The Question
Can a city with $424K median top-quintile income and $15.5K bottom-quintile income produce upward mobility — or does the gap itself prevent it?
Figure 1
Upward Mobility: Bottom-to-Top Quintile, DC Region
DC's 4.7% upward mobility rate is among the lowest in the country for major metros. Montgomery County — 7 miles north — is more than 3x higher. This is not a regional pattern. It is a tract-level divergence that maps almost perfectly onto the Anacostia River.
The Claim
DC is a low-mobility city. Not a high-mobility city with a few bad neighborhoods — a structurally low-mobility city whose aggregate statistics are inflated by the federal workforce, suburban commuters counted in metro-area figures, and gentrification-driven income growth that displaces rather than uplifts.
For children born into the bottom quintile in DC proper, the probability of reaching the top quintile by age 35 is 4.7%.[1] That is less than a third of Montgomery County's 16% — a jurisdiction 7 miles north with a shared metro system. It is below the national average of 7.5%. It is below Prince George's County, which has a third of DC's per-capita income.
The pattern is not random. It maps onto race, geography, and the Anacostia River with near-perfect precision. 75% of the lowest-mobility tracts in DC are east of the river,[11] in Wards 7 and 8, which are 95%+ Black. The highest-mobility tracts are in NW DC — Georgetown, Dupont Circle, Cleveland Park — which are majority white and have median household incomes above $140K.
"What would change my mind"
Evidence that children born east of the Anacostia who stay in DC have meaningfully different outcomes than the tract-level data predicts. If the exposure effects are less than ~40% causal (current estimate: ~60%), or if post-2015 investments in Wards 7/8 show measurable mobility gains in the next Opportunity Atlas update.
The Mechanism
Low mobility is not a single failure. It is a five-step causal chain that reproduces itself across generations. Each link has independent empirical support; together they form a self-reinforcing system that Raj Chetty's team has documented at the tract level.
Figure 2
The Neighborhood Mechanism: A 5-Step Causal Chain
Low mobility is not a single failure. It is a chain: segregation concentrates poverty, concentrated poverty produces underfunded schools, underfunded schools limit networks, limited networks block employment, and blocked employment reproduces the starting conditions. The loop arrow is the point — this is a self-reinforcing system. Breaking any single link is insufficient.
Step 1: Residential Segregation
DC's residential patterns are among the most racially sorted in America. Wards 7 and 8 are 95%+ Black with concentrated poverty. This is not self-selection — it is the product of redlining, public housing siting decisions, highway construction (the Anacostia Freeway), and exclusionary zoning that persisted through the 1990s. Massey & Denton (1993)[7] showed that American residential segregation is unique in its intensity and persistence. DC is a textbook case.
Step 2: School Divergence
Concentrated poverty produces under-resourced schools. Schools in Wards 7/8 have lower test scores, higher turnover, and fewer advanced courses. Reardon (2011)[8] documented the tightening link between income segregation and achievement gaps nationally. In DC, the school quality gap between NW and SE is not subtle — it is a 2x difference in proficiency rates. Private and charter options in NW further drain resources and social capital from the public schools that serve the poorest students.
Step 3: Network Isolation
Schools are not just about instruction — they are network formation mechanisms. Children in well-resourced schools build connections to professionals, institutions, and pathways. Children in under-resourced schools build connections to each other. Sampson (2012)[5] documented how "collective efficacy" — the social capital of a neighborhood — compounds across generations. In Wards 7/8, the adults children interact with have fewer professional connections, lower educational attainment, and narrower horizons. This is Bourdieu's habitus made geographic.
Step 4: Employment Barriers
Limited networks produce limited employment options. Youth unemployment in Wards 7/8 exceeds 35% — compared to under 8% in NW DC. Wilson (1987)[6] showed that concentrated joblessness is self-reinforcing: when most adults in a neighborhood are unemployed, the behavioral norms, information channels, and expectations that support employment atrophy. Employers in NW DC are 7 miles away and require transit connections that take 45+ minutes each way.
Step 5: Intergenerational Reproduction
Blocked employment reproduces the starting conditions. Adults who grew up in low-mobility tracts and remain in DC raise children in the same tracts, under the same constraints. Sharkey (2013)[4] showed that neighborhood disadvantage persists across generations — children of parents who grew up in poor neighborhoods are themselves likely to live in poor neighborhoods, even controlling for income. The cycle is the mechanism. Breaking it requires intervention at the neighborhood level, not the individual level.
Evidence & Method
The core data comes from the Opportunity Atlas — Chetty, Friedman, Hendren, Jones & Porter (2018)[1]. The Atlas links Census data to IRS tax records for every American born between 1978 and 1983, tracking them to adulthood. This is not a survey. It is the universe of tax filers, covering 20+ million children at the census tract level (~73,000 tracts nationally).
Methodology
The Opportunity Atlas measures outcomes by childhood tract of residence, not current residence. This matters enormously: it captures where children grew up, not where adults ended up. The key metric — probability of reaching the top quintile from the bottom — is measured at age 35, conditional on parental income quintile.
Chetty & Hendren (QJE, 2018)[2] estimated that approximately 60% of the variation in outcomes across tracts is causal — using a quasi-experimental design based on the age at which children move between neighborhoods. The remaining ~40% is selection (families who move to better neighborhoods may differ in unobserved ways). The 60% causal estimate is based on the linearity of the exposure effect: each additional year in a better neighborhood increases income by ~4%, and this relationship is monotonically linear until approximately age 13.
Figure 4
Exposure Effect: Years in a Better Neighborhood vs. Adult Income Gain
Chetty & Hendren (QJE 2018) found that each additional year a child spends in a higher-opportunity neighborhood increases their adult income by approximately 4%. The effect is roughly linear until age 13, after which it flattens sharply. A child who moves to a better neighborhood at age 5 gains ~32% more income than one who stays. After 13, the critical window has largely closed. This is the strongest evidence that neighborhoods are causal, not just correlated.
The exposure effect is the strongest evidence for causality.[2] If neighborhoods merely sorted families by unobserved characteristics, the age-at-move profile would be flat. Instead, it slopes at ~4% per year — meaning a child who moves from a low-mobility tract to a high-mobility tract at age 5 gains roughly 32% more income than one who moves at age 13. After 13, the effect flattens. The critical window closes.
The Anacostia Divide
The Anacostia River is not a natural boundary in any economic sense — it is a 7-mile-wide chasm in human outcomes. On the west side: Georgetown, Dupont Circle, Capitol Hill. Median household income above $140K. College graduation rates above 80%. On the east side: Congress Heights, Anacostia, Deanwood. Median household income below $38K. College graduation rates below 19%.
The comparison below uses Chetty's five key variables — the ones most predictive of intergenerational mobility. Every single one diverges across the river. This is not a gradient. It is a cliff.
Figure 3
NW DC vs. East of Anacostia: Head-to-Head on Five Chetty Variables
These are not different cities. They are 7 miles apart in the same jurisdiction. NW DC tracts (Georgetown, Dupont Circle) have household incomes 3.7x higher, college graduation rates 4.3x higher, and teen birth rates 14x lower than tracts east of the Anacostia River. Incarceration rates are 5.3x higher east of the river. Every single Chetty variable diverges.
NW DC (Wards 2/3)
Among the highest-opportunity tracts in the country. Children born here have outcomes comparable to the wealthiest suburbs. Proximity to federal institutions, professional networks, and elite schools creates a self-reinforcing advantage loop.
East of Anacostia (Wards 7/8)
Among the lowest-opportunity tracts in the country. Comparable to deep rural poverty in Mississippi and Appalachia — but in the nation's capital, 3 Metro stops from the Capitol. Geographic proximity to wealth does not produce exposure to opportunity.
DC's Income Structure
DC's income distribution helps explain why high aggregate wealth coexists with low mobility. The top quintile captures 54.6% of all income[10] — driven by the federal workforce, lobbying, law, and consulting. The bottom quintile captures 2.2%. That ratio — 24.8:1 — is among the highest in America.
Figure 5
Income Distribution: DC vs. National (Share by Quintile)
DC is one of the most unequal jurisdictions in America. The top quintile captures 54.6% of income — the bottom quintile just 2.2%. The national numbers are bad; DC's are worse. The bottom quintile's share is 27% smaller in DC than nationally. The federal workforce and lobbying sector inflate top-line income statistics while the composition of that income is among the most skewed in the country.
The federal workforce creates a structural floor for high-income employment that inflates DC's median income without creating pathways for residents born into poverty. GS-13 and above positions — the entry point for six-figure federal salaries — require college degrees, security clearances, and professional networks that are systematically absent from Wards 7 and 8. The jobs exist in the same city. The access does not.
Competing Explanations
Three alternative framings deserve serious engagement. Each contains a partial truth; none is sufficient.
"It's selection, not causation"
The argument: families in low-mobility tracts differ in unobserved ways. The neighborhoods don't cause low mobility — they just attract low-mobility families.
Counter: Chetty & Hendren's exposure design[2] estimates ~60% of the variation is causal. The age-at-move linearity is the key evidence: if it were pure selection, moving at age 5 and moving at age 15 would produce identical outcomes. They don't. The slope is ~4% per year.
"The federal workforce distorts the statistics"
The argument: DC's low mobility rate is an artifact of high-income federal workers pulling the top quintile threshold up, making it harder for bottom-quintile children to reach it.
Counter: The Opportunity Atlas uses national quintile thresholds, not local ones. A child in DC is compared to the same income cutoffs as a child in Montana. The 4.7% rate is not an artifact of local inequality — it is a measure of absolute mobility against a national benchmark. Additionally, the Atlas tracks children born in DC, not current residents — suburban commuters are excluded.
"Gentrification is improving conditions"
The argument: rising property values and new development in historically disadvantaged areas signal improvement. Navy Yard, H Street, even parts of Anacostia are seeing investment.
Counter: Gentrification changes who lives in a neighborhood, not the outcomes of who was there. Brummet & Reed (2019)[9] found that gentrification displaces low-income residents to other low-opportunity tracts rather than improving their outcomes. Tract-level statistics improve because the people changed, not because the people improved. The children who grew up in pre-gentrification Anacostia are now in PG County or further out — their outcomes are not captured in DC's improving tract data.
What Would Falsify This?
A thesis is only as strong as the conditions under which it would fail. Four concrete tests:
Causal share below 40%
If improved methodology (e.g., sibling designs, randomized housing voucher follow-ups) showed neighborhood effects explain less than 40% of outcome variation, the structural argument weakens significantly.
Post-2015 mobility gains east of Anacostia
DC has invested significantly in Wards 7/8 since 2015 — new schools, transit, mixed-income housing. If the next Atlas update (2028 cohort) shows meaningful upward mobility gains in these tracts, the "fixed structure" claim needs revision.
Within-tract variation exceeding between-tract variation
If children in the same low-mobility tracts show wildly different outcomes based on family characteristics, the neighborhood mechanism is weaker than the family mechanism. Current data suggests both matter.
Moving to Opportunity long-term results diverge
The MTO experiment showed mixed results at the 10-year mark. Chetty's reanalysis of MTO (2016) found large effects for children who moved young — consistent with the exposure framework. If long-term MTO follow-ups showed no lasting effects, the neighborhood causal story weakens.
So What?
Personal Coda
I walk past the divide every day. I live in a city where the probability of escaping poverty depends on which side of a river you were born on. Not which country. Not which state. Which side of a river, in a city that is 68 square miles.
I went to Brookings and listened to researchers present Chetty's data on a screen. The maps were clean. The gradients were smooth. The tract-level variation was rendered in tasteful color scales. But what the maps show is not tasteful: it is that a child born in Congress Heights has a 4% chance of reaching the top quintile, while a child born in Georgetown — who rides the same Metro system, lives under the same mayor, is counted in the same city — has a 30%+ chance. Same city. Same year. Different world.
The data says ~60% of that is causal. Not family. Not genes. Not culture. Neighborhood. The physical, social, institutional environment that a child cannot choose and largely cannot escape. Every year in a better neighborhood is worth ~4% more income. After age 13, the window closes. That means by the time a child in Ward 8 is old enough to understand their situation, the critical period for intervention has already passed.
This connects to habitus — the inability to dream what you have never seen. To agglomeration — the way cities concentrate opportunity and extract rent for proximity to it. To housing — the financialization that determines who lives where. The Anacostia River is not the cause. It is the symbol. The cause is structural, reproduced at every level, and visible in the data if you are willing to look at it tract by tract.
Sources
Related Reading