A young man holds a sign as people participate in a dance party on North Ave., a day after Baltimore authorities released a report on the death of Freddie Gray, May 2 in Baltimore. (Photo by Patrick Smith/Getty Images)

Before Harvard economists Raj Chetty and Nathaniel Hendren released their groundbreaking new research this week on the role that geography plays in shaping a child's chances of future success, they collected a list of the 100 largest counties in America.

At the top were places — Dupage County in suburban Chicago; Fairfax County, Va. — that dramatically improve a poor child's odds of moving up the economic ladder. At the bottom were the places that have the opposite effect, counties that exert a kind of negative pull on children, where every year of childhood whittles away at a poor kid's odds of thriving as an adult in a way that can be measured by his potential earnings in adulthood. Ranked last on this list: Baltimore City.

"Baltimore is at the bottom," Hendren says. "But it’s really at the bottom for boys."

Every year a poor boy spends growing up in Baltimore, this research found, his earnings as an adult fall by 1.5 percent. Add up an entire childhood, and that means a 26-year-old man in Baltimore earns about 28 percent less than he would if he had grown up somewhere in average America. And that's a whole lot less than the very same child would earn if he had grown up, 50 miles away, in Fairfax County.

That one result — among data Chetty and Hendren have calculated for every county in America — marks a remarkable convergence this week of slow-going social science and current events. If young men in Baltimore who have been protesting for the last two weeks are lashing out at a long legacy of inherited disadvantage, they are also reacting to a reality today that empirical data now confirms: Baltimore is a terrible place to grow up as a poor black boy.

So are cities like it — places with a high degree of economic and racial segregation, where the schools are weak, crime is high and social capital is scarce (scroll below for the full list of the largest 100 counties in the U.S.). Baltimore also reflects a larger racial pattern that emerges in this new analysis -- a pattern that, in our enthusiastic, bipartisan discussion of economic opportunity, we can often gloss over. The places in America where poor kids have the most economic mobility are, by and large, not where blacks live.

Among the 100 largest counties in the U.S., those that appear to offer the greatest social mobility for poor kids tend to have few blacks, such as Dupage County in suburban Chicago. Counties with the worst mobility, on the other hand, have among the largest black populations, like Baltimore City and Wayne County in Detroit.

Only three large counties with black population shares above 20 percent offer positive social mobility for poor kids, as  you can see in the chart below.


The three outlier counties -- those with a sizable black populations and a very small but still positive effect on a child's future incomes -- are Prince George's County, the District of Columbia, and Duval County around Jacksonville, Fla. In all three places, that positive effect amounts to about a 2 percent income gain at age 26 or less.

"Why is it that places with larger African-America populations have more negative effects on upward mobility?" Chetty asks. "We think there’s a set of correlated factors here, where places with larger African-American populations have a lot of other disadvantages. They tend to be much more segregated. They tend to have less investment in things like public schools, and this lack of investment has adverse effects on both African-Americans and on whites."

That means that the downward drag that Baltimore exerts on poor black kids applies to poor white kids, too.

Across the country, these place-based disparities effectively amplify racial inequality. We already know that a black child in America faces longer odds of success than a white child — but this is even more so true for a poor black child in Baltimore. Chetty and Hendren in fact estimate that 20 percent of the earnings gap between blacks and whites in America is due to the differences between where black and white children grow up.

These findings aren't merely a correlation between a place and an earnings outcome; Chetty and Hendren believe they're causal. It's not simply that successful families chose to live in Fairfax and unsuccessful ones pick Baltimore. Baltimore itself appears to be acting on poor children, constraining their opportunity, molding them over time into the kind of adults who will likely remain deeply poor.

That may be because, in Baltimore, poor children must also contend with struggling schools and less social capital, because racial and economic segregation further isolate them away from good schools and better neighborhoods. Change where these children live, though, and you might well change their outcomes.

That finding is remarkable for two reasons. It allows us to begin thinking about what we need to do to turn harmful places into healthy ones (how can Baltimore be more like Fairfax?). And it provides renewed evidence for one straight-forward but long-contested solution to help kids living in poverty: Give their families vouchers to move to better neighborhoods.

This research, in fact, suggests the younger children are when they move, the better. Because the effects of place accumulate over time, a child who spends 10 years in Fairfax is better off than one who spends only her high school years there.

Chetty and Hendren were able to determine this thanks to anonymous tax records they examined from more than five million families who moved from one county to the next between 1996 and 2012. They exploited the difference in ages between siblings in the same family to gauge how the causal effect of place differs with the length of time a child spends there (if you were 3 when your family moved to Fairfax and your brother was 10, you'll get more of an income boost at age 26 than he will).

Chetty, Hendren and Lawrence Katz also found similar results when they re-examined data from a famous government experiment called "Moving to Opportunity" (MTO) that started in the mid-1990s in which HUD offered housing vouchers to poor families to move to better neighborhoods. Research about that experiment has been largely disappointing. The families who moved (relative to others in the experiment who didn't) didn't see dramatic gains. The parents didn't earn more money as a result, and the children didn't appear to do better in school, although there were some health gains for families.

None of those studies, though, have had the benefit of as much passage of time as we have today. When Chetty, Hendren and Katz revisited the children of MTO, now as adults, they found significant benefits for kids who moved before the age of 13 (consistent with the idea that earlier is better). Those kids today earn about 30 percent more than the children who didn't move — in fact, enough that the added taxes they now pay more than offset the costs of the program. They also have higher college attendance rates, they live in better neighborhoods themselves, and they're less likely to be single parents.

Those findings — combined with the much larger national study of American movers — now give renewed hope to the idea that voucher programs could be a powerful anti-poverty tool for some families. For years, it's been puzzling that MTO didn't show those results, that families who left troubled, high-poverty ghettos didn't fare better when they reached the low-poverty suburbs.

"One of the reactions we’ve gotten," Hendren says of this latest work, "is 'well thank God, this doesn’t fly in the face of logic.'"

That is the hopeful side of this big new study. The discouraging part is the map of American geography it reveals today: True social mobility does exist in many places, but those places often aren't where poor black children live.

The percentage gain, or loss in income (at age 26) from growing up in each of the 100 largest counties in the U.S. for children in low-income families

Rank County State County effect, all kids County effect, boys only County effect, girls only
1 Dupage Illinois 15.1% 12.2% 18.2%
2 Snohomish Washington 14.4% 13.9% 14.6%
3 Bergen New Jersey 14.1% 16.6% 11.2%
4 Bucks Pennsylvania 13.3% 16.8% 9.2%
5 Contra Costa California 12.1% 14.5% 9.4%
6 Fairfax Virginia 12.1% 9.2% 15.1%
7 King Washington 11.3% 11.1% 11.4%
8 Norfolk Massachusetts 10.8% 12.4% 8.9%
9 Montgomery Maryland 10.5% 7.5% 13.6%
10 Middlesex New Jersey 8.6% 7.8% 9.4%
11 Montgomery Pennsylvania 7.9% 4.4% 11.6%
12 Ventura California 7.4% 10.9% 3.5%
13 Middlesex Massachusetts 6.5% 7.6% 5.2%
14 Macomb Michigan 5.6% 2.5% 8.9%
15 San Mateo California 5.6% 4.2% 7.0%
16 Hudson New Jersey 4.9% 10.4% -1.1%
17 Salt Lake Utah 4.4% -0.9% 10.2%
18 Pierce Washington 3.8% 5.5% 2.0%
19 Providence Rhode Island 3.8% 6.5% 0.8%
20 Kern California 3.7% 6.0% 1.1%
21 Monmouth New Jersey 3.6% 0.6% 6.8%
22 San Diego California 3.3% 1.6% 5.2%
23 Worcester Massachusetts 2.7% 1.2% 4.4%
24 Hennepin Minnesota 2.7% 4.8% 0.3%
25 Hartford Connecticut 2.6% 5.0% -0.1%
26 Erie New York 2.6% 7.2% -2.6%
27 Maricopa Arizona 2.5% -2.6% 8.1%
28 Clark Nevada 2.5% -1.2% 6.6%
29 Multnomah Oregon 2.2% 7.9% -4.2%
30 Prince Georges Maryland 2.0% 6.5% -2.9%
31 Honolulu Hawaii 1.9% 0.5% 3.4%
32 Santa Clara California 1.6% -2.2% 5.8%
33 Pinellas Florida 1.4% 1.8% 0.9%
34 San Francisco California 1.1% -5.5% 8.4%
35 Suffolk New York 0.9% 6.6% -5.5%
36 Duval Florida 0.6% 5.2% -4.5%
37 District Of Columbia D.C. 0.5% 2.8% -2.0%
38 Tarrant Texas 0.3% 2.0% -1.5%
39 El Paso Texas -0.3% 0.6% -1.2%
40 Oakland Michigan -0.5% -6.1% 5.7%
41 Sacramento California -0.6% -2.9% 2.0%
42 Dade Florida -0.6% -6.6% 6.0%
43 Nassau New York -0.7% 1.5% -3.1%
44 Harris Texas -0.7% -3.3% 2.1%
45 Tulsa Oklahoma -1.4% -6.2% 3.9%
46 Essex Massachusetts -1.5% -2.2% -0.6%
47 Lake Illinois -1.8% -7.3% 4.2%
48 Allegheny Pennsylvania -2.1% -2.5% -1.6%
49 Franklin Ohio -2.1% 7.6% -12.9%
50 Westchester New York -2.3% -9.6% 5.8%
51 St Louis Missouri -2.4% -2.7% -2.0%
52 Queens New York -2.5% -6.1% 1.5%
53 Cuyahoga Ohio -2.9% 1.7% -7.9%
54 Philadelphia Pennsylvania -2.9% -7.3% 2.0%
55 Hamilton Ohio -3.1% 2.9% -9.7%
56 Orange California -3.2% -10.7% 5.1%
57 Suffolk Massachusetts -3.2% 4.3% -11.5%
58 Jefferson Kentucky -3.5% -0.2% -7.1%
59 Broward Florida -3.8% -3.4% -4.2%
60 San Joaquin California -3.9% -7.1% -0.4%
61 Baltimore Maryland -4.4% -11.0% 2.9%
62 Jackson Missouri -4.6% -6.9% -2.0%
63 Hidalgo Texas -4.6% -9.2% 0.4%
64 DeKalb Georgia -4.7% -1.4% -8.2%
65 Kent Michigan -4.9% -7.8% -1.7%
66 Dallas Texas -5.1% -7.9% -2.0%
67 Oklahoma Oklahoma -5.6% -7.6% -3.4%
68 Alameda California -5.6% -6.7% -4.4%
69 Kings New York -5.8% -8.5% -2.7%
70 Gwinnett Georgia -6.0% -14.2% 3.0%
71 Cobb Georgia -6.9% -11.6% -1.7%
72 Bexar Texas -7.1% -12.0% -1.6%
73 Travis Texas -7.5% -10.4% -4.3%
74 Bernalillo New Mexico -7.6% -14.0% -0.4%
75 Davidson Tennessee -7.8% -5.7% -10.0%
76 Fairfield Connecticut -8.3% -13.5% -2.5%
77 New Haven Connecticut -8.4% -15.0% -1.0%
78 Essex New Jersey -8.6% -4.8% -12.7%
79 Montgomery Ohio -8.9% -9.0% -8.7%
80 San Bernardino California -9.0% -11.9% -5.6%
81 Monroe New York -9.1% -13.9% -3.7%
82 Shelby Tennessee -9.6% -9.0% -10.1%
83 Jefferson Alabama -10.1% -10.8% -9.3%
84 Los Angeles California -10.6% -13.0% -8.0%
85 New York New York -10.9% -7.0% -14.9%
86 Riverside California -11.2% -17.0% -4.7%
87 Palm Beach Florida -11.3% -16.5% -5.5%
88 Wake North Carolina -11.4% -13.4% -9.1%
89 Fulton Georgia -11.6% -11.6% -11.5%
90 Marion Indiana -12.1% -8.8% -15.5%
91 Pima Arizona -12.2% -23.0% 0.0%
92 Bronx New York -12.3% -15.2% -9.0%
93 Milwaukee Wisconsin -12.3% -14.8% -9.4%
94 Wayne Michigan -12.5% -17.4% -6.9%
95 Fresno California -12.9% -16.8% -8.5%
96 Cook Illinois -13.3% -13.7% -12.8%
97 Orange Florida -13.5% -14.6% -12.0%
98 Hillsborough Florida -13.5% -16.3% -10.2%
99 Mecklenburg North Carolina -13.8% -12.8% -14.7%
100 Baltimore City Maryland -17.3% -27.9% -5.4%

Dan Keating and Chris Ingraham contributed to this post.