boston, music in Nashville, autos in Detroit, entertainment in LV & LA, or finance in NYC). When it comes to truly spectacular, world-renowned places and institutions, we are simply not in driving range.

More than cultural and geographic resources, though, the paucity of people gives KC its feeling of isolation. Here’s a chart of the forty largest US metro areas ranked by the number of people who live within the same 500km radius. Despite our centrality, Kansas City lands 34th on the list. We are between a lot of people, but those people are mostly pretty far from us.

Implications

You can make an argument that none of this matters all that much. I can almost hear commenters typing out the word “elitist” as I sit here. And there’s definitely something to that.

But the reality is that it does matter. There is immense value in having access to the world’s great places and scenes.

There are also three economic consequences worth considering.

The first is network effects. Urban econ researchers have found that midsized cities in dense regions are sometimes able to ‘borrow size’ from the broader population around them; that is, they are able to host more urban functions than their own size could otherwise support. The purest example of borrowing size in the US is Wilmington, DE, which hosts a cluster of financial and legal firms much larger than you’d typically find in a city of its size. Because of its position along the Northeast Corridor between DC, Baltimore, Philly, and New York, Wilmington was able to market itself to firms that needed general proximity but not immediacy to those cities. Without many people nearby, we struggle to punch above our weight in any one area (bbq excepted).

A second is attraction challenges. One key source of metropolitan economic growth is in-migration of highly productive firms and workers. Elite cultural institutions matter a lot to these types. Crystal Bridges is a great example of an elite institution serving as an attractor—the Fayetteville/Bentonville area is counting on it and other high end investments to grow its corporate sector beyond Walmart, something they have seen a lot of success with in the past few years. Quality outdoor rec opportunities matter, too—think of Silicon Valley and Miami, or the emergence of Denver and Salt Lake City as tech hubs in the past few decades. Without proximity to other cities’ amenities, Kansas City can only sell potential movers on its own attractions.

Third, elite universities are particularly important for long run economic growth, both for the firms they attract and the human capital of the graduates they produce. With respect to this region’s quality land grant universities, WashU Law is our only academic institution that consistently produces top-crust professionals. Mizzou and KU aren’t spinning off startup scenes like MIT and Harvard do in Boston or Carnegie Mellon and Pitt do in Pittsburgh. We don’t even do a particularly good job of keeping their grads.

Finally, there are cultural challenges. Our market size is a big mark against us for attracting large events and tours. We do a great job with what we have—the World Cup in particular is a signal of quality regional leadership on this front—but undoubtedly suffer because of our limited population base.

Fin

There are worse things than this, for sure. It’s a constraint, but every place has constraints. I wrote this post more to put a couple figures behind this feeling that I had than to say anything definitive about our region’s potential or my quality of life here.

We’re in the center of everything, but it’s all far away.

Share

This post follows the same format as that one, giving quick high-level thoughts about what another year of change means for our region. It’s quite delayed, apologies — I’ve been on the road a ton and this month started a new job.

Quick Hit #1: Kansas City is growing robustly

Let’s start with the headline figure: Kansas City is growing robustly again. After a slow start to the decade it added nearly 5,000 people in each of the past two years. You can arguably feel the inflow if you follow the news: there are more residential towers in the permitting or construction process now than at any point in the city’s history, the West Bottoms and Berkley Riverfront each have billion+ dollar projects underway, and a number of large Northland developments are in the early stages of coming together.

KC’s mirrors growth for the metro, which picked up over the past few years after a slow start to the decade. As I mentioned in the post on metro growth, however, we are likely to see slower growth later this decade as international migration slows. We don’t have disaggregated migration data at the city level but we know from the county data release that migration has made up more than half of JaCo’s increase in population since 2020. It’s reasonable to think that’s mirrored in the city proper.

But still, robust growth. 27% of the metro’s growth last year in a city that comprises just 23% of its population.

Quick Hit #2: …but slower than before

A 13,092 person increase since 2020 gives KC its slowest half decade of growth since 2000-2005, at just a 2.6% increase. That’s disappointing, but if we had grown at the average rate of just the past two years we would have added 4.8% to our population, a faster-than-average pace. Like all cities, KC struggled to attract and retain population during the pandemic. And as the stat at the end of the last paragraph demonstrates, the slowdown is more a result of slower metro growth than KC losing ground against its neighbors.

Still, we’re growing slower and are fairly far behind the 2010s.

Quick Hit #3: Good news for the urban core

It’s hard to know where growth has happened within the city until we get richer data at the Decennial. You can get a hint from looking at housing permits though: the general KC development pattern is that single family homes are built in the Northland and multifamily happens in the historic core. There was a modest single family building boom on the east side of the core and in West Plaza up until a couple years ago but the energy code changes seem to have brought it to a halt. So looking at permits by unit count can give us a decent sense of where growth is happening.

The next chart tells the story: with the exception of 2023/2024, each year just under 2/3s of new housing permitted in the city is multifamily. Not all of that is in the historic core (there are always a couple multifamily buildings going up north of the river, and the occasional project down south of Bannister) but a large majority is. The core is almost certainly adding population, and that should continue in the coming years as several large towers begin construction and major projects finish in the West Bottoms and Berkely Riverfont.

Quick Hit #4: KC is adding more residents than any of its suburbs. But they’re growing faster.

The largest gains since 2020 have been in KCMO, Olathe, Overland Park, Lees Summit…and Raymore, which is the fastest growing locality in the metro. Raymore, Gardner, and Belton, all growing from a small base, lead the metro in relative growth. In relative terms KC falls all the way from first to eleventh and Overland Park falls from 3rd to 10th.

On the other end of the spectrum, Raytown leads in relative decline, while Indendence leads in absolute decline after losing more than twelve hundred residents in five years. Population decline in these two cities (and on the east side of KCMO) is one of JaCo’s two most serious medium-term problems.

Most of the fast-growing small cities are adding single-family housing overwhelmingly, while the larger (including KC, OP, Lenexa, Olathe, Lees Summit) are more mixed. Overland Park has permitted at least 1,500 apartments in large buildings since 2020. Lees Summit permitted more than 2,200, including more than 1,100 last year.

Close1

Population growth is among Kansas City’s biggest challenges over the next few decades. Improving amenities (the streetcar being the headliner) and the pull of centrality have drawn people into the core over the past decade and a half, but any number of challenges (violent crime being the headliner) could start to slow or reverse gains. Even if the Northland and the streetcar spine (Riverfront→Plaza) keep growing, further decline east of Troost could offset gains and leave the city with deepened fiscal liabilities that will be especially challenging if its post-covid downtown employment base remains weak.

With that said, it’s hard to feel too pessimistic about a year where the city added 4,700 new residents. That’s 4,700 more people renting apartments, shopping in local stores, paying the income tax, and enlivening our parks, sidewalks, and events. Growth is positive sum. I’m happy to have it here.

There are a number of different ways of measuring city extent; that is, where the city ends and what it encompasses. It’s somewhat unintuitive if you haven’t considered it before but which you choose depends on how you define what a city is (not as easy as it sounds) and can include some rather arbitrary line drawing. Is a city a geographic area defined by jurisdictional boundaries? That’s an easy definition, but somewhat unsatisfying for most purposes. Is a city a continuous built up area of homes and businesses? A collection of people living at some minimum population density? A market for jobs or housing? Each definition is likely to give you a different city extent. I have a previous post that maps applications of these and other city definitions in the Kansas City area.

Economists, sociologists, governments, and others apply and operationalize different definitions to match their purposes. An economist is likely to use the labor or housing market definition, which bounds the city in such a way that you can study and compare it as a quasi-independent economic unit. A sociologist might use the built up area or population density definition to compare cities across national boundaries or across time, reasoning that a city as a collection of people and buildings is how people have colloquially thought about cities throughout history. A government might care most about the jurisdictional definition, for that is where various structures of state control—taxation, policing, infrastructure ownership—are bounded.

As an average person, I find myself most drawn to the built up area definition. It’s not how we use the word ‘city’ colloquially today (most people default to jurisdictional boundaries) but best measures the city as I experience it. I cross jurisdictional lines frequently to see friends and family. My wife crosses three on her 15 minute drive to work every morning. Outside of local services provision, mundane signs along the roadway, and a small share of the taxes I pay, jurisdictional lines just don’t feature prominently in my day-to-day experience. I visit places throughout the built up area of greater Kansas City all the time, and don’t feel like I’m leaving something akin to a city until I hit farm fields at the edge. My experience of urban life is not experienced within the arbitrary boundaries of Kansas City proper.

The built up area also tracks better with Kansas City’s amenities, culture, and status than jurisdictional boundaries do and does about as well on those measures as do labor markets or other population-within-driving-range measures. Major amenities—sports teams, the fine arts, the restaurant scene, etc—scatter throughout a region’s built up area and draw customers from all parts of it.

Finally, I feel the built up area is the most useful measure for comparing cities across countries and over time. How jurisdictional boundaries are drawn and what they encompass varies widely across the US and the world, and labor and job market data is hard to assemble beyond the US and in the past. The built up area of any contemporary city can be roughly determined with satellite imaging and that of cities in the past from maps, making it the most legible of all common measures.

Urban Areas

The easiest way to operationalize built up area is with the Census Bureau’s Urban Area designation. An Urban Area (UA) is a continuous area of at least 5,000 people created by grouping together contiguous census blocks radiating out from a central core that meet a minimum housing unit threshold of 200 units per square mile.

It’s not a perfect definition (any delineation of the built environment will have edge cases) but tracks nicely with the line between subdivisions and large-lot rural housing scattered among farmland.

Mapping UAs give the most granular and even view of the extent of large settlements in the US. Unlike MSAs, which commonly include huge undeveloped areas, and city limits, which fragment continuous development into dozens or hundreds of places, UAs are self contained and limited to what looks physically like part of the city.

The map below, taken from my previous post about urban extent measures, shows what the edge of a UA looks like. This is southern Johnson County KS.

As you can see, the UAs edge corresponds pretty clearly with the end of the continuous landscape of suburban subdivisions and the beginning of unbuilt farm fields. Where there is divergence it is mostly because UA boundaries are only updated decennially and thus don’t include post-2020 subdivisions.

Kansas City

I have a forthcoming post exploring the Kansas City UA in more detail but since this is a blog about KC here’s some quick facts:

The Kansas City UA has ~1.75 million residents across 751 square miles, for a population density of 2,330 people per square mile.1 It is the 29th largest UA by population, the 20th largest by size, and the 35th densest of the 44 with at least a million residents.

This, to me, is Kansas City.

Wrap

Urban Areas are not always the right way to measure city extent but they make the most intuitive sense to me and are most useful for the kind of intercity comparisons that I enjoy making on this blog. You can play around with an interactive map of them here or see more data about them here.

MSAs

Let’s start with the headline finding: the MSA added 17,395 people (0.77%) last year. That’s slightly down from the previous two years but above our five year average.

Since 2020 we’ve added 75,441 people (3.44%), which puts us on track for moderately slower absolute growth than the 2000s and 2010s. Slow growth in ‘20 and ‘21 have been almost completely offset by faster growth in the past three years.

Relative to other large US metros, KC has seen solid growth in both the past year and the first half of this decade overall. We were 20th/50 in 2025 and 22nd/50 overall, far outpacing St Louis (44th, 42nd) and similar to Midwestern peers like Cincinnati (23rd, 29th), Columbus (15th, 20th), and Indianapolis (13th, 16th). Fellow railroad town Denver (31st, 19th) had a slower year but has grown similarly overall.

As is typical, Sunbelt growth dominated the charts while the Rust Belt and Coastal California bled residents. Midwestern metros were mostly somewhere in the middle.

Counties

The Kansas City MSA encompasses 14 counties, about half of which are very small. Most of the signal here is in the larger counties.

Johnson and Jackson Counties unsurprisingly grew most in both periods, though the Northland counties (Platte, Clay) collectively outgrew Jackson. Jackson is the most bimodal of the region’s counties, with fast-growing suburbs on its periphery (Blue Springs, Lees Summit) and stagnant or shrinking city and inner ring neighborhoods in its northeast. Johnson, meanwhile, is both densifying along key corridors—Metcalf, Nall—and sprawling outward. Growth is beginning to jump over the Blue River/Wolf Creek Valley all the way down at 183rd St.

Cass County has added >1k/year as Belton and Raymore have continued to urbanize. Miami, too, is feeling the city’s reach and adding residents fairly quickly relative to its size.

Wyandotte has had a great couple of years and turned around its early 2020s population loss. Almost all of this has been in the greater I-70/I-435 interchange area, presumably.

The Census release decomposes the county data into several components: natural increase/decrease, net domestic migration, and net international migration.

The only really noteworthy thing here is that Jackson and Wyandotte have both seen net domestic outmigration over the past five years. No doubt driven by core population loss, it’s a worrying trend. If the Kansas Cities can’t stop the bleeding they will suffer as international migration collapses through the rest of the second Trump administration.

Finally, a look at our growth in the 2020 against MARC’s most recent population forecast. There’s not yet much divergence—we are 6k people north of expectations. I expect that to change as the forecast ages but it’s a useful grounding for understanding what the future of our region is likely to look like.

Overall, a nice release for KC. Population growth is not an unalloyed good but it brings many benefits. Ours is both strong and reasonably well shared.

What is a city? Is it a legally incorporated municipality? A collection of people who live near each other? A cultural identity? A labor market? A group of buildings and infrastructure? Something else? 

You might pick one of these definitions for one purpose, another for another. And that's fine. It doesn't make much different in casual use. But if you want to be more rigorous about it—say to compare two cities to each other, something I do frequently on this blog—it matters what you choose, because city extents vary a lot. In some places, like Honolulu, city limits includes almost everyone you might plausibly say lives in that city. In others, like Salt Lake City, very few of the people who would tell someone they meet elsewhere "I'm from Salt Lake City!" live in city proper.1

The most common way to think about a city’s extent is by the administrative boundaries of the core municipality (what you might call "Kansas City proper"). Core municipalities are typically the strongest and most influential unit of government in urban areas, which matters, but this definition otherwise obscures as much as it reveals. The US is a suburban country. Most people who call US cities home live outside the core municipality in sprawling suburban regions, and our cities and their suburbs are codependent on each other for jobs, housing, transportation infrastructure, and other amenities. Judging where the city ends and the region starts, or where the city-region ends and rural areas start, is a largely a matter of taste and need; you can think about this in many different ways.

 This post looks at the most common ways we define city boundaries.

The Post in a Table

The following sections describe and map these measures.

Municipal Boundaries

As noted above, municipal boundaries are a common delineator for what the city is and what it is not. The core municipality of greater Kansas City is KCMO, which has the following extent:

Image: Kansas City proper. Map data © OSM, Mapbox. Created in RStudio.

The Kansas City limits, as I discuss in my post on pre-annexation-area population decline, expanded rapidly through annexation in the second half of the 20th century. Much of the land at the northern, southern, and eastern extremes is still undeveloped, and—unusually—the city includes land in no fewer than four separate counties. The two most populous areas are from the river south to I-435 and east to the Blue River Valley, and from the river north to 152 Highway (though the northlands are now filling in rapidly).

Images: On the left, the inclusion of the Little Blue Valley in a 1960 annexation created an unusual pattern for a core municipality. The undeveloped land at center is inside city limits, while the built-up areas on three sides are not. On the right, the red dashed line marks city limits north of the Missouri. They encompass a sprawling suburban and agricultural area that built out almost entirely after the Second World War. There are four enclave cities north of the river: Birmingham, Avondale, North Kansas City, and Gladstone. Images via Google Maps.

MSAs

The Federal Office of Management and Budget produces the second most commonly used definition, the Metropolitan Statistical Area (MSA, said colloquially in the form ‘the Houston metro’). MSA boundaries are much larger than core cities, typically including anywhere from two to more than twenty counties. Their boundaries are measured by labor market relationships between counties: if more than a 25% of residents of an outlying county commute into the core county for work, or vice-versa, that county is considered a part of the metro area. 

Image: the Kansas City MSA. Map data © OSM, Mapbox. Created in RStudio.

MSAs are useful for comparing labor markets between cities. They also capture something fundamental about what a city is: it would seem strange to say that commuting a few miles a day between your house and your office involves a multi-city trip. Moreover, one of the core insights of urban economics is that cities primarily exist because they aggregate people into labor markets. The extent of the labor market therefore seems like a good way to measure what a city is. If you live in Overland Park, it's easy to work in downtown KCMO. But it would be very unusual to live in Overland Park and commute to Columbia. They're two different cities.

A major downside of MSAs, however, is that they are absolutely gigantic. MSAs include hundreds or thousands of square miles of agricultural and forest land. Go for a drive down rural roads in northern Platte County and try to convince yourself that you're still in Kansas City. It seems absurd.

CSAs

OMB has a related definition, the Combined Statistical Area (CSA), that’s based on a looser version of the same rules (15% employment interchange). CSAs are typically 1.5x to 3x as large as their corresponding MSAs. 

Image: the Kansas City CSA. Map data © OSM, Mapbox. Created in RStudio.

The Kansas City CSA encompasses the Kansas City MSA, the St Joseph MSA, the Lawrence MSA, and several micropolitan statistical areas (equivalent of MSAs for places <50k in population). Its 15 counties spread over more than 11,000 square miles, an area larger than Vermont.

To me, CSAs are close in size to what you'd call a region. There are 175 CSAs in the US; collectively, they contain close to 90% of the country's urbanized population and an even higher share of its GDP. If we were a nation of city-states I'd imagine they'd be around the size of a CSA.

MPOs

CSA and MSA boundaries correspond with county boundaries for ease of administration and analysis. Another city definition that uses county limits as boundaries are Metropolitan Planning Organization service areas. MPOs are transportation planning entities set up by the federal government to promote regional cooperation on road and transit projects. In most cases MPOs serve fewer counties than MSAs, focusing on those that contain significant transportation assets (major arterial roads, for example, don’t typically get built in semi-rural areas at the region’s fringe). 

Image: the MARC service area. Map data © OSM, Mapbox. Created in RStudio.

The Mid-American Regional Council, MARC, is Kansas City's MPO. MARC serves nine central counties of the Kansas City metro area: Miami, Johnson, Wyandotte, Leavenworth, Platte, Clay, Ray, Jackson, and Cass. With the exception of Ray, this is where the region's significant roadways (and therefore travel and commerce) are focused.

Urbanized Areas (UAs, GHSL cities)

MSAs, CSAs, and MPO areas are primarily concerned with activity—where people commute and how they do it. But there’s another way to look at what a city is, one that’s more intuitive for me. That is to look at the built environment directly.

In my mind, the city is where most land is used for buildings, and the countryside begins where most land is used for agriculture or nature preserves. So I look at the city (really the city-region) as being a continuous blob of built stuff. This is the satellite map enthusiast’s favorite definition.

At least two entities produce rigorous data based on definitions like this one.

The first is the Census Bureau’s Urban Areas, which I’ve seen well-defined as being created “by taking a city center and eating up adjacent blocks until it reaches areas that are not built up (or another urban area.)”* Blocks are included or not based on their housing unit density, a strong but imperfect proxy for the built environment overall.

You can think of UAs as being the entire central city plus its suburbs. The edge of the UA is where the subdivisions stop. Here's the map of the Kansas City UA:

Image: the Kansas City UA. Map data © OSM, Mapbox. Created in RStudio.

This version of the city limits is much smaller, and corresponds much more closely with the places I think of as being Kansas City or its suburbs. Unfortunately, it's missing Lee's Summit for a reason that is essentially technical (the Little Blue Valley cuts off enough of it from touching the other easter suburbs that it gets counted as a separate UA). I usually tack the Lee's Summit UA onto the KC UA when comparing us with other UAs.

Image: the Kansas City and Lees Summit UAs. Map data © OSM, Mapbox. Created in RStudio.

Combining the two gives a population of about 1.8 million, nearly as much as the MSA population (2.2 million) but on 1/10th the land. Here's what the edge of the UA looks like:

Image: the edge of the KC UA in southern Johnson County. Map data © OSM, Mapbox. Created in RStudio.

That's pretty good! With a couple minor exceptions (no doubt due to more recent construction), the UA scoops up all the continuous built up area of suburban subdivisions, institutions, and strip malls, and stops where it meets farm fields.

The Global Human Settlement Layer (GHSL) produced by the European Commission's Joint Research Centre has a similar definition of cities that it has used to classify their extent and development in every country on the globe. The GHSL methodology is fairly similar (though it uses satellites rather than censuses to classify land cover) and is based on a classified 1km^2 grid of the globe, as well as some population estimation work.

Image: the Kansas City GHSL area. Map data © OSM, Mapbox. Created in RStudio.

The GHSL area is now a bit dated, so it's not worth looking at the exact edges, but it covers ~60% of the area of the UA. The UA ends up bigger because it has a somewhat higher population cutoff and is based on Census blocks, which are not a perfect grid like the GHSL units.

UAs and the GHSL are density based measures, which raises an interesting question: why not use a higher density for your city boundary cutoff? After all, it feels odd to define cities in a way that includes all of their suburbs. People think of cities and suburbs as different things, with the former being denser and taller and more urban than the latter. If population density is a good metric for a cutoff at the edge of the suburbs it could be a good metric for cutting off the urban from the suburban, too. Unfortunately no one publishes a definition using a higher population density cutoff.2

Media Markets 

Just for fun, I’ve included one final definition, this time a cultural one. Media markets once were much more determinative of regional culture than they are today, but over very large areas they distinguish different US regions from each other. These days they matter most for local news, sports distribution, and tv marketing. 

Image: the Kansas City Media Market. Map data © OSM, Mapbox. Created in RStudio.

Media Markets are a good way to delineate who and where shares Kansas City's culture. But they're too large to really connect people—my friends in Sedalia may get local channels but they don't claim any connection with the city. And owing to Kansas City's relative isolation (the subject of an upcoming post on here), adding all that land doesn't substantial raise the population.

Fin

That covers all the common ways to define the limits of Kansas City and its region. Here's the comparison table again, in case you skipped it earlier.

Again, there's no measure here small enough to just look at urban-ness. But if you did measure us that way, Kansas City would be tiny. As I've written before, we just don't have much density in our core.3

Which one of these definitions you pick when you compare us to other cities matters a lot. Each has upsides, each has downsides. For this blog's purposes—mostly analysis of urban development, economics, and demographics—the UA or the MSA is typically most useful. I'm careful to mention which I'm using in various posts.

If I've missed a definition you like, feel free to comment or message me on twitter.

This is unfortunately a much more annoying question to answer. The only reliable source of data on homelessness in the US is HUD, which contracts with Continuums of Care (CoCs) to go out and count the number of literally homeless people in the US once per year. Those figures are widely criticized for a number of reasons, but they’re all we’ve got.1 

Unfortunately, this data isn’t very good for comparing between places. CoCs were allowed to incorporate somewhat arbitrarily and have extremely varied geographies. If you look at a map of CoCs you’ll see that some cover tremendously large populations, in some cases whole metro areas, while others are split by county or even locality. The Houston CoC oversees the entire Houston MSA; the DC metro has five separate CoCs. 

So comparing between places is difficult. So is program evaluation in general—housing markets are regional, so if you have a sub-regional CoC it's difficult to evaluate its performance against selection effects. Perhaps your CoC is great at keeping homelessness rates down, but if it’s sub-regional it’s more likely that the area is just good at pushing the homeless into the next suburb. 

People who write seriously about this stuff tend to narrow their data down to the CoCs that have similar geographies, like citywide CoCs (eg St Louis City) or countywide CoCs (conveniently, eg St Louis County). This approach gives you cleaner data, but it doesn’t fix the selection effects problem or let you compare across most cities. And neither approach works well here. As the map shows, our CoC boundaries don’t follow either city or county lines.

The boundaries of the Kansas City CoC are typically random. Johnson County has one CoC and the rest of the built up part of the metro has another. Less developed parts of the MSA fall into Missouri’s statewide CoC. The JoCo CoC counted 250 homeless people in 2024 and the larger CoC counted 2,215.¹

Fortunately, there’s a decent basis for comparison here. Given that our two CoCs together encompass the whole built-up area of greater Kansas City (there’s an OMB definition for this, the Urban Area), and almost certainly an even higher share of its homeless population, I’ll compare us here to close-to-whole-of-Urban-Area CoCs. From a casual look at HUD’s map, here is a list:

Seattle, Omaha, Phoenix, Houston, Denver, Columbus, San Diego, Las Vegas, Charlotte 

There’s decent variety here. There's a growing midwestern city (Columbus), two cities that have a lot in common with us (Denver, Omaha), three sunbelt cities, and two west coast semisuperstars. That’s enough to give a sense of where we stand overall in the country. 

Conveniently, the Brookings Institute has a blog from last year that runs the numbers for these places (in a larger list, one that has the geography problem and is therefore not useful for drawing comparisons). They report these figures as the rate per 100k. 

Combining figures for the two Kansas City CoCs gives a rate of 146.3 per 100,000 people. This table shows how many homeless residents we would have at other cities’ homelessness rates.

Once again, we do very well as a region. That’s not surprising—as my housing post covers, we have a modest cost of living. Housing affordability explains a lot of the variance between these cities. Seattle and San Diego have price-to-income ratios of 7.3 and 9.5, respectively, while ours sits at 4.2. 

It’s pretty clear that housing costs are a much larger driver of homelessness rates than is CoC performance. CoC performance does matter though. Unfortunately, I don’t have a sense of how it contributes here. Perhaps I’ll dive into it in a future post. 

If Kansas City had Seattle’s homelessness problem, we would have 4x as many people living on our streets. If we had Houston’s, we’d have 1/3rd. 

We should strive to be more like Houston. Build!

1. Annexations

The typical population trajectory for an older northeastern or midwestern US city is of continuous growth from its founding until 1950, then several decades of decline from white flight and suburbanization, followed by growth after 1990 or so. The most fortunate (New York) have recovered all their lost population; the least (Detroit, Baltimore) are still spiraling downward. 

Kansas City’s population trajectory doesn’t follow that pattern. Yes, our growth nearly stalled after 1950, but it didn’t begin to decline until three decades later. Until 2020 our population peak was 1970, 20 years after similar late-19th-century industrial boomtowns like Chicago, Saint Louis, Milwaukee, Cincinnati, Pittsburgh, etc. Ours looks more similar to that of Denver, a much younger city, or even Columbus, a city of an entirely different kind.

There’s a simple reason for Kansas City’s missing decline decades: annexations. Before the Second World War, Kansas City covered just 59.8 square miles. Between 1947 and 2010 it swallowed up its suburbs rapidly, eventually adding nearly 259 square miles. That level of annexation is unusual for ‘old’ US cities; Detroit, for example, lost its practical ability to annex in 1926. Saint Louis hasn’t expanded its boundaries since it separated from the county in 1876. 

The huge growth of our city limits makes it hard to compare like for like. Did Kansas City really avoid the worst of the urban crisis after 1945? Or did we just grow enough to mask it? There are clearly neighborhoods in the city that have declined, but by how much? Does our urban core’s experience look more like that of Detroit, with extreme flight and disinvestment, or are we more like Denver, a 20th century success story? 

 It’s surprisingly difficult to find solid answers. This isn’t covered in any planning department work (that I’ve been able to read), or on city websites, the local media doesn’t seem to have covered it before, and even when Strong Towns did a big report on the Kansas City suburban experiment a few years back they didn’t track the figure down. So I did, pulling census data from NHGIS and adopting it to the old boundaries. I’ll walk through the specific steps at the bottom of this post.  

2. Population Collapse

The population of the pre-1945 boundaries of Kansas City peaked in 1950 at 430,535 people. In 2022, it was (approximately) 187,902. That means pre-annexation Kansas City has lost  242,633 residents, 56% of its previous population. 

A 56% population decline is an incredible figure.  

This chart shows the population trajectories of all major US cities that lost a significant share of their population after 1950. Each of the cities on this list have generated national headlines as they’ve declined, and those at the top have practically become synonymous with urban failure and population loss. None annexed substantial land or population after 1945. 

If not for the growth of areas annexed after 1945, Kansas City would be among the very largest population losers of any major city in the US. In percentage terms, we lost the fourth most population; in absolute terms, 7th. Had Kansas City not annexed so much land it would be near the top of every urban decline listicle, CityLab op-ed, and glossy magazine think piece written in the past fifty years. City planning students would pile up masters theses on rethinking the city’s land use; think tanks would talk about ‘right-sizing’ services. Conservatives would use its name as a pejorative; political careers would be made on ‘turning the city around’. It’s hard to overstate how different Kansas City’s reputation would be without the postwar annexations. 

3. 187,902 Fewer People 

These maps show the extent of population decline within the old city boundaries. In 1950 several tracts reached a density of near 40,000 people per square mile, mostly on the east side of downtown. The core residential section stretched from Downtown to the Plaza area, State Line Road to roughly Swope Parkway and for the most part was built to transit-supportive densities. Higher density areas traced the path of the extensive streetcar network; these neighborhoods were generally built out with a mix of small apartment buildings, single family homes on small lots, and one-to-two story shops on commercial streets. The colonnaded apartment buildings lining the extensive network of boulevards gave the city a distinctive feel. 

By 2022, this core region had lost hundreds of thousands of residents. Downtown itself and the near east neighborhoods—Paseo West, Parkview—lost the most people in absolute terms, while neighborhoods further southeast generally lost the highest share of residents. Highway construction and urban renewal was the most common cause of severe local population loss; according to historian Kevin Fox Gotham, at least 12,000 residents were displaced by freeway building from 1950-1980 alone (it’s hard to say how many left the city entirely, of course). 

Newer sections southwest of Brush Creek were much less affected; the dynamics of white flight and racialized decline were held back there to a much greater extent than elsewhere in the city. Population decline here was almost entirely driven by falling household sizes. 

If you’ve never thought about demographic change you might not realize how much neighborhood population loss is driven by changing household compositions. The median household size in the US dropped from an average of 3.5 to 2.6 people per household during this period. Household sizes dropping by ⅓ explains why many places don’t look like they’ve lost population—there are nearly as many homes, they just have fewer people living in them. To keep its 1950 population, the old city would have needed to substantially add to its housing stock.  

This next graphic shows the impact to the built environment of population decline in one six-block area. The Census Tract home to these blocks lost 91% of its population between 1950 and 2022, dropping from 5,397 to just 515 residents. In just this small area the city bought and demolished two blocks for US-71 right of way, a church replaced a face block of single family homes, and a number of vacant lots replaced homes throughout.  

I posted this next photo on twitter recently, of Kansas City's core looking north right as the collapse began. There were roughly 50,000 people living in frame here (south of the river) in 1950. Fewer than 20,000 call this part of the city home today.

4. Learning From A Hidden Crisis 

Population loss is bad for remaining residents

Falling population creates a whole range of problems for people who don't leave. 

Property values drop, worsening racial inequality and leaving people whose homes are their largest investments without the appreciation they hoped would fund retirement or be handed down to children. Abandoned homes catch fire, either on their own or through arson by vandals. Unmaintained homes and property attract crime and vermin. Not knowing what else to do with vacant, distressed homes, cities incur large costs demolishing them. Budgets stretch as assessments fall and infrastructure per capita rises. Small businesses move away, or close. Social networks fragment. Wealthier residents, who have more options for where they live, leave first, taking their social resources and buying power with them and leading to rising poverty rates. All of these issues take a psychological toll on residents, serving as an additional push to move away. 

Annexation served to mask the old city’s decline but didn’t prevent it

This is a subtle point but I think the most important one here. 

We don’t know if the annexations increased or decreased population decline in the old city boundaries. My guess is neither, but here’s my best case for each:

Increased decline: people living in the declining area of the city saw that they could move to nicer housing without the increased psychological and bureaucratic hurdles of leaving their hometown, and did so more than they would have otherwise. 

Decreased decline: the growing population outside the distressed core kept the city better resourced than it would have otherwise been by propping up the tax base. This prevented the kind of public service collapse that cities like Detroit saw during their depopulation eras, where trash wasn’t picked up, roads weren’t paved, and police and fire sometimes took hours to respond to emergencies. In Detroit’s case, decline eventually brought it to bankruptcy. Kansas City’s fiscal picture was stronger and therefore it remained more livable, even in neighborhoods that were losing population quickly.  

Of the two cases, the latter is the more credible. It is very obviously true that a decline in neighborhood quality—driven in part by lower quality public services—has been one of the drivers of urban out-migration in US cities over the past seventy years. It is plausible to me that on the margin decline may have been worse absent the annexations. 

Still, there’s good reason to think the effect of the annexations on decline in the old city is not very large. Academic work has generally found six major causes for urban population loss after 1950, none of which match those cases well. They are: favorable mortgage lending standards for suburban housing, increased access to suburban land because of funding for expressways, ‘white flight’ away from neighborhoods that African Americans were gaining access to in the context of rising incomes and civil rights legislation, a sharp rise in violent crime in urban centers, the creation of higher quality and closed access school districts in the suburbs, and economic restructuring with deindustrialization and the growth of the service economy (old job centers declined, and new job centers rose in different geographies). Public service quality is downstream of those major structural shifts, and likely played a minor role in population movement. The exception to this is school quality. 

Schools are enormously important

One thing that didn’t change with annexations is the boundaries of the Kansas City school district, which has expanded very little beyond the old city limits. A combination of new and merged districts serve the annexed lands to the north and south of the old city. As the neighborhoods it serves declined and became increasingly disadvantaged, the KC school district suffered declining test scores, enrollment, and facility quality. And because they are extremely capital intensive, schools are highly susceptible to decline spirals (fewer students = less funding = buildings fall apart = fewer students = less funding = ). 

It’s not plausible that an expanded school district could have alone stopped population decline in the old city. But on the margin, expanding its boundaries likely would have helped (though there is also the flipside to consider - doing so could have slowed growth in the annexed lands).   

Comparative governance implications

I’ve long been sympathetic to the idea that struggling legacy cities could improve their fortunes by merging with their suburbs, normalizing resources across what have become very socioeconomically unequal geographies. This finding has made me reassess that view.

It’s naive to imagine that you can fix the deep structural issues driving decline in neighborhoods like these solely by bringing more outside investment into them. This has failed in other big cities — see the ongoing debate about demolishing Detroit’s Renaissance Center. Planners and economic development bureaucrats in my opinion overrate the potential of place based policy; in many cases it’s a better idea to invest in residents directly than to dump money into ‘revitalization’ projects.     

The lack of school district expansion does limit the value of the comparison, though. City-county mergers in Louisville, Jacksonville, and Indianapolis remain better cases. 

Narratives matter

This one is hard to pin down as an empirical matter but I do think that popular narratives about cities are important for their economic prospects. Ecdev consultants increasingly talk about ‘storytelling’ as a strategy to get people excited about a place, but really it’s the stories that already exist and are broadly known that influence how people feel the most. Positive narratives (such as ‘we are a growing, vibrant city’) can inspire pride and excitement, encouraging people who live in the city to stay and contribute and people in and from beyond the region to want to visit or invest in it. Conversely, decline narratives can be extremely difficult to change. When I told people I was moving to Detroit five years ago, despite a decade-long campaign by the city and developers to tell people it was coming back, my friends looked at me like I had three heads. Legacy city biases are deeply rooted. Even if it wasn’t the intention of the Kansas City annexors to create an illusion of resilience, it was effective. 

As an aside, the same thing is true of Kansas City Kansas and its Unified Government. Downtown KCK is among the most distressed centers I’ve ever visited, but when I talk about KCK with people here they think about the growth out in the county and so don’t have a negative impression of it as a city.

5. Solutions

I may write a future post about what the city can do for its declined neighborhoods. I’m new to the region and don’t know enough about the politics locally to have an informed opinion. In general, my recommendations for governing declining cities are:

1. Invest in people more than place

2. Make it as easy as possible to renovate houses

3. Be as efficient and communicative as possible about solving residents’ needs. Ask them what they need frequently

4. Make sure your tax foreclosure process isn’t shooting you in the foot

5. Devote every dollar you can find, and make every approval process as easy as you can, for infill construction (tax breaks, fees waived, cash grants for certain infrastructure or land clearance expenses, flexible zoning, fast approvals, etc)

6.  Don’t fall for single big projects that claim to be quick fixes

7. At the same time, don’t let your biggest problem spots define your agenda. Invest in strong neighborhoods and green shoots

6. Fin

As always, reach me at @2024dion on twitter with comments

My proposal: a three-level framing of metro areas that brings us closer to actual comparisons. The first level is the core city, the original location that lent its name to the entire region. The second level is the urban area, which is comprised of contiguous built-up areas with at least 1,000 people per square mile residing in it. The third level is the area beyond the urban area but still within the boundaries of the broader metro area (which draws its lines at counties).

To clarify, when Saunders says urban area and metro area he is referring specifically to Census-defined Urban Areas and Metropolitan Statistical Areas. The city is the city proper. The suburbs are the Urban Area minus the city proper. The exurbs are the rest of the MSA.

This definition works nicely for quick comparisons across American cities but it ignores a lot of complexity, especially with city limits.1 Some US cities are huge, incorporating urban, suburban, and even conservation lands. Jacksonville covers 875 square miles. Others are tightly limited to their core and immediately surrounding neighborhoods. Boston has just 48 square miles and doesn’t include many of the region’s dense urban neighborhoods.

Applying it to Kansas City, you get this map.

Kansas City has half a million people within its city limits. It covers a relatively large share of the Urban Area on the Missouri side, leaving off just Independence, Blue Springs, Oak Grove, Raytown, Lees Summit (Lees Summit actually comprises its own UA but I always fold it into Kansas City’s because it is so close and so linked otherwise), Raymore, Grandview, Liberty, Gladstone, Smithville, and Parkville. A number of other very small enclave cities carve chunks out of the Northland.

Those comprise its Missouri suburbs. On the other side of the state border, the suburbs include most of Johnson County out to Olathe (with a break between I-435 and K-7) and Wyandotte out to the Speedway. In total, the suburbs house 1.3 million people.

The rest of the 14-county MSA falls into the exurbs, which is the least precise of these definitions, incorporating small towns, scattered and large-lot subdivisions, and traditional-looking rural farmsteads. 400,000 people (~18% of the metro area’s population) live in the exurbs.

A Somewhat Different Approach Works Best for KC

This scheme is no doubt useful for comparing between cities but as I noted before ignores a lot of the complexities of city limits. Kansas City in 1950 covered just 80 square miles. The decades of annexations that followed have swollen its size to more than 300 square miles.

To me, the part of KC that feels ‘urban’ is the old core as it existed before this rash of postwar annexations. This part of the city—bounded by the Missouri to the north, State Line Road to the west, the Blue River to the east, and ~77th St to the south—was built out before 1945 and has a traditional American city feel with gridded streets, relatively small lots, pedestrian and streetcar-oriented streetscapes and neighborhood shopping districts. Areas annexed to the city after this period tend to be more automobile-oriented in character, with winding subdivision streets, generous arterial roads, and strip retail on major corridors.

This narrows the city population to <200k and increases the suburbs to 1.6 million. It’s not perfect either—some neighborhoods near the core don’t have the urban building blocks I’ve described, and some neighborhoods elsewhere in the metro (downtown Lees Summit, central KCK, etc) do. But it’s basically my mental model of these three categories here.

I have a forthcoming post looking at this urban core area as a distinct place. Stay tuned.

One of the first things that struck me about Kansas City as a new resident is the missing rivers. The Missouri, I knew, is the raison d'être for the rise of a city here. And the Kansas is a significant regional channel that famously divides Kansas City from the other Kansas City. The rivers are defining characteristics of the city when you look at it on a map, as I often did before the move. But from the ground, living here day to day, you hardly see them, hardly interact with them. Maybe you drive across on a highway bridge on your commute or to the airport but that’s about it. There is only one stretch of riverfront that’s really accessible (Berkley Riverfront) and it’s currently a construction site that’s mostly cut off from the rest of the city.

You can see the reason for this from a satellite map view: industry. The Missouri and the Kansas are lined with mile after mile of industrial buildings—warehouses, power plants, factories, even an airport. These facilities form a thick barrier between the residential and commercial city to one side and the waterfront to the other.

Of course, every older American city has some waterfront industry. That’s much of the reason they were founded where they were—to onload and offload goods transferring between regional and long-distance transportation networks. The geography of 19th and 20th century shipping networks created American cities to a greater extent than any other factor.

What’s striking about Kansas City, though, is the persistence of its waterfront industry. Most cities saw a steep decline in the utility of this type of development after the invention of railroads and then the rise of trucking across a ~60 year span in the late 19th and early 20th century. After a lull, these cities converted their outmoded waterfronts to other uses—parks, condos, office towers—that turned the views and space into amenities that improve the quality of life of urban residents. Kansas City isn’t like this. Its waterfronts stayed industrial.

The basic reason for this is the region’s topography and the influence that it had on our railroad and industrial geography. The two rivers both cut channels through higher ground as they flow through the Kansas City region. These channels are lined with river bottoms—flat, sedimented floodplains the rivers have meandered through before changing course over many millenia. Behind the bottoms, which are anywhere from a few dozen feet to several miles wide, are bluffs of 50’ or more at the height of the surrounding land.

Though they line the rivers, our industrial facilities have largely not been reliant on waterborne transportation.1 What they really trace is the routes of the railroad lines, which themselves follow the rivers because their bottoms are both flat and out of the way of the commercial and residential city that sprung up on the bluffs behind them.

Though there was some decline in the value of railroad adjacency for industry in the late 20th century, areas like these never became as desolate as urban docklands did. And even where they did, the railroads remained, barriers unto themselves.2

The persistence of this barrier between the people of Kansas City and their riverfronts is a source of significant weakness for the region. Converted waterfronts in other cities make for iconic recreational spaces and urban vistas, attracting billions of dollars in real estate development and millions of visitors each year. They’ve been instrumental in turning urban living into a viable middle-class lifestyle preference once again. But outside of the small-but-exciting Berkley Riverfront, Kansas City doesn’t have anything equivalent. Of course, we have other strengths. But the delta between the river city we could be and the geographically unremarkable city we are is a bit maddening. A piece of bad luck.

/fin

More view of our riverfronts:

Here's a simple post: a series of dot density maps showing the distribution of population by race in the Kansas City metro area from 1950 to 2020. Each dot equals 100 people; the dots are randomly distributed within the tract those 100 people live in. The modern KC city limits are in black; the boundaries of the five central counties in the region (Jackson, Johnson, Wyandotte, Platte, Clay) are in gray.1

You can learn any number of things from this data.

For starters, the region has become more diverse, but slowly. The white population dropped below 70% for the first time in 2020 (US Average: 58%) but only 12% of our population is Hispanic (US average: 19%) and 4% are Asian (US: 6%). The Black population share has risen slowly to 14% (US: 12%). Other groups comprise less than 1% of our population (US: 4%).

Here is a chart showing how the region has diversified. Note that this data only includes the five core counties of the metro area, not the metro area as a whole. Note also that 1990 is the first year we have relatively accurate counts of Hispanic residents in this geography.

Setting race and ethnicity aside, the gif also gives a nice visualization of how the region has expanded. Decade by decade, our population has spread out. The 2010s were the first decade since 1950 that our pw density started to rise again, and we still sprawled mightily.2 I've written previously about MARC's regional population projections; the upshot is that we should expect the region's extent to keep moving out at a rapid pace for the next few decades.

Here are the maps decade by decade for ease of comparison. In reverse order:

2020

2010

2000

1990

1980

The 1980 Census included a separate question about Hispanic ancestry for the first time, and it appears that very few people filled it out. According to NPR, this was common -- the formatting of the survey confused respondents. I only found 656 in the entire five county area for the 1980 Census. By 1990, this figure was recorded accurately.

1970

1960

1950

A cautionary note about interpreting the 1950 map: the Census Bureau introduced tracts (which this project is based on) for the first time in 1950, and only built them for close-in areas of large cities. Because US cities really exploded outward in the next decade, these areas are fairly small. For Kansas City they cover the city proper, all of Mission, a bit of Shawnee and Overland Park, a bit of the northlands, and some of Independence. Use this map to see how the population of the core city changed from 1950; don't use it for regional interpretation.

You can find dozens more stories and patterns in these maps, and I hope eagle-eyed readers do. Please feel free to email me at ddavoli.234@gmail.com or find me on twitter @2024dion if you have any questions, or data/visualization requests.

The simplest way to compare the cost of living between different metros in the US is to use the price-to-income ratio.1 This easy calculation (median home price divided by median HH income) is the best COL metric because a) housing is the biggest expense for the typical household, and b) housing costs and wages vary regionally more than almost any other prices in our economy. Regional variation in the cost of most services (like childcare) is downstream of wages; most durable goods (like cars) cost ~ the same everywhere in the country. 

Your choice of home price index will impact the PIR you come up with (there are a bunch of ways to think about the average cost of a home and several firms and gov agencies give their own figures) but they’re all similar enough and because it’s a metro comparison tool it doesn’t matter which you use so long as you’re consistent. For this post, I looked at a dataset assembled last year by Harvard’s Joint Center for Housing Studies available here. 

This gives a PIR for the Kansas City MSA of 4.2. The median home price in the Kansas City metro area (~$330k, to give an estimate of their paywalled data) is 4.2x our median household income (~$78k).

Because housing prices don’t directly correlate with buyability, there’s no standard for what a good or bad PIR is. A reasonable rule of thumb is that a ratio of 2-4 makes for a low COL market, 4-6 medium, and 6+ high. By this measure, KC is just over the limit for a low COL city. 

Per the report, the PIR for the US overall was 5.6 in 2022, up from 4.1 in 2019. You can tell a lot of big stories from that jump *cough cough 2024 election*.

Past

The JCHS dataset gives PIRs for every metro area from 1980 to 2022. Here’s Kansas City:

We’ve long been a low cost of living region, with housing prices only rising significantly in three periods: during the runup to the financial crisis, after 2010 as inventories grew slower than population, and then sharply after 2020 in a pattern that fits the US’s overall. It’s that final shock that pushes us out of the affordable bucket; its long-run dynamics aren’t yet clear but fortunately there are signs elsewhere that pandemic era price growth has peaked. 

Other Cities 

One interesting question is what housing prices would look like in Kansas City if we shared the cost crises of coastal cities. It’s exceedingly unlikely we reach those levels of unaffordability—the KC region has plenty of land to sprawl onto and is projected to keep going at a moderate pace. Still, the comparison is revealing. How bad could things be?

This table compares the PIRs of all 56 MSAs with a 2020 population of one million or more. The column “Imputed PIRs” gives what KC housing prices would be if we had the cost of living of these cities, given their current PIRs. I’ve also added a 2020-2023 population change column as an indicator for the general state of each city’s economy. 

(To spell this out a little more, I took KC’s median HH income and multiplied it by the PIR of each city. For example, if you multiply our PIR (4.2) by our median HH income ($78k) you get $327,600. Again, a high PIR tells you that housing prices are out of whack with household incomes.)

If Kansas City had the same cost of living as Honolulu our median home value would be $943,800, nearly triple what it is today. Now, Honolulu’s situation is somewhat unique: its tourism-driven economy doesn’t create particularly high wages, it’s tightly geography-constrained, and its housing stock is highly prized by the wealthy for second homes. 

San Jose doesn’t have those problems, and yet its PIR—12—would give us a $936,000 median home price. Their unique combination of a world changing economy and the growth strategy of a midsized suburb has created a cost of living disaster so bad and so stupid you don't know whether to laugh or to cry. 

To pull out some peer-ish cities:

Denver: PIR 7.1, imputed cost $553,800

Nashville: PIR 5.8, imputed cost $452,400

Charlotte: PIR 5.7, imputed cost $444,600

Jacksonville: PIR 5.6, imputed cost $436,800

Indianapolis, VA Beach: tie!

Columbus: PIR 4.1, imputed cost $319,800

St Louis: PIR 3.5, imputed cost $273,000

Pittsburgh: PIR 3.1, imputed cost $241,800

Kansas City has the 41st highest PIR of the 56 cities. That’s not bad! In absolute terms, as I noted up top, our cost of living is getting higher. But in relative terms we are doing pretty well. If you look at the population change column, only five of the cities with a lower PIR are growing at a similar rate as us. Most are cheap because they are stagnant; we have managed to stay cheap even as we have grown. 

Fin

Housing prices have been rising in the Kansas City region for the past several years but our cost of living is still relatively low, particularly given our continued economic and population growth. Still, there are steps we should take to ensure that costs don’t continue to creep upwards. The region does an admirable job of attracting high-value employers and jobs, though of course coordination could always improve. More importantly, housing reform, particularly in the neighborhoods people most want to live in, remains a desperately underrecognized regional need.

The 2024 Census population estimates for MSAs and counties were released this week. They found a one-year increase of 24,817 people in the Kansas City MSA, outpacing last year's 19,305. Here are a set of takeaways for our region.

All charts in this post are my own with data from the US Census Population Estimates Program

Quick Hit 1: Kansas City is growing faster than projected

Last year, MARC released a draft of its upcoming population and jobs forecast, which it produces to inform MPO planning. That forecast projected that the region would grow by 54,000 people from 2020-2025. But according to this latest release, we have already added 58,291 in four years. If 2025 is like the two years before it, we should expect to end 2025 with closer to 80,000 new people than 54,000.1

The 2024 population estimates have come in higher than anticipated in many places and the Bureau has even gone back to revise earlier 2020s estimates upward. They attribute this to higher than expected immigration during the Biden administration and note that it is unlikely to continue. Given this, it's reasonable to expect that Kansas City will revert back to growing at the projected pace in the second half of the 2020s.

Quick Hit 2: 2024's gain may have been the largest since the great recession

Decennial Census counts differ in methodology from the annual estimates (the decennial is mailed to every household while the annual estimates are compiled from surveys, other data sources, and statistical analysis). Therefore you should ignore the one when considering the other.

Without 2020, a decennial year, 2024's growth is the largest since before the great recession. That's quite impressive and speaks to our strong economy, healthy investment in place quality, low cost of living, and the good performance of our economic development institutions. But given demographic trends, it could be a high water mark for much longer than the 2007 figure's 16 years.

Quick Hit 3: All core counties saw increases

We also got county-level population change data. Here's cumulative change since 2020:

Jackson County's trajectory is the most interesting: from 2020-2022 we lost population before recovering to become the second largest gainer over the four-year period. Without more granular data it's difficult to know what happened exactly but it's likely that core city population flight during the pandemic--which was a trend for cities nationally--contributed. The same is likely true for Wyandotte, which saw a much sharper fall before more than rebounding.

Johnson County's extremely strong growth is an output of the same trend. Weak core, strong fringe.

Quick Hit 4: This tells us little about the local distribution of population

The distribution of population change informs a lot of questions not answered by its absolute value. For example, is remote work pushing people away from employment centers? Are neighborhoods near the core, where people tend to lead less emissions-intensive lifestyles, growing or shrinking? Are hyperlocal place-based investments (like the streetcar) paying off, or a financial albatross? The county data hints at some of these trends, but without more granular data we can't fully say.

Quick Hit 4: What about our peer cities?

I haven't yet settled on what I think are Kansas City's peers, so this set is:

1. Cities +- two rankings by population in 2020 (Cleveland, Cincinnati, Columbus, Las Vegas)

2. Cities with strong similarities or cultural ties to KC (Denver, Omaha, St Louis)

As you can see, our population growth through this period was quite respectable. While our neighbors across the state shed 8,000 residents, we added 3/4ths as many as Columbus (an economic powerhouse with both a state capitol and a strong state university) or Denver (a larger, more attractive sibling of ours, economically and culturally). We outgrew Cincinnati, though not by enough that we will challenge it for the 30th spot on the metro size rankings by the end of the 2020s.

Oh and shoutout to Omaha--they crossed the 1,000,000 resident threshold for the first time.

Quick Hit 5: Don't expect it to continue

I've covered this in a couple places here, but there isn't any good reason to expect that this strong one-year growth portends a longer-run rise in our growth rate. As the Census Bureau notes, the entire country grew faster than expected over the past four years because of a surge in immigration. That surge is over.

Moreover, the US is entering a long-term growth tailoff. In fact, the whole world is. Fertility rates are falling much faster than projected in most countries, including those with very different economic, social, and political characteristics from each other. Our TFR is now just 1.6, and without a large increase in immigration (which would be a challenge even without reactionaries trying to close the border, given the global fertility falloff) our population trajectory will start to level off and then fall in the next few decades. The CBO now projects that the US population will peak as soon as 2069 after years of very slow increases.

Cities, particularly those with strong existing growth and good economic prospects, will continue to outperform national population growth for so long as the modern economy grows on agglomeration. But they should expect growth to slow, too.

Another thing you can do with the jobs dataset I used in a recent post is to identify employment hotspots in the region. Employment, as I covered there, tends to concentrate to a much greater degree than does living. Jobs cluster into the region’s hospital districts, office parks, megamalls, and in strip centers along major arterials. This post will focus in on those clusters, looking at their extent, commute patterns, and the makeup of their workforces.

Kansas City is a weakly cored region with equally weak subcenters.1 Because we historically eschewed building a mass transit system, instead investing heavily in our freeway network, and because we are still relatively small and have few geographic constraints to outward growth, our office core doesn’t reach the skyscraper-and-subway densities that Chicago or even St. Louis does. Downtown Kansas City exceeds job densities of 50,000 psm for only a few contiguous blocks. Though the majority of white collar work takes place in our suburbs, we lack a true edge city like Clayton outside St. Louis or Tysons outside D.C. The closest thing we have is the Plaza and the large corporate parks along i435 in Johnson County. Corporate Woods, Aspiria, and Executive Hills contain tens of thousands of jobs but are suburban in form with large parking lots and buildings set back far from arterial roads. The Plaza is more plausibly urban, and will soon be served by a hybrid streetcar/light rail line, but has many thousands fewer workers.

Clusters

For this post I used LODES Origin-Destination and Workplace Area Characteristics data to identify job clusters, map their commute sheds (the neighborhoods where their workers live), and explore employment patterns within each of them. 

There are a variety of cluster-identification statistical methods available (spatial autocorrelation groupings by similarity, differences against adjacent areas, subtypes, etc) but what I wanted to capture with this analysis is less mathematically precise areas and more subjectively defined clusters based on common elements. These included places with distinctive topographic or geographic features (like the river bottoms along the Missouri), alike employment types (the Blue River Valley tracts are predominantly industrial), or relationships to significant regional infrastructure (the Independence and Blue Springs cluster is not quite contiguous but owes success to easy i70 access). I loosely applied those standards and the following definition: a job cluster is a group of five or more mostly contiguous block groups containing at least 1,000 jobs each and 15,000 jobs total.

This gave sixteen distinct job clusters.2 They are mapped here.

These sixteen clusters, as the map highlights, encompass just 8.49% of the region’s land while hosting 57.8% of its jobs. They are fundamental units for any regional transportation or economic analysis.

The three largest clusters are in Eastern Johnson County along i435 (offices, power centers), the Kansas City core (offices, hospitals), and along the southern extent of i35 in Overland Park and Olathe (offices, retail, industrial parks). Eleven of the sixteen have large shares of industrial employment, while six include dense office nodes. Ten are in whole or in part within the city limits of Kansas City; just three are north of the Missouri River. All except the Plaza and the Kansas and Missouri bottoms are either strung along major freeways or sit at their intersections. 

These charts give the total employment characteristics by sector, education attainment, race, and ethnicity3 of employment for jobs in the sixteen clusters:

The bulk of this post will look at each cluster individually.

First, the Kansas City Core 

The KC Core includes all of the Kansas City Loop, the Crossroads and Hospital Hill, Penn Valley Park and the Metropolitan Community College behind it, Crown Center, Hospital Hill, the greater 18th and Vine and Paseo West neighborhoods, and the River Market. 

Workers commute here from all five counties in close proportion to their population sizes (except Jackson and Johnson—Jackson tilts the scales between them). This cluster has the broadest commute shed, likely for two reasons: first, professional work done in class-A office space is the most specialized, so it attracts people who live across the metro area. These workers are unable to take similar jobs close to home; their occupations exist only in one or two places. Second, it is both at the center of the region and best-served by our hub-and-spoke freeway system. There are few places in greater Kansas City that are more than a 45 minute drive from downtown. The largest share of potential commuters here translates to the largest share of actual commuters.

There are collectively 79,917 jobs here, predominantly in white collar and healthcare roles. The graphic on the post lists the five most common of the 20 2-digit NAICS codes; more granular data is not available in LODES. Approximately 60% of total jobs are in office-based white collar professions; healthcare only rates at the top because it has a single NAICS code. Unfortunately, another limitation of LODES is that it lacks granular wage data. I suspect that this and the East Johnson County cluster have the highest mean and median wages, but can’t be sure from available data.

Plaza

The Plaza Cluster includes the Plaza itself and its western border, the offices along Mill Creek Parkway, parts of Westport, and UMKC and half of Rockhurst U (an annoying bg cutoff leaves the other half in a residential area). This is one of the smaller clusters, with just 22,000 jobs, and is the smallest besides downtown KCK to be almost exclusively white collar and retail employment. Consequently, it has among the highest rates of college attainment.

What’s most interesting though is the commute shed. Workers in the Plaza Cluster disproportionately live nearby, in the narrow north-south stretch of Kansas City that’s home to most of the city’s south of the river diverse middle-class neighborhoods. This is interesting and I don’t have a good explanation for it. University employees walking to work can’t alone explain it.

East Johnson County/I435

East Johnson County is the largest of the 16 clusters and likely the highest income, with tens of thousands of college educated, white collar workers on corporate campuses. Aspiria (T-Mobile), Corporate Woods, and Executive Hills dominate a pleasant landscape of well-maintained ridges and fountained retention ponds. There are several large shopping centers with more than 6,000 total workers; I bought a dress shirt here recently and was kinda sold on the lifestyle. 

Commute patterns are mostly unsurprising; Johnson County is overrepresented but it draws from across the metro area. Also unsurprising is that this cluster has the highest percentage of Asian workers. 

One thing I will note about commutes though is that, as a recent arrival in Lee’s Summit, I naively assumed that professional class LS residents would work more in Johnson County than in downtown KC. My logic was down to highway patterns: i470/i435 form a direct route between the two (as does M 150 Highway at the bottom of Jackson County) while the link to downtown is less direct. I don’t have granular enough data to be sure but it seems this is not the case. I wonder if this will remain true as growth in LS keeps shifting south.

Blue River North

Switching gears a bit, let’s zoom in on the northern Blue River Valley. My visualization process worked worst for this cluster, likely because it’s relatively small, in an area dense with block groups, and has a fairly well-spread workforce. Alas.

The major job centers here are the 187 bed VA teaching hospital, several small heavy manufacturing operations (engine parts, mining equipment, agricultural implants, etc), the sports complex, and a variety of light industry shops. It’s an older industrial district that, not suited to the spatial requirements of modern facilities, is dominated by smaller and low-productivity industrial sector firms. A lot of metal scrapping and auto storage. 

Of the 16 job clusters it has the highest share of Black workers, a product of its adjacent workforce and industry sectors. 

Blue River Valley South

The southern part of the Blue River Valley cluster stretches from Swope Park to the Grandview Triangle. Two major campuses dominate employment here - the Catalant biotech company and the Cerner Innovation Campus recently constructed on the old Bannister Mall site. The light industrial campus that replaced the Kansas City Plant, which once made Pratte & Whitney engines, and later materials for the nuclear stockpile, is also in this footprint.

The concentration of medical technology companies in this area creates a unique employment hub. Because these firms offer specialized jobs not found elsewhere in the metropolitan region—particularly in biotechnology and medical research—workers commute here from across the entire metro area, regardless of where they live. This explains the unusually wide commute shed. Highly specialized workplaces draw employees from a much broader geographic area than more common types of workplaces, since workers with these specific skills need to go where the specialized jobs are. The same phenomenon is visible with clusters in downtown and Johnson County. 

Grandview

The M 150 Highway/I49 interchange has generated a decently size employment cluster on the border between Grandview and Kansas City. The NNSA complex is down here, along with a CPKC transload yard. Smaller firms include auto parts and computer products manufacturers. Unsurprisingly, this cluster pulls the most workers from Cass County. Surprisingly, it also attracts a large number from Clay County. 

The NNSA complex is down here, a CPKC transload yard, and a bunch of auto and consumer products manufacturers. Workers mostly come from se jackson county or down i435 from the northlands. Only 1 in 4 have a college degree - a number that would be lower without the NNSA site. 

I70 East Independence/Blue Springs

The eastern side of I70 is a major retail corridor, with one of the region’s last remaining indoor malls and a full complement of American big box stores. One large hospital (280 beds) and a smattering of aging office campuses (mostly consumer services and healthcare) fill out a fairly diverse job market. Workers are heavily drawn from Jackson County and their demographics reflect its diversity.

Lees Summit

From my tweet about this cluster: “hospitals, retail, and a single massive corporate campus home to USCIS and a couple obscure companies. Mostly draws workers from Jackson Co, relatively low educational attainment.”

It’s interesting that more workers don’t come from Jackson County given that Lees Summit has such a strong highway link (i470/i35 put us 20 minutes away from the JoCo center of population). Instead, we pull workers from the eastern half of Jackson County in large numbers. 

I35 North

Back over in Johnson County, three clusters follow I-35 South, the region’s most important freeway link. These workplaces—light industrial, office, retail—sit uphill from the highway (which follows Turkey Creek out of downtown KC) and buffer the neighborhoods around them from its noise and pollution. It’s a great suburban land use pattern and one that surprisingly few places have tried to replicate. 

The largest employer here is the Shawnee Mission Medical Center; the Oak Park Mall similarly clusters thousands of workers into a single structure. Workers are overwhelmingly drawn from Johnson County and their demographics reflect the county’s. 

I35 South

This is the second largest cluster in the metro (and given the wide geographic spread, arguably better thought of as a general Johnson County blob of 200,000 workers together with the clusters to the north, east, and south). Some 70,000 people work here in a range of occupations; one of the largest employers in the region, Garmin, has more than 5,000 workers here and in the Gardner cluster to the south.  

Most work done in this cluster happens in bland “business parks”, a new type of place invented towards the end of last century that combines low profile office buildings with light warehouses and buildings that are variously used for distribution, contractors’ offices and storage, consumer storage, medical buildings, and fad fitness concepts, all set back behind pleasant enough vegetation-lined streets. There are also a number of very large distribution centers, including our biggest UPS hub and a surprisingly large JC Penny warehouse. 

Gardner

Gardner and its environs includes the Garmin HQ building, the airport and rail-served New Century AirCenter, and a couple of large aggregate quarries. Its demographics mirror the cluster to its north. Workers commute from Gardner and southern JoCo, particularly the I35 corridor. They have typical demographics for the region and occupational mix.

Downtown KCK

Downtown KCK doesn’t meet the criteria I set out here for clusters but I’ve chosen to include it because of its historic significance as one of the region’s earliest commercial centers. Much of its workforce followed its population in moving out to suburban environs, but it remains home to a large number of white collar workers, predominantly in government roles. Governments at every level of the US frequently have office siting rules that strongly favor urban and disadvantaged communities; those requirements explain the large number of state, local, and federal workers in various bureaus here. 

Eastern Wyandotte County is majority Black; the workforce reflects this, with the largest share of Black workers of any cluster in the region. Hispanic workers are also overrepresented. Not surprisingly, considering the separative terrain, workers mostly commute in from Wyandotte County rather than the counties to its north or south. Unlike the Plaza cluster, few live in the block groups directly adjacent—most commute from the Wyandotte suburbs.

Kansas River Bottoms

The bottoms (wide flats carved out by the region’s two major rivers) are considered as clusters because of their relatively homogeneity of employment types, geography, and economic raison d'être. Along the Kansas, these include the West, Armourdale, Argentine, and Rosedale Bottoms, along with several others further upriver. The railroads turned these flat floodplains into industrial hubs starting 150 years ago; for the most part they remain as such. The bottoms further west have more modern distribution centers; the West Bottoms is slowly moving through the deindustrializing-wasteland-to-hip-urban-neighborhood transition. 

Missouri River Bottoms

The Missouri River Bottoms cluster is one of the largest in the region in both size and employment. It follows the Missouri through the heart of the region, including the Fairfax Industrial District in KCK, the East Bottoms, and North Kansas City. The furthest eastern block groups in this cluster reach inland beyond the bottoms; interestingly, much of their industrial space is below ground in caverns cut for limestone mining. A hill in unincorporated Jackson County just south of the river contains 5,000,000 square feet of warehouse space alone. 

Like the Kansas Bottoms, the westernmost edge of the Missouri Bottoms is the most recently developed and consists primarily of large distribution warehouses, while bottoms closer to downtown are among the oldest in the region. The GM plant in Fairfax is the largest single employer with more than 2,000 workers. It is currently retooling to produce the Chevy Bolt EV from 2026. 

Workers in this cluster come from across the metro area but particularly the Missouri side. Most live in the Northlands, followed by the city proper south of the river and Independence. Like other industrial-dominant clusters, it has a relatively large share of minority workers.

Liberty

The Liberty cluster is an odd shape: it non-contiguously follows I35 N from the I435 junction, jumps over to Liberty proper, and then follows hwy H for several miles through what is mostly farmland. This again is an effect of block group definitions; the overwhelming majority of the cluster’s jobs are along the highway or adjacent to the town. 

This is another cluster with a high share of industrial, transportation, or distribution employment; there are also sizeable shopping and health facilities. It does not include the Liberty Triangle, a (well known? I can’t really tell) thicket of strip centers bounded by several major roads. The area is growing fast—a new collection of warehouses artfully termed the Liberty Commerce Center has popped up on its border so recently that google earth still shows soybean fields. 

The Ford Claycomo Assembly Plant on the southern edge of this cluster makes up almost half its total employment. Its 8000+ workers collectively churn out more cars than any other plant in the country in a massive multi-building operation spread across 50 acres. The complex, which recently added a fully tooled stamping plant, is one of two F-150 production plants and among the largest and most significant factories in the state.

MCI

The final cluster, MCI, includes the airport and industrial parks to its east and south. Transportation and retail trade unsurprisingly are the two largest job categories; the airport alone employs more than 1/5th of the workforce here. Nearby industrial firms include an auto parts plant and a cpg manufacturer. A number of small technical schools and e-learing firms are headquartered across the highway from the airport, and the KCI expo center takes advantage of proximity to hotels, flights, and cheap land here. Just 1 in 4 workers in this cluster have a college degree.

1.
My previous two posts discussed the present and future distribution of Kansas City’s population. Population distribution is an important component of human urban geography, but is in many ways downstream of employment distribution.

One of the fundamental insights of urban economics is that cities are principally labor markets, pools of people brought together because their skills are worth more as part of a dense network than they are on their own. Where people work (at the aggregate level) is thus more determinative of where they live than the other way around. If it was the inverse, we would expect people to live in relatively low density communities with little traffic and easy access to natural amenities like beaches or mountains. Instead, we increasingly cluster in cities that are located primarily according to the whims of economic geography.1 You wouldn’t choose to drop a city at the outlet of the Kansas River if you were seeking maximum human pleasure; we are mostly here for the opportunities afforded by our economic geography (or at least the legacy of that economic geography).

Labor markets continue to shape the distribution of population below the macro (super-regional, national, etc) scale. People’s decisions about where they live within cities is further shaped by their commute time to jobs they currently have or are likely to take.2 Another classic urban econ finding is Marchetti's Constant, a surprisingly durable rule of thumb that suggests people are willing to commute up to an hour per day to access jobs. If you’re an accountant, and most large accounting firms are downtown, you are likely to buy a home within a half hour drive of downtown. Within those constraints you might make different decisions (who you want to live around, what amenities matter to you, where your family/social network lives, etc), but the half hour commute rule holds for most people most of the time. Indeed, the median commute time in greater Kansas City is 23.3 minutes (a figure that also suggests to me we will continue to sprawl as we grow no matter our infill program).

In part because labor markets have returns to scale (agglomeration economies) and in part because workers want to be able to access as many jobs as possible and firms want access to as many workers as possible, jobs tend to cluster to a greater extent than do residents. This is why central business districts with big office towers exist even in regions where the overwhelming majority of people live in single family houses in the suburbs: CBDs maximize how many people can get to workplaces within that half-hour commute time. That dense employment clusters are efficient is another classic urban econ finding. In recent decades another benefit has gained attention: climate considerations. Driving commutes account for a large share of our ghg emissions; centralizing jobs minimizes driving and makes public transportation more viable.

For all these reasons—shaping the pattern of the region, agglomeration economies, climate considerations—it is important to understand where people work in a region. This post gives a basic outline.

2. Kansas City Jobs
The Census Bureau produces a few different datasets with workplace data. The ACS reports workplace location and employee characteristics through PUMS, but it is not available at small geographies and—in part because it is survey data—it has been criticized for having high margins of error. ACS employment data is considered to be more appropriate for assessing demographic, social, economic, and housing characteristics than for geographic analyses. 

More useful here is the Origin-Destination Employment Statistics (LODES) dataset produced by the Bureau’s Longitudinal Employer-Household Dynamics program (LEHD). LODES includes origin and destination data for individual workers, as well as data about their residence area and work area characteristics. Rather than surveys, it is collected through analysis of unemployment insurance records, state and federal tax records, and other administrative sources. Conveniently, it is available in geographies as small as blocks.

For this analysis I pulled LODES data for the five core counties of the Kansas City region at the block group level and joined it to TIGER spatial data.3 All areal calculations come from TIGER bg area estimates, which I understand (could be wrong about this) are not always perfectly consistent with other sources. Water area is left in for simplicity.4 Rivers and freeways were added as visual guides from TIGER shapefiles. Density categories are intended to capture significant clusters of employment.

There are 975,000 in-person jobs in this dataset.5 Their general structure will look familiar to anyone who has either spent time in KC or is broadly knowledgeable about the geography of midsized US metros. There is tight clustering in a small CBD, a well-defined ‘midtown’, a broad wedge out of downtown in one direction that reaches its widest point along a circumferential beltway (here, to the southwest), and then a smattering of variously-dense suburban hubs adjacent to freeways around the metro area.

The densest nodes outside greater downtown are suburban downtowns or large hospital districts. Downtown Olathe, downtown KCK, and the Plaza Area (arguably a suburban downtown) all clear 20k jobs psm, and (surprisingly) North Kansas City clears 10k psm. Hospital Hill, University of Kansas KCK, and the Shawnee Mission hospital districts stand out with 20k; hospital districts that clear 10k jobs psm include St. Joseph/St. Mary's at i435 in South Kansas City, Research Medical Center along US71 at 63rd St, and North Kansas City hospital. Others, including St. Lukes in KC, are too integrated into larger job centers to stand out here. I was somewhat surprised to see that the job density of hospital districts is one of the metro’s defining features, but I shouldn’t have been—hospitals are now small cities unto themselves.

Large industrial block groups have relatively high job densities. The river bottoms of the Kansas and Missouri Rivers, the area around Pleasant Valley and Claycomo north of the river, and Johnson County industrial parks along I35 all feature here. Of these, only the JoCo districts achieve densities of at least 5k jobs psm.

Finally, you can see retail-dominant suburban activity centers like those in northwest Lee’s Summit, Zona Rosa, Liberty Triangle, the Kansas Speedway, and along the eastern side of I70. Johnson County’s mix with large suburban office parks like Corporate Woods, Aspiria, and a variety of different clusters on Metcalf Ave. Some of these, like Oak Park Mall in Overland Park and Town Center Plaza in Lenexa, are among the largest retail centers in the region.

One thing you’ll notice is that workplaces are much more sensitive to quick highway access than are homes. Despite the harm the freeway loop does to quality of life and urban vitality around Downtown KC, it has helped the CBD and the crossroads retain the greatest density of jobs of any place in the region.6 I35 and I435 anchor most of the non-downtown job centers, eastern JaCo jobs are strung along I70, US50, and I470, and I29 and I35 in the northlands connect significant clusters. Us urbanists love to dream of living in places centered around walkable neighborhoods and sustainable transportation, but it is impossible to overstate how important the big roads are to the spatial order of contemporary Kansas City. Less than 3% of Kansas City area commuters get to work by bike, walk, or public transportation.

An eagle-eyed map reader here may notice that MCI does not appear, despite being a major job center. More than 3,000 people work at the airport or on its grounds, but because the Census codes the entire facility in a single block group, and block group size is determined by resident count, its area swamps job density. The nature of block groups obscures job density in other areas too; for example, if the area around Ford’s Claycomo facility were densely populated, its 9,000 employees would push its color scheme from blue to yellow. Large standalone industrial centers are underrepresented across the map.

3. Jobs + Residences

Kansas City is hardly a walkable region, but all else equal people really do prefer to live near jobs and amenities. This becomes evident when you map jobs and workplaces together. 

The one thing that really stands out in this visualization is the favored quarter pattern: the wedge following I35 southwest from downtown KC dominates the distribution, spread and reinforced by I435. Johnson County, with its high-performing schools, large homes, and quality public services dominates the market for both living and working in the region. 

The density of the gridded core city becomes evident here, with the northeast neighborhoods distinctively outlined and a solid wall of neighborhoods marching right up to the Blue Valley gap. Also clearly visible is the Independence-to-Grandview ridge between the Blue and Little Blue Valleys, one of the most important elements of early Kansas City geography. The Lee’s Summit and I70 patterns are fairly similar to that of their job distribution alone, though Independence and Blue Springs neighborhoods make themselves much more clear. 

North of the river, density stretches out towards Liberty and is otherwise mostly evenly distributed northwards. The rectangle formed by I435, I29, M152,& US 169 remains mostly farmland but is starting to fill in as Meta’s Golden Plains Technology Park and other industrial facilities go up just to the north. Over in Wyandotte County, development curiously follows a clear path between KS 5 & I70; anyone who can explain it (it isn’t geography or municipal lines!) is welcome to message me. 

Finally, though the ridge between the Blue and Little Blue is better known, an imitation of this pattern has developed between Cedar and Mill Creeks in Johnson County. There is no real significance to this region (anchored north-south by KS 7), I just find it interesting.

4. Fin 

Future posts about this data will explore more about what it means for regional transit, revitalization efforts in the city proper, and the region’s overall competitiveness. As always, email thoughts to ddavoli.234@gmail.com or message me @2024dion on twitter.

Few things determine the character of a region so much as its population: where people live, how many of them there are, and how their communities are structured. Regional population size and distribution shape everything from the size and accessibility of labor markets to eligibility for public funds to the competitiveness of your kid’s little league games. Thus, projecting the future size and distribution of the region’s population is of enormous importance for planners. Fortunately it is one we are decently good at, at least with large aggregation units.

In the Kansas City region (I’ll favor the terms region and area on this blog over those with a more specific meaning, like metro area), the agency tasked with estimating our population trajectory is our MPO, MARC. This makes sense–land use planning is treated as an arbitrary practice in US cities but transportation planning guides the investment of billions of dollars of infrastructure. You don’t want to put roads or subway systems in places they won’t get used.

The most recent MPO projections available online were released around 2020. They have a handy viewer here. You can also download their data, which gives 10-year population trajectories at the tract level out to 2050.

This post walks through my analysis of the projections and major themes I see. Note I’m not a demographer or much of a stats guy. This is all amateur work. If you are and you think I’m mistaken about something, get in touch.

This is the lead finding, of tract level population change from 2020-2050. Tracts (unfamiliar? Tracts are statistical units representing approximately 4,000 people. Think of a large urban neighborhood or several subdivisions–the River Market area makes up ~1 tract) in blue are projected to grow; red shrink.

The basic pattern is clear: the edges of the region are growing quickly while impoverished neighborhoods in the core slowly decline. Greater downtown is the exception to core population loss, with growth driven by continued inmigration of young professionals.

In all, the region will add 480,000 people by 2050, with growth gradually slowing each decade from the 9% rate of the past two decades to 6% in 2040-2050. The Kansas City Region of 2050 will have 2.55 million residents, approximately matching the population of metropolitan Portland today. That’s a 23% increase, significantly higher than the 9% that the Census Bureau projects for the country overall.

The eight counties that here make up the region (see then notes for an explainer) will all grow, albeit unevenly.

Interstate rivals Johnson and Jackson Counties will each climb to nearly 800,000 residents, with Johnson surpassing Jackson for the first time by the end of the 1940s. Steady population growth will continue in Independence, Lee’s Summit, greater downtown Kansas City, and Blue Springs, but out-migration from east side Kansas City neighborhoods will offset some of those gains. In Johnson County, growth will continue to overwhelmingly concentrate at the urban fringe, with wealthy Olathe, Overland Park, and Lenexa seeing thousands of residents move to subdivisions to be built out on prime farmland. The City of Overland Park will also add residents as it replaces aging power centers with low-rise apartments, particularly along Metcalf Ave.

Elsewhere, the northlands will see the continued outward march of subdivisions up to and past MCI. Wyandotte County will see enough suburban development to counteract decline in KCK, and Leavenworth County will see modest growth, particularly in its seat.

Growth will cool slowly over time in seven of the eight counties. Only Miami County (south of Johnson) will grow at an increasing rate each decade, though from a very low base. Platte, Clay, and Johnson will grow the fastest, particularly in the 2020s. On the other end of the scale, Jackson and Wyandotte will approach zero growth by 2050.

The most climate adaptive development pattern is to grow through infill construction in near-core neighborhoods, where people are likely to have shorter work commutes and can take advantage of a denser network of services and streets to shorten or substitute driving trips. Unfortunately, much of this development will happen on Kansas greenfields.

The next two charts plot each tract’s growth and distance from downtown. They’re faceted into decline/growth to make the trends more visible. The majority of the high growth tracts (though note these aren’t adjusted for area) are at least 10 miles from the core, and the highest growth are all at least 17 miles out. On the other end of the spectrum, most declining tracts cluster within five miles of the core.

You can see that a series of tracts in Jackson and Wyandotte Counties will lose people; in some, the number is substantial. 400 people is 10% of a tract’s current total—that’s a ‘homes left abandoned’ level of decline. Others will see small losses, presumably from a lack of new construction and the gradual decline in household sizes as homeowners age and their kids move away.

Another useful way to look at this data is to facet by income. The four graphs tell a simple story: people are moving to wealthy neighborhoods and moving away from poorer neighborhoods. There are a few wealthy Johnson and Platte County neighborhoods slowly losing population; these are again mostly the result of aging.

Finally, interstate competition. Is Kansas City more Kansas or more Missouri? The gap might be closing slightly, but the number still say Missouri.

Have ideas for how else I should visualize this dataset? Message me on twitter or leave a comment below.

A more useful way to understand population density is to look at the number of people in a region who live at a given density. You can do this by plotting population as a function of binned block group density, giving an easy-to-follow curve (note that this approach is location-agnostic; another common way to visualize density is to look at population-weighed densities by distance from the center of the region).

Here’s such a plot for the Kansas City area. The bins are logged; the labels are selected for clarity:

The distribution here is not particularly surprising: Kansas City is an overwhelmingly suburban region. A large majority of Kansas Citians live in neighborhoods with a density of ~2.5k – 7k ppl/mi^2, a common suburban pattern. For reference, Palo Alto has a density of 2.6k ppl/mi^2; Staten Island 8.1k. The suburban dominance holds for all of these counties except Platte, which is mostly still rural.

So suburban is KC, in fact, that it’s difficult to make out any of the dense neighborhoods near the center. Downtown KC and its environs collectively are all represented in the tiny extension and bump on the gold line beyond Johnson County’s curve. Even the two densest bgs barely clear 30k ppl/mi^2; by comparison, the entire upper east side of Manhattan has a density of 168k ppl/mi^2. KC lacks even a single small area that approaches big city urbanism.

Another way to look at this data is through simple maps of where these densities cluster.

I chose cutoffs of 2k/mi^2, 5k/mi^2, and 10k/mi^2, densities that serve as a decent proxy for transit service viability: local bus, 15min high capacity bus, light rail.

The 2,000 people per square mile cutoff captures virtually everywhere in the metro area that’s built in a recognizably suburban pattern of subdivisions connecting to collector roads. Density drops below these thresholds along the river valleys and in the industrial bottoms near downtown. The Blue River valley forms the clearest divide, cleaving most of the eastern suburbs from Kansas City proper. At the urban fringe, you can see leapfrog sprawl in Gardner, Grain Valley, Greenwood, and north of the speedway. You can also see pieces of suburban towns that have yet to be swallowed by sprawl, including Bonner Springs, Excelsior Springs, Liberty (kind of), and Platte City. There are a few notable breaks in Johnson County: industry follows i35 out of the city, and a combination of high-end office parks and mansions keeps densities low along i435.

Block Groups with a density of 5,000 people per square mile are much spottier. In Kansas City proper, downtown, midtown, the plaza area, and the historic northeast all easily clear this threshold. North of the river, a few NKC bgs do so and then a smattering of subdivisions stretching out to the the beltway. Despite its population challenges, many central KCK bgs still reach 5k. Older parts of Raytown, lee’s Summit, Blue Springs, and Independence are relatively dense; as are many newer subdivisions in southwest Johnson County.

At 10k ppl/mi^2 things break apart completely. Just a few downtown bgs in the Kansas Cities clear this density. The case for the east-west bi-state sustainable corridor is evident here even without considering downtown job densities; it would follow one of the most urban routes in the region. The streetcar route, unfortunately, is less visible. There should be a whole-of-region effort to direct new construction onto these two spines. Beyond city limits, the scattered bgs reaching 10k are mostly single or grouped apartment complexes; one odd bg in Overland Park actually clears 30k/mi^2.

Finally, here’s what this looks like if you compare to sets of peer cities. You’ll notice the KC distribution is slightly different–for ease of comparison, I used UAs here (LS is properly part of the KC UA, in my extremely biased opinion). The first compares Missouri’s pole cities; the second, the four closest UAs in size to the combined KC/LS UA.

Subscribe now