Sunday, November 14th, 2010

Are People Really Fleeing Shrinking Cities?

Update: The self-service tool that has easy query access to the IRS migration data, including at the metropolitan area level, is available at You can also read the launch announcement with more info on this tool and what it can do for you.

When people think of Rust Belt or other regions that are either shrinking or growing extremely slowly, what often comes to mind is the thought of rats fleeing a sinking ship. The concern over “brain drain” reinforces this exact type of narrative. And when you look at the net migration figures from the Census Bureau, it’s easy find the image confirmed. For example, here’s net domestic migration for Midwest metro areas from 2007 to 2008:

I’m sure that high out-migration was a big thing fueling the exodus to the suburbs, but are people really leaving the region? Are people really fleeing Detroit and Cleveland on a metro area basis?

Part of the challenge in answering this question is that the Census figures are net: in-migration minus out-migration. So you don’t know if the net figure is negative because lots of people left or because very few people came or both.

Fortunately, there is data out there that lets us disaggregate the number and look at in and out migration individually. It is the tax return data from the Internal Revenue Service. (This is actually one of the major sources the Census Bureau uses in estimating migration). Not many people really work with this data though because it is so painful. A friend who’s a well known demographer recently told me he “found it to be about the worst data I had ever seen, in terms of working with it.”

I believe I have cracked the code on this data, rendering it easily useful for real analysis by ordinary human beings for the first time. In addition to making the basic data easy to use, I also generated a ton of other data from it, including the MSA-MSA migration that I talked about in a previous post. I’m planning to launch a tool for this soon, so if you’re interested, you should definitely mail me so I can tell you when it’s ready.

In the meantime, I wanted to share some interesting things I’m finding in it, notably in the migration rates. I calculated these myself and the numbers are still experimental at this point, but I wanted to share some early results.

Looking at the most recent year for which data is available – 2008 (2009 data comes out in the spring) – here are the ten lowest metro areas for out-migration rate* between 2007 and 2008, along with the all metro are:

Not what you expected, is it? That’s right, Pittsburgh is dead last among all 366 US metro areas I’m tracking in terms of its out-migration rate. People aren’t leaving, just like they aren’t leaving a lot of other places famous for large absolute net domestic out-migration. Not even Cleveland (#13 from the bottom) or Detroit (#17). (In fairness, net migration did turn positive for Pittsburgh this year).

I’m still validating some of the numbers. Large metros seem to have lower rates than small metros, probably an artifact of many small metros being single county. But still, this was surprising even to me.

What’s really killing these places is that they have even lower in-migration. Here’s the bottom 10 MSAs in the US for in-migration rate.

Here’s all 12 of my Midwest MSAs on in-migration rate, as a percent of the US all metro average:

What I tend to see is the Rust Belt and other struggling regions have both very low in-migration and out-migration, and the net migration number is bad because in-migration is very low indeed.

I’m reminded of what Jim Russell once said about one city being a “cul-de-sac of globalization.” That seems to be the effect at work. These cities are not getting the human capital churn they need to build the talent networks necessary to connect them to the global economy. (Places like LA and Chicago are different in my view, since they are really two cities in one – a thriving global city core and a larger lumpen-city that more fits the Rust Belt model).

It made me wonder what the relationship really is between something like out-migration and population growth. Here’s a quick map of 2007-2008 population growth by county in Indiana, growing counties in blue, shrinking counties in red, color shading proportional to the percentage change.

Now here’s county out-migration rate from 2007 to 2008, grayscale, but again with intensity shading:

It’s a bit difficult to tell, but it does look like the belt of stagnant to declining population northeast and southwest Indiana roughly match the low out migration areas on that map.

I was curious so just before this post I dumped out-migration rate and percentage population change for all counties in the US and plotted them quickly:

Hmmm. This one I did in about two seconds right before this blog post when up, so while everything in here is caveat emptor right now, that one particularly so.

Anyhow, I think this analysis shows that we can’t fall into simplistic notions around people fleeing regions. It would appear that the opposite may in fact be true. People are stuck in them – either because they don’t want to leave or because they can’t. I’ve got to believe at some level that recovery means boosting the human capital circulation factor – in and out – significantly. I think these migration rates are a key indicator to watch in seeing if cities are turning the corner. I’ve got the data going back to 1996, and it looks to me like the rates are very stable over time for most of these places. That’s probably not a good sign.

* I calculate the migration rate as a rate per thousand people. The formula I used is (migrants) / ((non-migrants+out-migrants)/1000). If you have feedback on this methodology, I’d love to hear it.

Topics: Demographic Analysis

26 Responses to “Are People Really Fleeing Shrinking Cities?”

  1. Danny says:

    I wonder what the data would look like if we hadn’t pushed homeownership with strong policy measures for the last 40 years. Homeownership is all fine and dandy when you have economic security…it is nothing more than a ball and chain when you are unemployed.

  2. Jennifer Hill says:

    This analysis seems right to me. I grew up in IL and IN, lived on both coasts from college onwards and moved to Michigan in 2005, 20 years after I left the Midwest. Twice in the last 24 hours, I have interacted with people who said, “Wow, you’ve lived all over”. I don’t remember hearing that sentiment much on the coasts, instead we just compared notes on locations. If I did hear it, it was usually from someone I would guess did not attend college – I think education and the expectation that one “should” go away to school plays into this some? And homeownership, as previous Danny notes above. Lot cheaper to buy a house in the Midwest then the cities I lived in on the coasts.

  3. Pete from Baltimore says:

    I would make two points
    1 States like Pennsylvannia and Michigan have lost population.thats pretty much a regions have lost population
    2 I dont think that anyone is denying that most cities have lost a great percentage of thier population to thier suburbs. I really havent heard many people say otherwise.So im not sure what point this article is trying to make?

    Places like Detroit,Baltimore and Cleveland have lost massive amounts of population.and are filled with many abandoned houses. The fact that thier “Metro” area has prospered isnt much of a comfort in my opinion. Especially in places where there is a clear border. In Baltimore at least, we dont live in a county.Baltimore County is a completly different entity then Baltmore City. So we get no tax dollars from any resident or business in Baltimore County, or any other suburb. So thier success is not too helpful too those of us who live in the City

  4. DBR96A says:

    Pennsylvania has never lost population, not even in the 1980’s when the economy in Pennsylvania was at its worst.

    Furthermore, Pennsylvania is the only one of its closest “peer” states (IL, MI, NY, OH) to post positive net domestic migration in recent years. The other four are strongly negative.

  5. BrianTH says:

    Many older, larger central cities did lose some population after WWII, but many of those started growing again at some point. Meanwhile, many younger, now-larger central cities never lost significant population. Of course central cities are a tough unit to work with, since their borders aren’t determined in any standard way and can change over time.

    Anyway, the point of this particular study is not to look at how central cities have been interacting with their suburbs, but to look at how metro areas have been interacting with the rest of the world.

  6. Pete from Baltimore is 100% spot-on. If you only looked at the statistics for the Baltimore metro area, you’d think we have a pretty steady population. However, when you remove the surrounding areas and only look at Baltimore City, the story changes dramatically. Baltimore City has been losing population for decades, and we have blocks upon blocks of abandoned homes. Not just an abandoned home here or there…entire neighborhoods have been abandoned. And it’s only getting worse as time goes on.

    Shrinking Baltimore would save taxpayers millions on unnecessary infrastructure costs, police and fire services, and other city budget line items that are directly impacted by loss of population.

  7. COAST says:

    Bingo on the IRS data. What we found in Cincinnati is that population analysis isn’t very enlightening. Viewed next to incomes, and a whole new picture emerges.

    Cincinnati’s population has been pretty flat for the last decade or two, give or take a few percentage points. All the while our poverty rate has grown and surrounding suburbs have exploded with higher income former Cincinnatians.

    IRS data confirmed that lots of wealthier people have moved to the ‘burbs, while lots of low income people have moved in to replace them. Simultaneously, taxes and fees to live in Cincinnati and Hamilton County have grown at 2-3 times the rate of inflation. City & county government has been jacking up taxes to pay for more and more social services.

    It turns out that people who pay a whole lot in taxes (rich folks) don’t tend to use much in the way of social services. They are the ones fleeing to newer, safer, cheaper lives in the likes of West Chester, Milford, Mason, etc. Meanwhile poorer folks, who use lots of social services, but don’t pay much in taxes, are attracted to the urban core because it’s a better deal for them.

    The macro picture that emerges looks pretty bleak. Revenues continue to slide in Cincinnati and Hamilton County. In response, politicians cut core services like police and fire. This makes them even less attractive to folks with money. It’s a death spiral.

  8. Chris Barnett says:

    Coast, I think you have one essential element wrong. Low income people haven’t “moved in to replace” suburban migrants in Cincinnati (or any other central city). They’ve just grown up poor in the city and haven’t moved up the ladder and out to the ‘burbs. They’ve been left behind, as Pete from Baltimore points out.

  9. Shaun Smakal says:

    Well, I’m not really surprised, I would say, by what you’re presenting. However, this blog is horribly named. MSAs are not cities. They don’t represent ‘traditional’ forms of urban development, economies, or population/density/diversity. I understand you have to work with the information we’ve got, but we have to realize that MSA don’t functionally or figuratively represent the cities they are associated with and continuing to lump it all together and treat it like a “city” is a very bad idea, in my opinion.

  10. poor policy for more than 20 years however I’m glad to see Buffalo is not on the list. I’m also glad these stats are in MSA than just “City Proper” just due to places like Columbus Ohio alone, you never know are you in the city till you get to I-270

  11. George Mattei says:

    Fascinating. Two thoughts emerge:

    1. Columbus has spent a lot of time talking the past few years about stopping the “brain drain” of students either growing up here or coming here, attending OSU, and then leaving afterwards. Maybe that’s not such a bad thing. You want to keep SOME of those people, but if your thoughts about networks are true, then spreading some “apostles” to the winds might be a positive.
    2. Clearly if metro areas are not shedding people, but their core cities are decaying, we’re building way too much product and investing in new areas at the expense of older ones. I see that in Cleveland-their metro area population has been fairly constant over the past 50 years, but the land are has increased dramatically. It’s like musical chairs on a metro level.

  12. Anon says:

    The expansion to the exurbs is creating a new problems with the MSA definitions. For some reason, Cleveland and Akron are split. I work in downtown Cleveland with several people who commute from just over the line in Akron’s MSA. Their commute is only slightly longer than mine, and I live in the central city. Same thing east and west. People who live in “Non-metro area” are in the city everyday because its easy with our light traffic.

  13. George Mattei says:

    Anon, I agree. That’s been an issue in the Northeast for years. I grew up in New Haven, CT. I knew people that commuted to New Haven, Hartford, Bridgeport, Stamford, New York City and various suburban communities. Well, if you live inn a suburb near New Haven, but a large percentage of your workforce in that community commutes to Hartford, whose suburb are you?

    In places like Columbus or Indianapolis, which don’t bump up against any other major cities, it’s much easier to identify suburbs with their cities.

    This caused an interesting effect over time in the Northeast. When people from Connecticut traveled they would rarely say they are from Hartford, they would say they were from Glastonbury or Manchester. Anyone know where those towns are? Nope. However, people from suburban Columbus still say they are from Columbus, even if they’re address is Gahanna or Westerville.

    I get the sense that having one core city helps people to focus on the core of their metro area as the driver of the economy. When you have divided loyalties, it seemed that there was less concern about what happened to the core cities, because there was less identification with them.

  14. pete-rock says:

    It seems some people might be missing the bigger point that I think Aaron is trying to get across, at least the way I interpret it.

    I don’t think Aaron or anyone is denying that central city population is declining and has been declining for decades. But that decline is not simply because of the continued move of middle class residents from city to suburb. I think what Aaron is trying to get across is that American metro areas have long been places where a continual stream of low-income immigrants moved in, got jobs and homes, adopted middle class mores and values and moved out. That process kept churning like an escalator for decades, taking people to the next level.

    In some metro areas, that process continued with both domestic and international immigrants. Chicago, for example, where I live, will always be the hot spot for recent Big Ten school grads, even if Hispanic immigration remains as flat as I believe it is right now. However, in places like Detroit, Cleveland and Baltimore, there’s no new group hopping on that escalator anymore — just the same people, most often poor, who are unable or unwilling to jump on.

    The message I get from Aaron’s data is that in-migration is the real issue for metro areas in general and central cities in particular. Cities require the constant influx of new groups to remain viable.

    BTW, if my analogy seems simplistic, or if I’m misunderstanding Aaron’s points, please forgive me. It works for me.

  15. Wad says:

    Anon, something that’s compensating for the relatively inflexible MSA orbits is a category called the Combined Metropolitan Statistical Area (CMSA). This adds together MSAs with substantial economic ties, as well as micropolitan statistical areas.

    I’ve been studying these tables lately in the past few days.

    The New York City CMSA encompasses the giant New York City MSA (the 5 boroughs, Long Island, neighboring counties, and the northern counties of New Jersey), plus the Poughkeepsie and the Bridgeport-Stamford-Norwalk, Conn. MSAs.

    One reason for CMSAs is because employment is polycentric or acentric, for that matter. In New York, many major S&P 500 companies with thousands of employees are not in Manhattan, but giant campuses in the hinterlands. General Electric’s headquarters is in New York’s CMSA, Bridgeport’s MSA, but it’s a campus not near any city center, nor did GE create a center around it.

    Repeat this at hundreds or thousands of places within an MSA, and you’re bound to get a picture of the CMSA’s galaxy.

  16. rs says:

    Nice analysis… but I think you’re giving yourself a little too much credit for being able to manipulate the IRS data. Sparse matrix algorithms have made these types of flow matrices easy to deal with and have been around for years.

  17. Matthew Hall says:

    rs, I hope you realize almost no one on hear has any idea what you are talking about.

  18. rs says:

    The problem with large flow matrices like the data used for this analysis is that they are huge and so cumbersome to analyze due to the large amount of memory needed to handle them. However, the vast majority of the data elements are zero. In particular, I am willing to bet that over 90% of the non-zero data elemants are found very near the main diagonal of the matrix.

    Sparse matrix algorithms use that fact to get around the memory problem and so make these matricies relatively easy to manipulate.

    Matt, I am also willing to bet that Aaron knows exactly what I am talking about.

  19. Alice says:

    The problem is that you are taking a snapshot when you are telling a dynamic story.

    When you look at one of the rust-belt cities, you see only those who did not leave in the years and even decades before the year of the data. So a city that had suffered massive outflow in the past might be very stable now. Without looking at the whole time series of data, it is not possible to make statements about how much migration is or is not happening.

  20. Alice, I think that’s part of it. Maybe people did flee Pittsburgh in the past, and now they aren’t, but perception hasn’t caught up with reality. Nevertheless, I do have a time series going back to 1996, and the data hasn’t change much since then.

  21. Regarding the difficulty of working with IRS data, indeed, if you have SAS or similar tools and the ability to work with them, it isn’t difficult. 99.9% of America has no idea what a sparse matrix is, however, much less how to use one. If you work in an academic research institute or some such, doing analysis like this is no problem. But if you don’t…

  22. Matthew Hall says:

    rs, Aaron likely does understand your message, but this forum has a very diverse readership. Many readers and posters on Urbanophile have little or no formal statistical training. Yet, their insights and observations into regional demographic issues have been a valuable contribution to this forum in my opinion.

  23. Evan says:

    This post and the comments that followed are very interesting, but I see some discussion points missing. I will cite Portland, Oregon as an example but I suspect Portland is not alone.
    As Portland has “matured” in the last two or three decades, it has become a much more attractive place to live. However, as in many other places, the school system has not improved along with it. As a result, you see a lot of folks with kids who might prefer to live near downtown move out for the sole reason of a “better” school elsewhere. Those without kids are not concerned with the state of the schools, at least directly, and so are free to live in or near downtown and also near work, the social scene and services they desire. For some, they are able to live in places where they can walk, ride a bike or take transit to their favorite destinations – a major lifestyle benefit.
    But back to my point – these people statistically have fewer children, and may also have them later in life as well. So the natural replacement is less than in the suburbs, and the net population change is negative.

  24. Jerry Paytas says:

    The CMU Center for Economic Development used to do a lot of this research in the early 2000s:

    Destination Pittsburgh (PDF)

    Published June 2003
    Concerns about population loss have brought retaining and attracting talent to the forefront as a key policy issue. In 2000, the CED issued a series of reports examining migration trends in the region. These reports found that contrary to popular belief, Pittsburgh was not losing people so much that it was not attracting people. In-migration and out-migration were both extremely low. The difficulty of further reducing out-migration required a focus on in-migration. This report focuses on the in-migration half of the equation

    The Root of Pittsburgh’s Population Drain (PDF)
    Published November 2003
    Recent population estimates and migration data show that the Pittsburgh region is losing population. News coverage of the release of the annual estimates have explored why the region is losing people. The most-common speculation is that young people are moving out of the region. This report attempts to quantify the impact of net migration losses in terms of those that moved away, and the children and grandchildren born elsewhere to former residents of the region.

    Who is Leaving Pittsburgh? (PDF)

    Published April 2004
    Who is Leaving Pittsburgh? addresses the question of youth migration and demonstrates that the 20-29 age group is not the only category losing population.

    Origins and Destinations of Pittsburgh Migrants (PDF)
    Published April 2004
    Origins and Destinations of Pittsburgh Migranst provides a data chart breaking down the state of birth for those entering and exiting the region.

  25. wanderer says:

    Evan raises an important point, though I wouldn’t lay it all at the feet of the (real or perceived) quality of schools. There’s been a long run sorting out process of families with children from (North) American central cities. This started in the 1950’s, when cities weren’t perceived to have a school crisis. Families with children moved to the suburbs to escape a range of urban ills (real or perceived), to get larger, newer units, in some cases to get away from racial minorities (though that wouldn’t have been the big push factor in Portland). For decades, no decent housing was constructed in central cities, especially not for families. The only North American city I’m aware of that’s attracted significant numbers of families with children to its downtown is Vancouver, B.C. A good many of its immigrants are from Hong Kong, so they’re obviously not put off by high-rise living, which many Americans think is bad for children.

    Meanwhile the central cities refilled (when they did refill) with childless couples, (straight) singles, gays, immigrants, and African-Americans, not that I am by any means equating these groups. In general (immigrants are an exception) these groups had fewer people per unit. So population can fall even when all units are occupied.

    The best number to look at in gauging the trajectory of a city or place is the number of households, not the number of people. When you look at this, a lot of places that have lost population, like Philadelphia, haven’t actually lost very many households. Places like Detroit and Saint Louis have lost households over the decades.

    It’s an interesting analysis, but is it really news? Metropolitan areas have generally maintained, or almost maintained their populations, even with depopulating central cities like Detroit.

  26. Chris Barnett says:

    The problems with looking at households as a key indicator are twofold: depopulation and obsolescence.

    When mom & dad retired after the kids grew up and moved to the ‘burbs, the city was left with two aging (or aged) folks unlikely to contribute to income growth or growth in commercial or retail activity (depopulation), where once there were two working parents and several teens. Nor are mom & dad as likely to be able to maintain an aging and/or obsolete housing unit over the long haul (obsolescence).

    To the extent that older residents are replaced by a pair of younger roommates renting their home, or by a younger couple buying it, then these problems can be overcome: properties are maintained and upgraded, and their residents spend more of their (likely higher) incomes.

    So the numbers in each age cohort of the population are an important part of the analysis, perhaps as important as the population or household trends. Jim Russell has pointed this out repeatedly regarding Pittsburgh: young and educated trumps old and retired.

    Paradoxically, programs to “help” seniors in inner city neighborhoods “age in place” probably do more harm than good for the neighborhoods overall because they can’t fix everything and thus promote deferred maintenance and delay the inevitable (and necessary) turnover of housing units to new residents.

The Urban State of Mind: Meditations on the City is the first Urbanophile e-book, featuring provocative essays on the key issues facing our cities, including innovation, talent attraction and brain drain, global soft power, sustainability, economic development, and localism. Included are 28 carefully curated essays out of nearly 1,200 posts in the first seven years of the Urbanophile, plus 9 original pieces. It's great for anyone who cares about our cities.

About the Urbanophile


Aaron M. Renn is an opinion-leading urban analyst, consultant, speaker, and writer on a mission to help America’s cities thrive and find sustainable success in the 21st century.

Full Bio


Please email before connecting with me on LinkedIn if we don't already know each other.



Copyright © 2006-2014 Urbanophile, LLC, All Rights Reserved - Click here for copyright information and disclosures