The difficulty of applying inequality measurements to cities
By Joe Cortright
Earlier this year, our friends at the Brookings Institution released a new tabulation of Census data on levels of inequality in the nation’s largest cities. Inequality, in this case, is measured by dividing the income of a household at the 95th percentile of the population by the income of a household at the 20th percentile. The higher the ratio, the greater the degree of income inequality.
The post has generated a lot of interest in the urban policy world: Some cities, it appears, have a lot more inequality than others.
But a closer look at this data suggest that it paints a misleading picture of the nature of inequality, and some important respects, gets the role of cities in fighting inequality—and, importantly, in reducing concentrated poverty—exactly backwards.
Inequality is a big, national problem
First let’s stipulate a central point: inequality is a big and growing problem in the United States. By virtually any measure, income inequality is as high as it’s been at any time since 1929—the high water mark following the last Gilded Age. The chief aspect of the growth of inequality has been the prodigious gains realized by the top one percent, and among their number, the top tenth, and even one hundredth of one percent. We should further stipulate that Brookings has accurately reported the data that have been tabulated by the Census Bureau. There’s nothing wrong with the math here.
But does computing an income disparity ratio for every city in the United States add anything to our understanding of the extent, geography, or underlying causes of inequality? Is our national inequality problem merely the sum of a vast series of local inequality problems? If a city has a large number of people at the high end and at the low end of the income distribution, does that mean that the city is contributing to the nation’s inequality problem?
In some respects, these data lead us to exactly the wrong conclusions about the nature and geography of inequality. They mute what is the true geographic aspect of inequality: income segregation.
Cities don’t cause poverty
As Ed Glaeser has pointed out, poor people concentrate in cities precisely because they have good transit systems and plenty of jobs. Even if their current income is lower (not necessarily in absolute terms, but relative to the rich people who skew the income distribution) than in smaller places, the lower cost of transportation coupled with job opportunities means that they have a better chance to improve their economic condition over time:
It is critical to recognize that cities rarely make people poor. Rather, cities attract poor people, with economic opportunity, a better social safety net, and the ability to get around, usually without owning cars.
But concentrated poverty does create real problems. Most recently, the major studies released by Raj Chetty and his colleagues have shown how poor neighborhoods reduce the likelihood of economic mobility for their residents. Our own work, including Lost in Place, has shown how durable these neighborhoods can be.
A major contributor to this kind of economic segregation is driven by the secession of the rich. The latest research from Stanford’s Sean Reardon and his colleagues shows income segregation is being driven by the decisions of higher-income families to increasingly isolate themselves in higher-income enclaves, often in exclusive suburbs and established high-income areas.
National inequality is not the sum of local inequality
Moreover, the commonly-cited reasons for growing income inequality have little to do with local policies: Falling value of the minimum wage, R>G, falling effective tax rates on the highest income households, skill-biased technological change, superstar payment, crony-compensation setting, financialization of the economy. Aside from subsidies for sports franchises owned by billionaires, and restrictive zoning that tends to drive up housing prices, there’s precious little cities have to do with generating income inequality per se.
What cities do influence, however, is who lives within their boundaries.
The way this measure is constructed, however, describes places where people have very similar incomes as having lower rates of inequality: if everyone in your community is very low income (i.e., Gary, Indiana), you have income equality. Likewise, if everyone in your community is very high income (i.e., Beverly Hills) you have income equality. The cities in the United States with the highest levels of income equality are exclusive, high income enclaves, and cities of unrelenting poverty.
But if your community contains a mix of high income and low income people, your community will be scored by the 95/20 ratio as having a high level of income inequality. Another word for this might be “diverse and inclusive.”
Are localized inequality statistics a good guide to policy?
From a policy standpoint, the question ultimately has to be whether the measured inequality in cities is susceptible to any meaningful policy solution at the city level. Here it’s helpful to remember that one can attack income inequality at either end of the economic spectrum. City policies that raise the incomes and wages of lower income households (or which lower their cost of living) could clearly ameliorate at least some of the inequality in a city. But it’s a dubious proposition to suggest that cities can (or should) look to address income inequality by reducing the incomes of the well-off. The primary problem is practical: the rich are generally under no obligation to live in a given community, and so easily have the option of simply moving away if faced with effective re-distribution.
The irony here is that policies that encourage the rich to leave your city (or to not live there in the first place) invariably reduce measured inequality. It’s worth noting that Detroit has one of the lowest levels of measured inequality of any large city in the United States.
The lesson here is that for cities, a focus on inequality, while distinctly in rhythm with a serious and growing national malaise, is poor guide to municipal policy. On the other hand, cities ought to have a laser like focus on poverty, especially concentrated poverty.
Is there are “right” geography for localized inequality measures?
In our view, cities are plainly the wrong geography for thinking about inequality. Municipal boundaries of the nation’s largest cities are widely variable; sometimes they cover a majority of a metropolitan area (Jacksonville, San Antonio) and in other cases they are barely 10 percent or more (Atlanta, Miami). Comparing different sized fragments of metro areas can lead to misleading conclusions.
So while Atlanta has the highest 95/20 ratio of any US city, the Atlanta metropolitan area has a level of inequality that is actually below the national average. Atlanta’s high score is influenced by the fact that its municipal boundaries include only about eight percent of the metropolitan area population, and has a bigger share of both high-income and low-income households than the metropolitan area as a whole.
But more fundamentally, the problem is not simply the choice of the optimal geographic units for analysis.
Even at the metropolitan level, the policy implication of the variations in equality is that one ought to have a very, very expensive metro housing market. The metropolitan areas with the most expensive housing in the country (San Jose, Washington, Boston, San Francisco) have some of the lowest levels of inequality. Why? Because poor people can’t afford to live there. At the metro level—as at the municipal level—one way to improve one’s measured equality is simply to exclude the poor. If anything, when measured at the local level, income equality is an indication of income segregation.
And that’s bad on both ends. Having more high income people in your city may increase measured inequality, but it doesn’t make the poor people who live in your city poorer. In fact, at the extremes, having some higher-income people is important to having a tax base that can support the kinds of services that low-income residents rely on. As Alan Berube has acknowledged: “Detroit does not have an income inequality problem—it has a poverty problem. It’s hard to imagine that the city will do better over time without more high-income individuals.”
On the other end, when the rich secede to their own gated suburbs and cities, whether Grosse Pointe or, increasingly, San Francisco, they’re creating relatively “equal” municipalities—but the low-income people left behind are hardly benefiting.
Local inequality measures may be a classic case of “drunk under the streetlight”—we’re looking at the problem because the light shed by the data is good, but it turns out that its not where the problem, and more importantly, the solution can be found.
The difficulty of applying inequality measurements to cities
Earlier this year, our friends at the Brookings Institution released a new tabulation of Census data on levels of inequality in the nation’s largest cities. Inequality, in this case, is measured by dividing the income of a household at the 95th percentile of the population by the income of a household at the 20th percentile. The higher the ratio, the greater the degree of income inequality.
The post has generated a lot of interest in the urban policy world: Some cities, it appears, have a lot more inequality than others.
But a closer look at this data suggest that it paints a misleading picture of the nature of inequality, and some important respects, gets the role of cities in fighting inequality—and, importantly, in reducing concentrated poverty—exactly backwards.
Inequality is a big, national problem
First let’s stipulate a central point: inequality is a big and growing problem in the United States. By virtually any measure, income inequality is as high as it’s been at any time since 1929—the high water mark following the last Gilded Age. The chief aspect of the growth of inequality has been the prodigious gains realized by the top one percent, and among their number, the top tenth, and even one hundredth of one percent. We should further stipulate that Brookings has accurately reported the data that have been tabulated by the Census Bureau. There’s nothing wrong with the math here.
But does computing an income disparity ratio for every city in the United States add anything to our understanding of the extent, geography, or underlying causes of inequality? Is our national inequality problem merely the sum of a vast series of local inequality problems? If a city has a large number of people at the high end and at the low end of the income distribution, does that mean that the city is contributing to the nation’s inequality problem?
In some respects, these data lead us to exactly the wrong conclusions about the nature and geography of inequality. They mute what is the true geographic aspect of inequality: income segregation.
Cities don’t cause poverty
As Ed Glaeser has pointed out, poor people concentrate in cities precisely because they have good transit systems and plenty of jobs. Even if their current income is lower (not necessarily in absolute terms, but relative to the rich people who skew the income distribution) than in smaller places, the lower cost of transportation coupled with job opportunities means that they have a better chance to improve their economic condition over time:
But concentrated poverty does create real problems. Most recently, the major studies released by Raj Chetty and his colleagues have shown how poor neighborhoods reduce the likelihood of economic mobility for their residents. Our own work, including Lost in Place, has shown how durable these neighborhoods can be.
A major contributor to this kind of economic segregation is driven by the secession of the rich. The latest research from Stanford’s Sean Reardon and his colleagues shows income segregation is being driven by the decisions of higher-income families to increasingly isolate themselves in higher-income enclaves, often in exclusive suburbs and established high-income areas.
National inequality is not the sum of local inequality
Moreover, the commonly-cited reasons for growing income inequality have little to do with local policies: Falling value of the minimum wage, R>G, falling effective tax rates on the highest income households, skill-biased technological change, superstar payment, crony-compensation setting, financialization of the economy. Aside from subsidies for sports franchises owned by billionaires, and restrictive zoning that tends to drive up housing prices, there’s precious little cities have to do with generating income inequality per se.
What cities do influence, however, is who lives within their boundaries.
The way this measure is constructed, however, describes places where people have very similar incomes as having lower rates of inequality: if everyone in your community is very low income (i.e., Gary, Indiana), you have income equality. Likewise, if everyone in your community is very high income (i.e., Beverly Hills) you have income equality. The cities in the United States with the highest levels of income equality are exclusive, high income enclaves, and cities of unrelenting poverty.
But if your community contains a mix of high income and low income people, your community will be scored by the 95/20 ratio as having a high level of income inequality. Another word for this might be “diverse and inclusive.”
Are localized inequality statistics a good guide to policy?
From a policy standpoint, the question ultimately has to be whether the measured inequality in cities is susceptible to any meaningful policy solution at the city level. Here it’s helpful to remember that one can attack income inequality at either end of the economic spectrum. City policies that raise the incomes and wages of lower income households (or which lower their cost of living) could clearly ameliorate at least some of the inequality in a city. But it’s a dubious proposition to suggest that cities can (or should) look to address income inequality by reducing the incomes of the well-off. The primary problem is practical: the rich are generally under no obligation to live in a given community, and so easily have the option of simply moving away if faced with effective re-distribution.
The irony here is that policies that encourage the rich to leave your city (or to not live there in the first place) invariably reduce measured inequality. It’s worth noting that Detroit has one of the lowest levels of measured inequality of any large city in the United States.
The lesson here is that for cities, a focus on inequality, while distinctly in rhythm with a serious and growing national malaise, is poor guide to municipal policy. On the other hand, cities ought to have a laser like focus on poverty, especially concentrated poverty.
Is there are “right” geography for localized inequality measures?
In our view, cities are plainly the wrong geography for thinking about inequality. Municipal boundaries of the nation’s largest cities are widely variable; sometimes they cover a majority of a metropolitan area (Jacksonville, San Antonio) and in other cases they are barely 10 percent or more (Atlanta, Miami). Comparing different sized fragments of metro areas can lead to misleading conclusions.
So while Atlanta has the highest 95/20 ratio of any US city, the Atlanta metropolitan area has a level of inequality that is actually below the national average. Atlanta’s high score is influenced by the fact that its municipal boundaries include only about eight percent of the metropolitan area population, and has a bigger share of both high-income and low-income households than the metropolitan area as a whole.
But more fundamentally, the problem is not simply the choice of the optimal geographic units for analysis.
Even at the metropolitan level, the policy implication of the variations in equality is that one ought to have a very, very expensive metro housing market. The metropolitan areas with the most expensive housing in the country (San Jose, Washington, Boston, San Francisco) have some of the lowest levels of inequality. Why? Because poor people can’t afford to live there. At the metro level—as at the municipal level—one way to improve one’s measured equality is simply to exclude the poor. If anything, when measured at the local level, income equality is an indication of income segregation.
And that’s bad on both ends. Having more high income people in your city may increase measured inequality, but it doesn’t make the poor people who live in your city poorer. In fact, at the extremes, having some higher-income people is important to having a tax base that can support the kinds of services that low-income residents rely on. As Alan Berube has acknowledged: “Detroit does not have an income inequality problem—it has a poverty problem. It’s hard to imagine that the city will do better over time without more high-income individuals.”
On the other end, when the rich secede to their own gated suburbs and cities, whether Grosse Pointe or, increasingly, San Francisco, they’re creating relatively “equal” municipalities—but the low-income people left behind are hardly benefiting.
Local inequality measures may be a classic case of “drunk under the streetlight”—we’re looking at the problem because the light shed by the data is good, but it turns out that its not where the problem, and more importantly, the solution can be found.
Related Commentary