If your city isn’t unequal, it’s either poor or exclusionary

Measured income equality, which is sensible goal nationally, is a perversely misleading indicator of which cities are the most just and and inclusive

Income and wealth inequality in the United States are large and growing problems. In the past several decades most of the increased economic output has gone to a tiny fraction of the wealthiest Americans. A long overdue debate about the merits of a wealth tax is finally starting to play out on the national stage.

With the mindset of “thinking globally and acting locally” many people have sought to apply the same kind of metrics used to gauge national income inequality to measure local inequality. Its a seductive analogy: If we should strive to get greater economic equality nationally, then the first step–particularly if you’re frustrated with the federal government or national politics–is to try to achieve equality in your community.

The trouble is that the usual measures of inequality–the 90/10 ratio, the Gini index and others–produce a kind of “fun-house mirror” effect when applied to small geographies. In fact, measuring inequality at the local level gets the identification of which places are inclusive and which are exclusive exactly backwards.

In a nation of widespread and deep inequality, places become economically equal in one of two ways: either by being so undesirable that very few people with the means to live elsewhere will stay, making a community equal in its poverty, or by being so expensive (and usually thanks to land use restrictions) so exclusive that only people of means can live there. Rankings of cities based on inequality measures produce anomalous and inexplicable results: central cities generally show up as “more unequal” than their surrounding suburbs, and the places with the highest levels of equality are often rich suburban enclaves (Bethesda, Maryland or Flower Mound Texas, or economically distressed cities (Detroit, Milwaukee).

The trouble with ranking measured levels of city income inequality

A fairly standard exercise in urban wonkdom is to choose some standard social or economic performance statistic that is available for every city in the US, and then to rank cities, top-to-bottom based on their score on that statistic. (We do this regularly ourselves at City Observatory). If you get the data and concepts right, a ranking tells you who’s doing well, and who’s doing poorly, and by understanding the policies and context of the highly ranked places, you may get some insights one what it takes to improve performance.

In an effort to shed light on inequality, several analysts have ranked US cities or metropolitan areas by their measured levels of inequality. The latest of these comes from Richard Florida and his colleagues at CityLab.

Florida and his team used data from the American Community Survey to rank the nation’s 50 most populous cities according to their Gini Index (a statistical measure of income inequality). By Florida’s reckoning, the City of Atlanta has the highest level of income inequality in the nation and the City of Arlington Texas has the lowest.

It’s worth reflecting for a moment what a statistic like the Gini Index tells us about income levels and income distribution within a city. The Gini Index is a statistical measure of dispersion, it tells us whether all of the observations in a group are very similar or are very different. The Gini statistic varies from 0 to 1, with 0 representing no dispersion (every observation has the same value, in this case, all households have the same level of income) to 1 (complete concentration, in the case of income, one household has all the income). Another way to think about the Gini index (and other measures of inequality) is they are measuring homogeneity: is everyone the same?

It’s also worth noting that the Gini Index makes no distinction between levels of income: If everyone in a city has a very high level of income, it has a low Gini Index; similarly if everyone in a city has avery low level of income, that city, too, has a low Gini Index. So while a Gini index may tell us that a city has homogenous incomes, it really doesn’t tell us whether a city is inclusive or not. In fact, a city that is open to a mix of residents (because it is both attractive to high income households and provides affordable housing options for low income households) will have a higher Gini score (i.e. greater measured inequality) than a city that consists of mostly of households from a single income strata.

As a result, in practice, and especially at the geography of cities, inequality metrics like the Gini Index are misleading measures of social justice and inclusion. If anything, high levels of measured income inequality in small geographies (cities, neighborhoods) are a sign of inclusiveness and income mixing.

The geography of inequality is not fractal

Florida describes urban inequality having a “fractal” nature.

The New Urban Crisis is fractal, recurring at every scale and level of geography, across metros and within them as well. Take the Dallas-Fort Worth metro area for example: Dallas ranks among the cities experiencing the worst of the New Urban Crisis, but two other nearby communities, Arlington and Fort Worth, rank as some of the least unequal cities.

As applied to inequality, this isn’t just incorrect, it’s backwards. “Fractal” means that the pattern repeats itself regardless of scale. If that were true, local inequality would be just a microcosm of metropolitan or national inequality. If anything, the geography of inequality is anti-fractal: high levels of measured inequality at small geographies mean exactly the opposite of what they mean at large geographies. (Also, the fact that there’s so little correlation between metro and city inequality measures on its face indicates that the phenomenon is decidedly not fractal).

This problem also affects Florida’s “New Urban Crisis” Index, which also heavily weights these misleading measures of local inequality and implicitly treats higher levels of measured inequality as indicative of worse performance. The Urban Crisis index consists of four measures, wage inequality, income inequality, economic segregation and housing costs, so half of the index is inequality.

Florida and CityLab are not the first analysts to fall into this trap. Previously, we’ve pointed out similar problems with inequality rankings published by the Brookings Institution’s Metropolitan Policy Program (in 2015) and more recently in a series of posts challenging a ranking of cities published by the Urban Institute (2018).

The real problem: Economic segregation and concentrated poverty

The weakness of city or metro inequality rankings doesn’t mean that the spatial patterns of wealth and poverty aren’t important. It just means that we need to take a more nuanced view of the problem, and regard communities with a robust mix of incomes as inclusive, rather than unequal. And as we and others such Alan Mallach and David Rusk have suggested, our focus ought to be on economic segregation within cities and metros. Concentrated poverty has been repeatedly shown to magnify all of the negative effects of living in poverty, and to transmit these effects to the next generation. Raj Chetty’s research shows that kids from low income families that grow up in mixed income neighborhoods have better lifetime outcomes. And contrary to popular belief, low income households living in mixed income neighborhoods are happier than low income households who live in low income neighborhoods. New experimental evidence from Seattle shows that low income households moving to mixed income areas are more satisfied with their neighborhoods.

Mixed income neighborhoods are a key to promoting more widespread opportunity, and improving the well-being of low income households. As a result, at the local level our first priority should be breaking down patterns of economic segregation–which is the aspect of inequality over which local governments have the most control. It’s ironic and to many, counterintuitive, but if your city is going to be inclusive, it has to have a high level of measured inequality.

Appendix: City Observatory analyses of city inequality

Ranking cities and metro areas by inequality is, unfortunately, a recurring meme in urbanism. We hope it goes away. We’ve repeatedly pointed out the problems with this methodology and its interpretations, and will, if necessary, continue to do so. For reference here are the previous commentaries we’ve published on the subject:

The key points in these posts are summarized as follows:

Cities don’t generate income distributions among their populations, so much as they include (or exclude) different income groups. City inequality is not a linear microcosm of national income inequality.

Highly equal cities are almost always either exclusive suburban enclaves (that achieve homogeneity by rigid zoning limits that exclude the poor) or impoverished cities that have been abandoned by upper and middle income households, leaving them homogeneous but poor. (For example, 8 of Urban’s Institute’s ten “most inclusive cities” are suburbs like Bellevue, Naperville and Santa Clara–among the wealthiest 20 cities in the US; while Detroit and Cleveland are also highly ranked for inclusiveness.

Small geographies, neighborhoods/cities that have high levels of measured income inequality (90/10 ratio, Gini Index) are generally much more inclusive than comparable geographies that with lower levels of measured inequality. Rich and poor are living closer together.

Even ignoring annexation issues, municipal boundaries are so varied as to make city rankings of such economic variables meaningless. The City of Atlanta is roughly 10% of its metro area; San Antonio and Jacksonville are nearly coterminous with their MSAs. A ranking that compares the densest one-tenth of one metro area to ninety-percent of another metropolitan area tells us almost nothing.

## Why its important for your city to be unequal

By Joe CortrightIf your city isn’t unequal, it’s either poor or exclusionaryMeasured income equality, which is sensible goal nationally, is a perversely misleading indicator of which cities are the most just and and inclusiveIncome and wealth inequality in the United States are large and growing problems. In the past several decades most of the increased economic output has gone to a tiny fraction of the wealthiest Americans. A long overdue debate about the merits of a wealth tax is finally starting to play out on the national stage.

With the mindset of “thinking globally and acting locally” many people have sought to apply the same kind of metrics used to gauge national income inequality to measure local inequality. Its a seductive analogy: If we should strive to get greater economic equality nationally, then the first step–particularly if you’re frustrated with the federal government or national politics–is to try to achieve equality in your community.

The trouble is that the usual measures of inequality–the 90/10 ratio, the Gini index and others–produce a kind of “fun-house mirror” effect when applied to small geographies. In fact, measuring inequality at the local level gets the identification of which places are inclusive and which are exclusive exactly backwards.

In a nation of widespread and deep inequality, places become economically equal in one of two ways: either by being so undesirable that very few people with the means to live elsewhere will stay, making a community equal in its poverty, or by being so expensive (and usually thanks to land use restrictions) so exclusive that only people of means can live there. Rankings of cities based on inequality measures produce anomalous and inexplicable results: central cities generally show up as “more unequal” than their surrounding suburbs, and the places with the highest levels of equality are often rich suburban enclaves (Bethesda, Maryland or Flower Mound Texas, or economically distressed cities (Detroit, Milwaukee).

## The trouble with ranking measured levels of city income inequality

A fairly standard exercise in urban wonkdom is to choose some standard social or economic performance statistic that is available for every city in the US, and then to rank cities, top-to-bottom based on their score on that statistic. (We do this regularly ourselves at City Observatory). If you get the data and concepts right, a ranking tells you who’s doing well, and who’s doing poorly, and by understanding the policies and context of the highly ranked places, you may get some insights one what it takes to improve performance.

In an effort to shed light on inequality, several analysts have ranked US cities or metropolitan areas by their measured levels of inequality. The latest of these comes from Richard Florida and his colleagues at CityLab.

Florida and his team used data from the American Community Survey to rank the nation’s 50 most populous cities according to their Gini Index (a statistical measure of income inequality). By Florida’s reckoning, the City of Atlanta has the highest level of income inequality in the nation and the City of Arlington Texas has the lowest.

It’s worth reflecting for a moment what a statistic like the Gini Index tells us about income levels and income distribution within a city. The Gini Index is a statistical measure of dispersion, it tells us whether all of the observations in a group are very similar or are very different. The Gini statistic varies from 0 to 1, with 0 representing no dispersion (every observation has the same value, in this case, all households have the same level of income) to 1 (complete concentration, in the case of income, one household has all the income). Another way to think about the Gini index (and other measures of inequality) is they are measuring homogeneity: is everyone the same?

It’s also worth noting that the Gini Index makes no distinction between levels of income: If everyone in a city has a very high level of income, it has a low Gini Index; similarly if everyone in a city has avery low level of income, that city, too, has a low Gini Index. So while a Gini index may tell us that a city has homogenous incomes, it really doesn’t tell us whether a city is inclusive or not. In fact, a city that is open to a mix of residents (because it is both attractive to high income households and provides affordable housing options for low income households) will have a higher Gini score (i.e. greater measured inequality) than a city that consists of mostly of households from a single income strata.

As a result, in practice, and especially at the geography of cities, inequality metrics like the Gini Index are misleading measures of social justice and inclusion. If anything, high levels of measured income inequality in small geographies (cities, neighborhoods) are a sign of inclusiveness and income mixing.

## The geography of inequality is not fractal

Florida describes urban inequality having a “fractal” nature.

As applied to inequality, this isn’t just incorrect, it’s backwards. “Fractal” means that the pattern repeats itself regardless of scale. If that were true, local inequality would be just a microcosm of metropolitan or national inequality. If anything, the geography of inequality is

anti-fractal: high levels of measured inequality at small geographies mean exactly the opposite of what they mean at large geographies. (Also, the fact that there’s so little correlation between metro and city inequality measures on its face indicates that the phenomenon is decidedly not fractal).This problem also affects Florida’s “New Urban Crisis” Index, which also heavily weights these misleading measures of local inequality and implicitly treats higher levels of measured inequality as indicative of worse performance. The Urban Crisis index consists of four measures, wage inequality, income inequality, economic segregation and housing costs, so half of the index is inequality.

Florida and CityLab are not the first analysts to fall into this trap. Previously, we’ve pointed out similar problems with inequality rankings published by the Brookings Institution’s Metropolitan Policy Program (in 2015) and more recently in a series of posts challenging a ranking of cities published by the Urban Institute (2018).

## The real problem: Economic segregation and concentrated poverty

The weakness of city or metro inequality rankings doesn’t mean that the spatial patterns of wealth and poverty aren’t important. It just means that we need to take a more nuanced view of the problem, and regard communities with a robust mix of incomes as inclusive, rather than unequal. And as we and others such Alan Mallach and David Rusk have suggested, our focus ought to be on economic segregation within cities and metros. Concentrated poverty has been repeatedly shown to magnify all of the negative effects of living in poverty, and to transmit these effects to the next generation. Raj Chetty’s research shows that kids from low income families that grow up in mixed income neighborhoods have better lifetime outcomes. And contrary to popular belief, low income households living in mixed income neighborhoods are happier than low income households who live in low income neighborhoods. New experimental evidence from Seattle shows that low income households moving to mixed income areas are more satisfied with their neighborhoods.

Mixed income neighborhoods are a key to promoting more widespread opportunity, and improving the well-being of low income households. As a result, at the local level our first priority should be breaking down patterns of economic segregation–which is the aspect of inequality over which local governments have the most control. It’s ironic and to many, counterintuitive, but if your city is going to be inclusive, it has to have a high level of measured inequality.

## Appendix: City Observatory analyses of city inequality

Ranking cities and metro areas by inequality is, unfortunately, a recurring meme in urbanism. We hope it goes away. We’ve repeatedly pointed out the problems with this methodology and its interpretations, and will, if necessary, continue to do so. For reference here are the previous commentaries we’ve published on the subject:

The key points in these posts are summarized as follows:

Discussionblog comments powered by Disqus## Related Commentary

History shows IBR modeling is simply wrongThe Week Observed, November 9, 2024IBR: Planning for a world that no longer existsIBR’s DSEIS uses the least accurate forecastIBR contradicts region’s climate commitmentsNeedless purposes: How IBR violates NEPAThe Week Observed, November 1, 2024Metro’s Kate Model: 25,000 phantom cars a day on the I-5 bridge