Where is income mobility high and low?

Climbing the income ladder occurs less often in the Southeast and industrial Midwest, the data shows, with the odds notably low in Atlanta, Charlotte, Memphis, Raleigh, Indianapolis, Cincinnati and Columbus. By contrast, some of the highest rates occur in the Northeast, Great Plains and West, including in New York, Boston, Salt Lake City, Pittsburgh, Seattle and large swaths of California and Minnesota.

Check out the map at the NYT link.  Based on eyeballing, western North Dakota seems to do best and northwestern Mississippi seems to do worst.

This is based on work by Raj Chetty, Patrick Kline, and Emmanuel Saez, and the other results are quite interesting:

The researchers concluded that larger tax credits for the poor and higher taxes on the affluent seemed to improve income mobility only slightly. The economists also found only modest or no correlation between mobility and the number of local colleges and their tuition rates or between mobility and the amount of extreme wealth in a region.

But the researchers identified four broad factors that appeared to affect income mobility, including the size and dispersion of the local middle class. All else being equal, upward mobility tended to be higher in metropolitan areas where poor families were more dispersed among mixed-income neighborhoods.

Income mobility was also higher in areas with more two-parent households, better elementary schools and high schools, and more civic engagement, including membership in religious and community groups.

Regions with larger black populations had lower upward-mobility rates. But the researchers’ analysis suggested that this was not primarily because of their race. Both white and black residents of Atlanta have low upward mobility, for instance.

Of course that is all correlation and not causation per se.  The Google link to the original research ought to be here, but right now the available links are down, perhaps soon they will come back up again.


When you see "western North Dakota", your mind should think "Bakken shale". That's why the economy is booming there.

Yup, "western North Dakota" = overnight millionaires (lucky landowners)

That immediately makes the whole 'income mobility' study highly suspect.

Their definition and measures of "income-mobility" look very subjective.

Personally, I would suspect it would be the middle deciles most affected, but that's based merely on half a decade in the industry. There are rather more people raking in ~$100k gross than there are landowners becoming millionaires.

I agree a lot of other blue areas are in areas with a lot of recent oil and gas development. E+P companies' spending lifts all boats in these areas whether it is landowners, specialized oil field work, or the tons of new businesses that get created in response to new demand.

It seem to me that if children with parents in the bottom fifth have more upward mobility then there must be more children with parents with higher income with downward mobility.

Not entirely, since the study is based on national income numbers, not relative position within the city. If the city becomes wealthier compared to the rest of the country, it can easily have more people moving up than moving down.

This is based on the "change that a child raised in the bottom fifth of the national income distribution will raise into the top fifth." (again, in national income quintile) Specifically looking at children born in 1980 and 1981 who were US citizens by 2011.

I'd be interested in seeing how the numbers changed, if at all, when adjusting for standard of living.

Among other things, the papers claims that high housing prices (and thus high average mortgage deductions) are correlated with higher intergenerational income mobility. It also finds that, unsurprisingly, higher mean income and higher income growth rate in an area are correlated with higher mobility.

The paper finds very strong positive correlations of religiosity and Social Capital Index with mobility, but very strong negative correlations of the divorce rate and single motherhood with mobility, as strong as anything else in the paper. Those didn't seem to be headlined in the New York Times article, mostly because it wasn't Ross Douthat writing the article.

Roughly speaking, it's interesting that many of the areas that do poorly have high internal migration within the US during the time period studied (whereas many of the areas that do well have high internal migration out, but many of them have high immigration from outside the country.)

Similarly to adjusting for standard of living, I'd like to see what the mobility was for people moving around the income quintiles of the city itself, not the rank based on national income.

There certainly are cities where the percentage of households in the top 20% of national income is so low that it's much more unlikely that anyone would move up into that range. (Some of these areas have so much lower housing costs that for the average budget and preferences, someone can easily live like someone in the top 20% of national income without being there.) Even some of the cities with high per capita and median household incomes, like Raleigh and Charlotte, have lower salaries of the nationally competitive top 20% type jobs, but people are glad to move or live there anyway due to housing costs.

+1 If you are in a poor region, your chances of remaining poor are better than if you start poor in a richer region.

The map looks closely related to the geography of ethnic distribution.

Yep, hard to ignore when you compare the map to this.


Which isn't that surprising, rich people in heavily black areas seem far more likely to have very high barriers for poor outsiders (of any race) to share their social capital. Rich people in monocultural areas are far more likely to have lower barriers and thus more income mobility.

Am I the only one that thinks that a 10% chance of catapulting from the bottom quintile to the top quintile ain't too bad? Where are we expecting these numbers to come out?

Well realistically the maximum upper bound is 20%, right? (Which would assume that your ending quintile is totally random). 10% may not be that bad, but 4% (the number in Atlanta) is a different story.

Upper quintile may not be the best endpoint to use, for the reasons defined by J Thacker above. A better question is how many are able to escape crushing poverty - hit the middle quintile, at least.

I think the maximum upper bound is 100%. You are comparing a locality to the national quintiles. If everyone in the locality started making $1M a year (which is implausible, but wouldn't move the national numbers) they'd all move to the top income quartile nationally. So more realistically, you would expect poor areas that experienced some kind of broad based boom (hello North Dakota!) to have numbers higher than 20%.

Yeah, but don't be distracted by North Dakota. Practically speaking, on a national basis, anything north of 20% would suggest there's a better chance in reaching the top quintile starting from the bottom than from some higher starting point, which is kinda like expecting cold water to boil faster than hot water.

Even a 1 in 20 chance of a kid breaking out from bottom to top looks hopeful. Even this exaggerates the odds. Consider that some candidates prolly look around their classroom and legitimately size up their chances as a lot better than that.

I think the data don't jibe with the gloom-and-doom narrative.

I think I agree. Nationally, this number should be pretty low, and as I mention below, I'd expect it to go down and not up with progress. Locally, you can have weird things happen that aren't very meaningful.

On average, across the nation? It's not obvious to me what you'd want that number to be.

Imagine you had perfect schooling and perfect equality of opportunity. Then you'd get pretty strict stratification by skill endowment, right? With a little allowance for luck. Presumably, with time and as society gets "better," this number will go down, not up.

I suspect you'd 'want' it to be random with respect to endowments other than abilities that you'd want it to be completely explained by. This is probably easier done than said. If a person becomes a doctor in good standing, notwithstanding the education signaling discussion, then they pretty much deserve that position.

We're getting excited that the "top" of the entire national mobility spread is 27% chance of going bottom-20-to-top-40 as though that's so much more successful than the average of around 20% chance? Boy is that sad.

The NW Mississippi Delta is stunningly poor (same is true for some parts of Arkansas and Louisiana on the other side of the river). I would imagine that the top decile earners in those areas may be around national median. In other words, you could go from being the poorest guy you know, to being the richest guy you know, and you might not qualify as having moved up under the NYT definition.

Right, the Mississippi Delta is the home of the blues, the Crossroads, Highway 61.

I grew up there (Poinsett Country, Arkansas), and while it's true that there's a lot of poverty, the income distribution is skewed by some very well-off farmers and landowners. But there's a mobility story there, too. When I was growing up, the rice farmers, in many cases, did very well indeed, but some of them got badly overextended and took a fall in the early 1980s. So like the Forbes list of richest people, it may not be the same set of names every time you look.

"All else being equal, upward mobility tended to be higher in metropolitan areas where poor families were more dispersed among mixed-income neighborhoods."

My story here would be that upwardly mobile people are gentrifiers.

Income mobility from bottom to top 20% is when a smart kid is born to a poor family. I think one relevant question is how well such a smart person does relative to smart kids that had better parental resources.

Or not smart kids that had better parental resources.

But you 'should' separate those 'parental resources' that provide human capital endowments versus network effects and maybe even more disaggregated than that.

The answer is "not well," there was a depressing NYT article about it a while back.

Is there a relationship to geographic mobility? Are those who grow up in the Plains states and end up in higher percentiles still in the Plains states? Or have they moved to other areas of the country with higher incomes relative to the national average (and perhaps costs of living too)?

Good point. The rural areas, particularly the Plains, have been losing population steadily (except for ND). Although agriculture is losing its importance even in rural areas, the income for full time farmers is now above the national average, so the kid who stayed on the farm and bought/rented his neighbors farms has almost certainly risen in the income scale.

Here's the URL for the income study I was thinking of. http://farmdocdaily.illinois.edu/2013/07/Farm-Policy-Income-Farm-Nonfarm.html--household income Farm is now 120 percent of nonfarm.

I'm not sure about interregional brain drain. At least in the upper Midwest, the bigger cities tend to collect the ambitious kids from smaller towns. I'm from northern Wisconsin, most of my (relatively) high earning high school classmates are in Minneapolis, with a few in Madison or Chicago in our early thirties.

I suspect immigration is an important part of the story. Places like CA have a lot of immigrants that are poor when they arrive and wealthy within a generation or two.

Does the study account for out-migration as well? Because I suspect a lot of the low-mobility places are areas where if you get rich, you leave.

It does, in a way. They note that if the parents move to a richer area earlier in the kid's life, then the kid is more likely to end up rich by a national scale.

But note that many of the areas that do poorly are precisely the ones with very large internal in migration (driven in many cases by the cost of living, especially housing.)

My experience tells me that the highest upward mobility is enjoyed by an educated young man who forgoes marriage and breeding and moves to Silicon Valley, Austin, Wall Street, North Dakota or even China.

Today, the educated young man probably comes from educated parents, and therefore, not from the bottom quintile. I.e., this focusing might work to take the 60th percentile guy to the 90th percentile even if he really has only 60th percentile skills, but it doesn't do much for the 7th percentile. And it's sort of an accounting trick, because the non-work sorts of productivity are hard to measure, so we make ourselves look good if we shirk on them.

Further, your description doesn't provide a good argument for using this as a measure of society's goodness. Not that you were trying to do that.

So, is education attainment a cause or effect, or put differently, which parts are which and to what "degree"?

Maybe obtaining the top quintile is simply a two-generation process.

> So, is education attainment a cause or effect, or put differently, which parts are which and to what “degree”?

Ask Bryan Caplan. All I read is what Tyler wrote, so as you might expect I'm absolutely confused.

> Maybe obtaining the top quintile is simply a two-generation process.

Maybe, but it seems to me that skill is given opportunity today. I could believe that this _was_ a two-generation process. To some extent the generational improvement measure might be indicating how poorly success and skill correlated 30 years ago. But even so, that was 1983. It's not like we're talking about the Dark Ages.

So it would seem that politics that emphasize creating racial constituencies (and the associated government programs aimed at those constituencies) have the opposite effect on those constituencies than what is claimed by the politicians and civic leaders. Economic mobility is decreased by concentrating racial minorities...better that they should be distributed throughout the general population and throughout the middle class in particular. But then, that would not benefit politicians and "civic leaders".

The patterns seem to mirror the pattern of mean IQ by region across the country. It would have been nice if they had controlled for mean state IQ as many policy type variables are also correlated with mean IQ and may not have any inependent effect.

Right, from reading the report the "quality of schooling" variable seems to be mainly test scores and drop out rates. Certainly neither of those is independent of IQ and neither is probably that good of an indicator of educational value added.

Anyway I seem to recall reading somewhere, recently, that wealth is far more important than income. So even if that NYC-born son of Puerto Rican natives is pluckily able to raise himself up to the top quintile of national income (say $100 grand pa) how long will it take him to amass any meaningful wealth -- let alone own property! -- in Man-fucking-hattan?

The mainstream economist would have us believe that debt slavery is a good. More debt increases growth and therefore we should ignore wealth and focus on the income that governments will be able to steal from the working class debt slaves(bottom 99%)

"Income mobility was also higher in areas with more two-parent households, better elementary schools and high schools, and more civic engagement, including membership in religious and community groups.

Regions with larger black populations had lower upward-mobility rates."

In other news the world is roundish and etc.

Guess I'd better make it clear that's not an accusation against anyone, just a captain obvious moment for anyone who has lived in this country more than two minutes.

Pointing out the obvious is considered rude in some circles. ;)

Does this study adjust for the massive cost of living differences between areas? Consider the basics of middle class family life -- making enough to afford a house with a yard in a decent public school district? How much income do you have to have for that in New York or Los Angeles vs. the Southeast?

Yes. I saw Raj specifically asked about this in a talk and he said that:

1. They tried applying a few standard of living adjustments and none of the results changed.


2. This was likely due to limited migration and so both parents and children in a place have the same relative standard of living compared to parents and children in other areas.

Interesting that none of the results changed.

2) doesn't really seem like an explanation. The complaint would be that in some areas fewer people are in the top quintile of national household income, so of course a smaller percentage of people starting out poor ends up there.

Top 20% of household income in the USA is roughly $105,000 / year. Various cost of living calculators suggest that a $105k income in the Washington-Arlington-Alexandria metro area is equivalent to about $70,000 / year in the Atlanta, GA metro area. $70,000 / year is 65th percentile household income. People in the DC metro area (of all sorts) may be lot more likely to end up with $105,000 / year than people in the Atlanta, GA metro area, but I would expect people in Atlanta, GA to be more likely to end up with at least $70k / year. (Judging from their internal migration stats, and median and mean income levels.)

I still think that the study is conflating two things-- the income of an area, and income mobility. Mobility to me (if migration were low) would be how many people in the bottom quintile of the local area ended up in the top quintile of that local area. Migration obviously affects that.

Would you be sensible, ladies and gentlemen?

Probably, but I like to hope not.

Looking at family income makes this analysis of little use.

The recent Census Bureau data on “Money Income of Families–Percent Distribution by Income Level, in Constant (2009) Dollars” suggests that there is a group of families getting richer (which I bet are two earner families) and those staying about the same in terms of income (which I bet are single-earner households).

This data needs to be adjusted for single or dual earner families.

"This is based on work by Raj Chetty, Patrick Kline, and Emmanuel Saez"

And Nathan Hendren!

Comments for this post are closed