Race and Economic Opportunity in the United States An Intergenerational Perspective
Raj Chetty, Stanford Nathaniel Hendren, Harvard Maggie R. Jones, U.S. Census Bureau Sonya Porter, U.S. Census Bureau
March 2018
Disclaimer: The views expressed are not necessarily those of the U.S. Census Bureau. The statistical summaries reported in these slides have been cleared by the Census Bureau's Disclosure Review Board release authorization number CBDRB-FY18-195. All values in the tables and figures that appear in this presentation have been rounded to four significant digits as part of the disclosure avoidance protocol.
1 Median Household Income by Race and Ethnicity in 2016 80 $80,700
60 $63,200
$46,900 40 $38,600 $39,700
20 Median Median Household ($1000) Income
0 White Black Asian Hispanic American Indian 2 Note: We focus here and in subsequent analyses on four non-Hispanic single-race groups (white, black, Asian, American Indian and Alaska Native) and Hispanics. Source: American Community Survey 2016. Theories of Racial Disparities Family-Level Factors Parental Income Magnuson & Duncan 2006; Rothstein & Wozny 2012 Parental Human Capital & Wealth Oliver & Shapiro 1995; Orr 2003; Conley 2010 Family Structure and Stability McAdoo 2002; Burchinal et al. 2011
Ability at Birth Rushton & Jensen 2005 vs. Fryer & Levitt 2006 Structural Features of Environment Segregation, Neighborhoods Massey & Denton 1993; Wilson 1987; Sampson and Wilson 1995; Smith 2005 School Quality Card & Krueger 1992; Jencks & Phillips 1998; Dobbie & Fryer 2011 Discrimination in the Labor Market Donohue & Heckman 1992; Heckman 1998; Pager 2003; Bertrand & Mullainathan 2004 Discrimination in Criminal Justice Steffensmeier, Ulmer, Kramer 1998; Eberhardt et al. 2004; Alexander 2010
Social Alienation, Stereotype Threat Steele & Aaronson 1995; Tatum 2004; Glover, Pallais, Pariente 2017 Cultural Factors and Social Norms Identity and Oppositional Norms Fordham & Ogbu 1986; Noguera 2003; Carter 2005; Austen-Smith & Fryer 2005 Aspirations or Role Models Mickelson 1990; Small, Harding, & Lamont 2010
3 This Paper: An Intergenerational Perspective . Prior work has typically studied racial disparities within a single generation
– Exceptions: school district data, longitudinal survey data, qualitative studies [e.g., Card and Rothstein 2007, Reardon et al. 2016, Mazumder 2014, Lareau 2003]
. We take an intergenerational perspective, focusing on dynamics of income across generations
– Use new de-identified data linking parents and children covering nearly the entire U.S. population from 1989-2015
. Intergenerational approach sheds light on which disparities will persist in the long run and allows us to isolate the factors that drive persistent gaps
4 1 Data and Sample Definitions
2 Intergenerational Mobility by Race
3 Marriage Rates and Gender Differences
4 Family Level Explanations
5 Neighborhood Level Explanations
5 1 Data and Sample Definitions
2 Intergenerational Mobility by Race
3 Marriage Rates and Gender Differences
4 Family Level Explanations
5 Neighborhood Level Explanations
6 Data and Sample Definitions
. Data sources: Census data (2000, 2010, ACS) covering U.S. population linked to federal income tax returns from 1989-2015 [Akee, Jones, and Porter 2017]
. Intergenerational linkage: Children linked to parents who first claim them as a dependent on a tax return
. Target sample: Children in 1978-83 birth cohorts who were born in the U.S. or are authorized immigrants who came to the U.S. in childhood
. Analysis sample: 20 million children, 94% coverage rate of target sample
7 Income Measures
. Parents’ pre-tax household incomes: mean Adjusted Gross Income from 1994-2000, assigning non-filers zeros
. Children’s pre-tax incomes measured in 2014-15 (ages 31-37)
– Non-filers assigned incomes based on W-2’s (available since 2005)
– Begin with household income, then turn to individual (own) income
. Focus on percentile ranks: rank children relative to others in their birth cohort and parents relative to other parents
8 1 Data and Sample Definitions
2 Intergenerational Mobility by Race
3 Marriage Rates and Gender Differences
4 Family Level Explanations
5 Neighborhood Level Explanations
9 Intergenerational Mobility by Race
. Organize empirical analysis using a statistical model of intergenerational mobility and inequality [Becker and Tomes 1979]
– Let i index families, t index generations, and r(i) denote race of family i
– Model child’s income rank as a race-specific linear function of parent’s income rank:
– Evolution of racial gaps and steady-state disparities in mean ranks controlled by
rates of relative and absolute mobility (ar, br)
10 Intergenerational Mobility in the United States
Mean Child Household Income Rank vs. Parent Household Income Rank
100 80
Slope: 0.351 (0.003)
60
40
20
Mean Child MeanChild Household Rank Income 0 0 20 40 60 80 100 ($22K) ($43K) ($69K) ($104K) ($1.1M) 11 Parent Household Income Rank
Convergence in Black-White Gap if Intergenerational Mobility is Race-Invariant
100
80
60 40
Mean Black Mean White
20 Parent Rank Parent Rank
Mean Child MeanChild Household Rank Income 0 32.7 57.9 0 20 40 60 80 100
Parent Household Income Rank 12
Convergence in Black-White Gap if Intergenerational Mobility is Race-Invariant
100
80
60 40
Mean Black Gap = 25.2 Mean White
20 Parent Rank Parent Rank
Mean Child MeanChild Household Rank Income 0 32.7 57.9 0 20 40 60 80 100
Parent Household Income Rank 13
Convergence in Black-White Gap if Intergenerational Mobility is Race-Invariant
100
80 60
Mean Rank of Black Children
44.8 40
Mean Black Gap = 25.2 Mean White
20 Parent Rank Parent Rank
Mean Child MeanChild Household Rank Income 0 32.7 57.9 0 20 40 60 80 100
Parent Household Income Rank 14
Convergence in Black-White Gap if Intergenerational Mobility is Race-Invariant
100
80 60 Mean Rank of White Children 53.6 Mean Rank of Black Children
44.8 40
Mean Black Gap = 25.2 Mean White
20 Parent Rank Parent Rank
Mean Child MeanChild Household Rank Income 0 32.7 57.9 0 20 40 60 80 100
Parent Household Income Rank 15
Convergence in Black-White Gap if Intergenerational Mobility is Race-Invariant
100
80 60
53.6 Pred. Gap in Next Gen. = 8.8
44.8 40
Mean Black Gap = 25.2 Mean White
20 Parent Rank Parent Rank
Mean Child MeanChild Household Rank Income 0 32.7 57.9 0 20 40 60 80 100
Parent Household Income Rank 16
Convergence in Black-White Gap if Intergenerational Mobility is Race-Invariant
100
80
60 40 Next Gen.
Gap = 8.8
20
Mean Child MeanChild Household Rank Income 0 44.8 53.6 0 20 40 60 80 100
Parent Household Income Rank 17
Convergence in Black-White Gap if Intergenerational Mobility is Race-Invariant 100
If intergen mobility did not vary by race,
racial disparities would shrink rapidly across generations
80 60 Gen. 2 Gap = 3.1 52.1
49.0 40 Next Gen.
Gap = 8.8
20
Mean Child MeanChild Household Rank Income 0 44.8 53.6 0 20 40 60 80 100
Parent Household Income Rank 18 Mean Child Household Income Rank 0 20 40 60 80 100 0 White Black Diff. at12.6 p=25: Intergenerational Mobility for Whites vs. Blacks 20 ParentIncome Household Rank 40 60 Diff. at15.7 p=75: 80 Diff. 12.4at p=100: 100 19
Rates of Upward and Downward Mobility: Blacks vs. Whites 50
41%
40 30
20 18%
17% Percentage Children of Percentage 11%
10 9%
3% 0
Kids Q5 Given Kids Q1 Given Kids Q5 Given Parents Q1 Parents Q5 Parents Q5 White Black 20 Click here to view an interactive depiction of these transition rates
Intergenerational Mobility for Whites vs. Blacks
100 80
Whites' 60 Steady State
54.4
40
20
Mean Child MeanChild Household Rank Income 0 54.4 0 20 40 60 80 100
Parent Household Income Rank 21
Intergenerational Mobility for Whites vs. Blacks
100 80
Whites' 60 Steady State
54.4 40
35.2 20
Mean Child MeanChild Household Rank Income Blacks'
Steady State 0 35.2 54.4 0 20 40 60 80 100
Parent Household Income Rank 22
Intergenerational Mobility for Whites vs. Blacks
100
80
60
40 20
Mean Child MeanChild Household Rank Income Steady-State
Gap = 19.2 0 35.2 54.4 0 20 40 60 80 100
Parent Household Income Rank 23
Intergenerational Mobility for Whites vs. Blacks 100 Intergenerational gaps racial disparities persist in steady state
Current gap is close to steady state intergenerational gaps
80 (not transitory factors) drive most of the black-white gap today
60
40 20
Mean Child MeanChild Household Rank Income Steady-State
Gap = 19.2 0 35.2 54.4 0 20 40 60 80 100
Parent Household Income Rank 24 Mean Child Household Income Rank 0 20 40 60 80 100 0 Mean Child Income Rank Parent vs. Income Rank by Race and Ethnicity 20 ParentIncome Household Rank 40 60 80 AmericanIndian Black White 100 25 Mean Child Household Income Rank 0 20 40 60 80 100 0 Mean Child Income Rank Parent vs. Income Rank by Race and Ethnicity 20 ParentIncome Household Rank 40 60 80 Hispanic AmericanIndian Black White 100 26 Mean Child Household Income Rank 0 20 40 60 80 100 0 Mean Child Income Rank Parent vs. Income Rank by Race and Ethnicity 20 ParentIncome Household Rank 40 60 80 Asian Hispanic AmericanIndian Black White 100 27 Mean Child Household Income Rank 0 20 40 60 80 100 0 Mean Child Income Rank vs.Parent Child and EthnicitybyMean IncomeRank Race Income 20 Children with MothersU.S.Childrenwith born inthe ParentIncome Household Rank 40 60 80 Asian Hispanic AmericanIndian Black White 100 28 Changes in Income Across Generations, by Racial Group
Mean for white children (born 1978-83)
Legend Black $23K Parents
Children American Indian $22K Steady-State Prediction
Hispanic $9K
30 40 50 60 Mean Household Income Percentile
29 Intergenerational Persistence of Racial Disparities: Summary
. All racial groups in the U.S. have similar rates of relative mobility will converge rapidly to steady state
. Key driver of disparities is therefore intergenerational gap in absolute mobility, e.g. between blacks and whites
. Why do black children have lower incomes than white children conditional on parent income?
. Rest of the talk: test a range of explanations for black-white intergenerational gaps
30 1 Data and Sample Definitions
2 Intergenerational Mobility by Race
3 Marriage Rates and Gender Differences
4 Family Level Explanations
5 Neighborhood Level Explanations
31 Mechanical Effects of Household Size
. Well-known that blacks marry at much lower rates than whites
. Do differences in marriage rates create mechanical differences between the household incomes of blacks and whites?
. Examine marriage rates and children’s individual incomes by parental income
32
Marriage Rates vs. Parent Income, Blacks vs. Whites
70
37)
-
60 50
Diff. at p=75: 34.2 40
Diff. at p=25: 32.1
30
20 10 White
Black
Percent of Children Married 32 Children of in (Ages 2015 Percent 0
0 20 40 60 80 100
Parent Household Income Rank 33
Black-White Gap in Child Individual Income Rank vs. Parent Income Rank
80 60
Diff. at p=75: 5.6 40
Diff. at p=25: 4.2 Mean Child MeanChild Individual Rank Income White
Black 20
0 20 40 60 80 100
Parent Household Income Rank 34 Black-White Gap in Child Individual Income Rank vs. Parent Income Rank
Male Children
80 60
Diff. at p=75: 12.0 40
Diff. at p=25: 9.7 Mean Child MeanChild Individual Rank Income White
Black 20
0 20 40 60 80 100
Parent Household Income Rank 35 Black-White Gap in Child Individual Income Rank vs. Parent Income Rank
Female Children
80 60
Diff. at p=75: -1.0 40
Diff. at p=25: -1.4 Mean Child MeanChild Individual Rank Income White
Black 20
0 20 40 60 80 100
Parent Household Income Rank 36 Hourly Wage Rates vs. Parent Income
Female Children
80 60
Diff. at p=75: 1.5 40
Mean Child Wage Rank (Age Rank >=(Age 30) MeanChild Wage Diff. at p=25: 1.9 White
Black 20
0 20 40 60 80 100
Parent Household Income Rank 37 Employment Rates vs. Parent Income Rank
Female Children
100
90 80 Diff. at p=75: -2.4
Diff. at p=25: -2.0
70 60
White
Percent of Children Working in ACS (Age >=(Age 30) ACS in Working Children of Percent Black 50
0 20 40 60 80 100
Parent Household Income Rank 38 Percent of Children Working (Age >= 30) 50 60 70 80 90 100 0 Employment Employment Rates vs.Parent Income Rank 20 Diff. at18.9 p=25: ParentIncome Household Rank Male Children 40 60 Diff. at11.4 p=75: 80 Black Males White Males 100 39 Percent of Children Working (Age >= 30) 50 60 70 80 90 100 0 Employment Employment Rates vs.Parent Income Rank 20 ParentIncome Household Rank 40 60 80 Black FemalesBlack WhiteFemales Black Males WhiteMales 100 40 College Attendance Rate for Children (%) 20 40 60 80 100 0 College Attendance College Attendance Rates Parent vs. RankIncome 20 ParentIncome Household Rank 40 60 80 Black FemalesBlack WhiteFemales Black Males WhiteMales 100 41 Incarceration Rates vs. Parent Income Rank
Male Children 32) - White
20 Black 15
Diff. at p=25: -8.2 10
5 Diff. at p=75: -3.2 0
Pct. of Children Incarcerated on April 1, 2010 (Ages April 27 (Ages 2010 on 1, Incarcerated Children of Pct. 0 20 40 60 80 100
Parent Household Income Rank 42 Pct. of Children Incarcerated on April 1, 2010 (Ages 27-32) 0 5 10 15 20 0 Incarceration Rates vs.Parent Income Rank 20 ParentIncome Household Rank Female ChildrenFemale 40 60 80 Black White 100 43 Incarceration and Intergenerational Gaps
. Differences in incarceration rates are substantial, but unlikely to “mechanically” explain entirety of black-white income gap for males – Income gaps remains substantial even among children in the highest-income families – Incarcerated individuals have low earnings even prior to incarceration [Looney and Turner 2018] – Would be useful to quantify impacts of incarceration directly using panel data on incarceration in future work
. We treat incarceration as an outcome determined by the same processes that shape labor market outcomes
44 Gender Differences in Racial Disparities: Summary
. Black-white gaps in earnings conditional on parental income are large for men, but small for women
. Does not imply that black women have the same level of welfare as white women
– Black women have lower household income, conditional on parent income
. Also does not mean that incomes of black women will converge to those of white women across generations
– Black women grow up in lower-income households in each generation
. But does suggest that addressing the unique challenges faced by black men may ultimately raise the incomes of both black men and women
45 1 Data and Sample Definitions
2 Intergenerational Mobility by Race
3 Marriage Rates and Gender Differences
4 Family Level Explanations
5 Neighborhood Level Explanations
46 Explaining the Black-White Intergenerational Income Gap Parental Education, Wealth, and Family Structure
. Do family-level factors (e.g., parental wealth) explain intergenerational gaps between black and white men?
. Condition on family-level characteristics to answer this question
47 Effects of Family-Level Factors on the Black-White Income Gap
Children with Parents at 25th Percentile 20
17.6 Male 15
10.0
10 9.3 9.1
8.4
5
0
Mean Rank of White MeanRank Black Minus White of
5 -
Controls: None Par. Inc. Par Inc. Par Inc. Par Inc. +Two-Par. +Two-Par. +Two-Par. +Educ. +Educ. +Wealth
48 Effects of Family-Level Factors on the Black-White Income Gap
Children with Parents at 25th Percentile 20 17.6 Male
Female 15
10.0
10 9.3 9.1 8.4
4.8 5
-2.0 -1.7 -1.9 -2.3
0
Mean Rank of White MeanRank Black Minus White of
5 -
Controls: None Par. Inc. Par Inc. Par Inc. Par Inc. +Two-Par. +Two-Par. +Two-Par. +Educ. +Educ. +Wealth
49 Explaining the Black-White Intergenerational Income Gap Differences in Ability
. Ability hypothesis is inconsistent with gender heterogeneity in intergenerational gaps
1. No ex-ante reason that racial differences in ability would produce differences in outcomes for boys but not girls
2. Prior arguments for ability diffs. based on test score gaps, but black- white test score gaps do not vary by gender
50 Test Scores at Age 9 for Low-Income (Free-Lunch Eligible) Students
National Assessment of Educational Progress 2012
.5 0
-0.16 -0.16
.5 -
Math Test Score at Age Age 9 Score at Test Math -0.61
In SD From National Average SD In From National -0.64
1 - Boys Girls
White Black 51 Explaining the Black-White Intergenerational Income Gap Differences in Ability
. Ability hypothesis is inconsistent with gender heterogeneity in intergenerational gaps
1. No ex-ante reason that racial differences in ability would produce differences in outcomes for boys but not girls
2. Prior arguments for ability diffs. based on test score gaps, but black- white test score gaps do not vary by gender
• Test scores may not be an accurate measure of ability for black children, e.g. because of test bias or stereotype threat [Steele et al. 1995, Jencks et al. 1998]
52 1 Data and Sample Definitions
2 Intergenerational Mobility by Race
3 Marriage Rates and Gender Differences
4 Family Level Explanations
5 Neighborhood Level Explanations
53 Neighborhood Environments and the Black-White Gap
. Do blacks have worse outcomes than whites because they live in different neighborhoods?
. Begin by examining broad geographic variation across commuting zones [Chetty, Hendren, Kline, and Saez 2014]
– Assign children to locations in proportion to the fraction of their childhood that they spent in each CZ
. Estimate expected rank of children with parents at the 25th percentile of national income distribution using linear regression within each CZ
54 Source: Chetty , Hendren
, Kline, , Mean Child Rank in National Income Distribution 20 30 40 50 60 70 Saez 0 2014 Mean Child Income Rank at Age 30 vs. Parent Income Rank 10 for Children Born in 1980 in and RaisedChicago 20 ParentIncome in National Rank Distribution 푦 ത + intercept+ of rank Predictoutcome for childin CZ c usingslope 30 25,Chicago,1985 40 = 푦 ത 50 0,Chicago,1985 - rank relationshiprank 60 + (Rank + 70 - Rank Slope) Rank 80 × 25 90 100 55 The Geography of Upward Mobility in the United States Average Individual Income for Males with Parents Earning $25,000 (25th percentile)
Seattle Salt Lake City $31.2k $27.6k Dubuque $31.7k Cleveland $24.6k
Newark $30.1k San Francisco Bay Area Washington DC $28.8k $27.8k
56.9 ($34.5k) Los Angeles Charlotte $27.6k $22.5k Atlanta $23.1k 49.0 ($27.6k)
42.9 ($22.8k) 56 Note: Green = More Upward Mobility, Red = Less Upward Mobility Two Americas: The Geography of Upward Mobility by Race Average Individual Income for Boys with Parents Earning $25,000 (25th percentile)
Black Men White Men Boston $31k Boston $24k Newark Newark $20k $32k
San San Francisco Francisco $19k $31k Atlanta Atlanta $26k $18k
<36.5 45.8 >56.9 ($17k) ($25k) ($35k)
Note: Green = More Upward Mobility, Red = Less Upward Mobility; Grey = Insufficient Data 57 Neighborhood Environments and the Black-White Gap
. Commuting-zone level variation illuminates broad regional patterns but does not directly test for “neighborhood” effects
. Blacks live in different neighborhoods from whites within CZs
. Zoom in to examine variation across Census tracts in the rest of the lecture
– 70,000 Census tracts with about 4,250 people per tract in the U.S.
58 Variation in the Black-White Gap Across Tracts
. Four key results:
1. Black boys have lower earnings than white boys in 99% of Census tracts in America, controlling for parental income
59 Black-White Gaps within Neighborhoods by Gender
Children with Parents at 25th Percentile 20 17.6 Male
Female 15
10.0 10 7.7 7.0
5 4.8
-2.0 -2.2 -2.7
0
Mean Rank of Whites Minus Minus Black MeanChildren Rank Whites of
5 -
Controls: None Par. Inc. Same Tract + Same Block + p=25 Par. Inc. Par. Inc. p=25 p=25
60 Black-White Gaps within Neighborhoods by Gender
Children with Parents at 75th Percentile 20 17.6 Male
Female 15
11.7
10 8.9 7.9
5 4.8
-0.9 -1.7 -2.6
0
Mean Rank of Whites Minus Minus Black MeanChildren Rank Whites of
5 -
Controls: None Par. Inc. Same Tract + Same Block + p=75 Par. Inc. Par. Inc. p=75 p=75
61
Distribution of Black – White Gap in Individual Ranks Across Tracts for Men
.08
.06
.04
Density .02 Raw Fraction < 0: 11.8% Signal Fraction < 0: 1.3%
Mean Gap: 7.5 pctiles 0
-40 -20 0 20 40
White Minus Black Rank Given Parents at 25th Percentile 62 Variation in the Black-White Earnings Gap Across Tracts
. Four key results:
1. Black boys have lower earnings than white boys in 99% of Census tracts in America, controlling for parental income
2. Both black and white boys have better outcomes in “good” (e.g., low-poverty, higher rent) neighborhoods, but the black-white gap is bigger in such areas
63 Correlations between Tract-Level Characteristics and Incomes of Black vs. White Men Children with Parents at 25th Percentile Economy Share Above Poverty Line Mean Household Income Employment Rate
Schools Mean 3rd Grade Math Score Mean 8th Grade Math Score Share HS Students Not Suspended
Educ. Share High School Grad. Attainment Share College Grad.
Housing Median Rent (2BR) Share Homeowners
Family Structure Share Married Share Two-Parent Healthcare Access Share Adults Insured
White Black 0 0.2 0.4 0.6 Magnitude of Correlation 64 Black – White Gap in Individual Income Ranks vs. Share Above Poverty Line
Males
8.5
8
7.5
7 6.5
Intercept: 3.57, Slope: 0.05
6 White Minus Black Mean Children Minus Rank Black MeanWhite Income Ind. Children
50 60 70 80 90 100
Share Above Poverty Line in Tract in 2000 (%) 65 Variation in the Black-White Earnings Gap Across Tracts
. Four key results:
1. Black boys have lower earnings than white boys in 99% of Census tracts in America, controlling for parental income
2. Both black and white boys have better outcomes in “good” (e.g., low-poverty, higher rent) neighborhoods, but the black-white gap is bigger in such areas
3. Within low-poverty areas, there are two factors associated with better outcomes for black boys and smaller gaps: greater father presence and less racial bias
66 Percentage of Tracts in which Predicted Rank of Black Males is above National Median
vs. Share above Poverty Line 15
Poverty Rate Below 10%
Inc.
-
10
5
Tracts with with MeanRank Low of Tracts
Black Males is Above National Median Black Above Males is (%) 0
40 60 80 100
Share Above Poverty Line in Tract in 2000 (%) 67 Association Between Tract-Level Characteristics and Black-White Gap Tracts with Poverty Rates Below 10% Black Father Presence (p25) White Father Presence (p25) Share Married Share Black Above Poverty Line Mean 3rd Grade Math Score Employment Rate White Mother Presence (p25) Share Black Insured (18-64) Share Homeowners Share Two Parents Black Mother Absence (p25) Share Above Poverty Line Mean 8th Grade Math Score Fraction Not Suspended from School Share Insured (18-64) Median Rent (2BR) Share White Above Poverty Line Share White Insured (18-64) Share HS Graduate Median Black Home Value Median White Home Value Mean Household Income Share College Grad.
-0.15 -0.1 -0.05 0 0.05 0.1 0.15
Correlation with White Minus Black Rank for Males (p25) 68 Association Between Tract-Level Characteristics and Black-White Gap Tracts with Poverty Rates Below 10% Black Father Presence (p25) White Father Presence (p25) Share Married Share Black Above Poverty Line Mean 3rd Grade Math Score Employment Rate White Mother Presence (p25) Share Black Insured (18-64) Share Homeowners Share Two Parents Black Mother Absence (p25) Share Above Poverty Line Mean 8th Grade Math Score Fraction Not Suspended from School Share Insured (18-64) Median Rent (2BR) Share White Above Poverty Line Share White Insured (18-64) Share HS Graduate Median Black Home Value Median White Home Value Mean Household Income Share College Grad.
-0.15 -0.1 -0.05 0 0.05 0.1 0.15
Correlation with White Minus Black Rank for Males (p25) 69 Black-White Gap in Individual Income Rank vs. Father Presence
Male Children with Parents at 25th Percentile - Poverty Share Less than 10% 54 White
Black
52
50 48
Diff: 6.1 46
Diff: 9.3
44
Mean Child MeanChild Individual Rank Income
42 40
20 40 60 80
Percentage of Black Children with Father Present 70 Black-White Gap in Employment Rates vs. Father Presence
Male Children with Parents at 25th Percentile - Poverty Share Less than 10%
90
88 86
Diff: 4.0 84
Diff: 7.2
82
Percentage of Children Working Children of Percentage 80 White
Black 78
20 40 60 80
Percentage of Black Children with Father Present 71 Male-Female Gap in Employment Rates vs. Father Presence
Black Children with Parents at 25th Percentile - Poverty Share Less than 10%
95 90
Diff: 7.7 85
Diff: 13.0
80 Percentage of Children Working Children of Working Percentage Black Male
Black Female 75
20 40 60 80
Percentage of Black Children with Father Present 72 Association Between Father Presence and Black Boys' Outcomes: Regression Estimates Dependent Variable: Mean Rank of Black Boys with Parents at 25th Percentile in Tract
Children with Black and White Baseline Two Parents Gender Ratio Father Presence
(1) (2) (3) (4) 0.0492 0.0450 0.0461 Low-Income Black Father Presence (0.0062) (0.0068) (0.0128) Low-Income White Father 0.0077 Presence (0.0076) Low-Income Black Father 0.0387 Presence (0.0043) Low-Income Black Male -0.0011 Adults Per Child (0.0011) Low-Poverty Tracts X X X
Results from OLS regressions at the tract level. Standard errors in parentheses. 73 Father Presence: Additional Results
. Greater presence of white fathers in tract is predictive of white boys’ outcomes
– Phenomenon is not unique to black boys; but rates of father presence are much lower for black boys
. Black father presence in childhood neighborhood is predictive even conditional on tract in which child lives as an adult
– Not a mechanical consequence of black boys and their fathers being subject to the same set of environmental factors (e.g., policing)
74 Racial Bias and Black Children’s Outcomes
. Now turn to another set of factors that are associated with both better outcomes for black boys and smaller black-white gaps: racial bias
. Racial bias measures unavailable at the Census tract level
. Instead focus on two measures available at county and media market level:
1. Implicit racial bias: index based on participants’ ability to match positive and negative words with black vs. white faces [Greenwald et al. 1998]
2. Explicit racial animus: index based on frequency of Google searches for racial epithets [Stephens-Davidowitz 2014]
75 Association Between Racial Bias and Black Boys' Outcomes: Regression Estimates Dependent Variable: Mean Rank of Black Children with Parents at 25th Percentile in Tract
Males White vs. State Fixed Females Males Females Dependent Variable: Baseline Black IAT Effects
(1) (2) (3) (4) (5) (6) Difference in IAT -0.0081 -0.0060 -0.0082 (0.0024) (0.0019) (0.0029) IAT for Whites -0.0080 (0.0023) IAT for Blacks 0.0047 (0.0023) Racial Animus -0.0263 -0.0191 (0.0056) (0.0080) State FE’s X Number of Cells 340 340 340 325 28 27 Number of Observations 492,200 492,200 492,200 491,700 386,600 386,600
76 Columns (1)-(4) are at the county level. Columns (5)-(6) are at the media market level. We restrict to counties (media markets) with poverty rates less than 10%. Variation in the Black-White Earnings Gap Across Tracts
. Four key results:
1. Black boys have lower earnings than white boys in 99% of Census tracts in America, controlling for parental income
2. Both black and white boys have better outcomes in “good” (e.g., low-poverty, higher rent) neighborhoods, but the black-white gap is bigger in such areas
3. Within low-poverty areas, there are two factors associated with better outcomes for black boys and smaller gaps: greater father presence and less racial bias
4. Neighborhoods have causal childhood exposure effects: black boys who move to good areas at a younger age do better
77 Identifying the Causal Effects of Neighborhoods
. Ideal experiment: randomly assign children to neighborhoods and compare outcomes in adulthood, by race
. We approximate this experiment using quasi-experimental design developed by Chetty and Hendren (2018) . Study families who move across areas in observational data . Exploit variation in age of child when family moves to identify causal effects of neighborhoods
. Identifying assumption: potential outcomes of children are orthogonal to age at which family moves to a better/worse neighborhood . Validated by Chetty and Hendren (2018) and Chetty, Hendren, and Katz (2016)
78 Childhood Exposure Effects on Income Rank at Age 30
White Males
0.8 0.6
Slope: -0.026
(0.003) 0.4
δ: 0.242
0.2
Coefficient on Predicted Rank Destination in Predicted on Coefficient 0
5 10 15 20 25 30
Age of Child when Parents Move 79 Childhood Exposure Effects on Income Rank at Age 30
Black Males
0.8
0.6 0.4
Slope: -0.027 (0.004)
δ: 0.119
0.2
Coefficient on Predicted Rank Destination in Predicted on Coefficient 0
5 10 15 20 25 30
Age of Child when Parents Move 80 Race-Specific Childhood Exposure Effects OLS Regression Estimates
Whites Blacks
(1) (2)
Prediction for Whites -0.023 0.003 (0.002) (0.004)
Prediction for Blacks -0.004 -0.029 (0.001) (0.004)
Note: standard errors in parentheses
81 Childhood Exposure Effects on Probability of Being Incarcerated in 2010
Black Males
0.8
0.6 0.4
Slope: -0.033
(0.004) 0.2
δ: 0.055
Coefficient on Incarceration Rate Incarceration on Coefficient 0
5 10 15 20 25 30
Age of Child when Parents Move 82 Summary: Impacts of Neighborhood Environments on Black Men
. Main lesson: childhood environment is an important driver of black-white gaps
. But the environmental factors that matter differ by race – Neighborhood effects cannot be reduced to a common set of factors that affect both black and white boys
. Black boys do well in nbhds. with good resources (low poverty rates) and good race- specific factors (high father presence, less racial bias)
. The problem is that there are essentially no such neighborhoods in America
83 Father Presence and Poverty Rates by Tract for Blacks vs. Whites
80 Low High
Fathers Fathers
70
60
50
40
30
20
10 Share of Children Neighborhood in Children of Share Type
66.3 9.8 19.7 4.2 34.1 62.5 0
High Pov. Low Pov. High Pov. Low Pov. Black White 84 Note: Low-Poverty: Poverty Rate < 10%; High Father Presence: >50% Father Presence Among Children of Own Race Father Presence and Poverty Rates by Tract for Blacks vs. Whites
80 Low High Low High
Fathers Fathers Fathers Fathers
70
60
50
40
30
20
10 Share of Children Neighborhood in Children of Share Type
66.3 9.8 19.7 4.2 1.0 34.1 2.4 62.5 0
High Pov. Low Pov. High Pov. Low Pov. Black White 85 Note: Low-Poverty: Poverty Rate < 10%; High Father Presence: >50% Father Presence Among Children of Own Race Examples of High Upward Mobility Neighborhoods for Low-Income Black Men
New York City, NY
Eastchester / Wakefield Bronx, NYC Queens Village / Laurelton Queens, NYC
Washington, DC
Downtown Silver Spring / Woodside Park / Silver Spring (MD) – Woodside Forest Washington DC CZ New Carrolton / College Park / Prince Georges’ County (MD) - Greenbelt Washington DC CZ
86 Conclusions
1. Mobility into and out of poverty is a central determinant of racial disparities
– Hispanics have relatively high rates of upward mobility increasing income across generations
– Blacks have much lower rates of upward mobility persistent gaps across generations
87 Conclusions
1. Mobility into and out of poverty is a central determinant of racial disparities
2. Commonly proposed policies likely to be insufficient to close black-white gap by themselves
– Changes in transfer programs and minimum wages unlikely to have persistent effects, unless they change rates of mobility [Cameron and Heckman 2001]
– Reducing residential or school segregation can improve the level of outcomes of both black and white children, but may not narrow gaps
88 Conclusions
1. Mobility into and out of poverty is a central determinant of racial disparities
2. Commonly proposed policies likely to be insufficient to close black-white gap by themselves
3. Reducing racial gaps requires policies that cut within neighborhoods and improves environments for specific subgroups, such as black males
– Ex: Mentoring programs, efforts to reduce racial bias, achieving racial integration within schools, criminal justice reform [Heller et al. 2015, Devine et al. 2012]
– Further development and evaluation of such efforts would be valuable
89 Supplementary Figures
90 Density of Parent Household Income Ranks, White and Black Children
White
.02 Black
.015
.01 .005
Mean Parent Rank = 32.7 Mean Parent Rank = 57.9
0 Share of Parents at Given Household Given Percentile Income at Parents of Share 0 20 40 60 80 100
Parent Household Income Rank 91 Density of Parent Household Income Ranks by Race
White Black .02 Asian Hispanic
American Indian
.015
.01
.005
0 Share of Parents at Given Household Given Percentile Income at Parents of Share 0 20 40 60 80 100
Parent Household Income Rank 92 Hours Worked vs. Parent Income Rank
Male Children 50
Diff. at p=75: 8.1
40
30 20
Diff. at p=25: 10.6 10
Weekly Hours Worked in ACS (Age >=(Age 30) ACS in Worked Hours Weekly White
Black 0
0 20 40 60 80 100
Parent Household Income Rank 93 Hours Worked vs. Parent Income Rank
Female Children
50 40
Diff. at p=75: -1.3 30
Diff. at p=25: -1.0
20 10
Weekly Hours Worked in ACS (Age >=(Age 30) ACS in Worked Hours Weekly White
Black 0
0 20 40 60 80 100
Parent Household Income Rank 94 High School Completion Rates vs. Parent Income Rank
Male Children
100 90
Diff. at p=75: 4.2
80 70
Diff. at p=25: 8.3 60
White
Black
Pct. of Children with Children High School of 19) >=Degree (Age Pct. 50
0 20 40 60 80 100
Parent Household Income Rank 95 High School Completion Rates vs. Parent Income Rank
Female Children
100 90
Diff. at p=75: 1.5 80
Diff. at p=25: 3.5
70 60
White
Black
Pct. of Children with Children High School of 19) >=Degree (Age Pct. 50
0 20 40 60 80 100
Parent Household Income Rank 96 College Attendance Rates vs. Parent Income Rank
Male Children
100 80
Diff. at p=75: 7.7 60
Diff. at p=25: 6.5 40
College Attendance Rate for Children (%) Children for Rate College Attendance White
Black 20
0 20 40 60 80 100
Parent Household Income Rank 97 College Attendance Rates vs. Parent Income Rank
Female Children 100
Diff. at p=75: 3.6
80 60
Diff. at p=25: 2.8 40
College Attendance Rate for Children (%) Children for Rate College Attendance White
Black 20
0 20 40 60 80 100
Parent Household Income Rank 98 Household Income Rank vs. Parent Income Rank
Male Children 80
Diff. at p=75: 16.6
60
40 Mean Child MeanChild Household Rank Income
Diff. at p=25: 13.6 White
Black 20
0 20 40 60 80 100
Parent Household Income Rank 99 Household Income Rank vs. Parent Income Rank
Female Children 80
Diff. at p=75: 14.9
60 40
Mean Child MeanChild Household Rank Income Diff. at p=25: 11.7 White
Black 20
0 20 40 60 80 100
Parent Household Income Rank 100 Black-White Gap in Child Individual Income Rank vs. Parent Income Rank
Male Children in Single-Parent Families
80 70
Diff. at p=75: 12.0
60
50 40
Diff. at p=25: 9.7
30 Mean Child MeanChild Individual Rank Income White
Black 20
0 20 40 60 80 100
Parent Household Income Rank 101 Black-White Gap in Child Individual Income Rank vs. Parent Income Rank
Male Children in Two-Parent Families
80 70
Diff. at p=75: 11.5
60
50 40
Diff. at p=25: 7.9
30 Mean Child MeanChild Individual Rank Income White
Black 20
0 20 40 60 80 100
Parent Household Income Rank 102 Black-White Gap in Child Individual Income Rank vs. Parent Income Rank
Male Children, Parents Do Not Own Home
80 70
Diff. at p=75: 11.9
60
50 40
Diff. at p=25: 8.1
30 Mean Child MeanChild Individual Rank Income White
Black 20
0 20 40 60 80 100
Parent Household Income Rank 103 Occupational Distributions Conditional on Parent Income, by Gender Black and White Children, Parents in 3rd Income Decile Male Female
Business Business
STEM STEM
Social Service Social Service
Healthcare Healthcare
Food/Service Food/Service
Administrative Administrative
Farming/Construction Farming/Construction
Maintenance/Repair Maintenance/Repair
Machine Operation Machine Operation
Transportation Transportation
0 10 20 30 0 10 20 30 Pct. of Workforce Pct. of Workforce
Fraction to Be Reallocated = 19.5% Fraction to Be Reallocated = 5.5%
White Black 104 Occupational Distributions Conditional on Parent Income, by Gender Black and White Children, Parents in 8th Income Decile Male Female
Business Business
STEM STEM
Social Service Social Service
Healthcare Healthcare
Food/Service Food/Service
Administrative Administrative
Farming/Construction Farming/Construction
Maintenance/Repair Maintenance/Repair
Machine Operation Machine Operation
Transportation Transportation
0 10 20 30 0 10 20 30 Pct. of Workforce Pct. of Workforce
Fraction to Be Reallocated = 13.2% Fraction to Be Reallocated = 5.4%
White Black 105 Effects of Family-Level Factors on the Black-White Income Gap
Children with Parents at 75th Percentile 20 17.6 Male
Female 15
11.7
11.4 11.4 11.0 10
5 4.8
-0.9 -0.7 -0.7 -1.2
0
Mean Rank of White MeanRank Black Minus White of
5 -
Controls: None Par. Inc. Par Inc. Par Inc. Par Inc. +Two-Par. +Two-Par. +Two-Par. +Educ. +Educ. +Wealth
106
Effects of Family-Level Factors on the Unconditional Black-White Gap 20 17.6 Male Female
15.2 15
13.3 13.1 10
5 4.8
2.4 1.8
1.0
0
Mean Rank of White MeanRank Black Minus White of
5 -
Controls: None Two Parent Parent Parent Education Wealth
107 Mean Child Individual Income Rank by CZ White Females with Parents at 25th Percentile
108 Mean Child Individual Income Rank by CZ Black Females with Parents at 25th Percentile
109 Mean Child Household Income Rank Given Parents at 25th Percentile by CZ Full Population
Mean Household Income Rank
110 Mean Child Household Income Rank Given Parents at 25th Percentile by CZ Whites
Mean Household Income Rank
Correlation with Baseline: 0.77
111 Mean Child Household Income Rank Given Parents at 25th Percentile by CZ Blacks
Mean Household Income Rank
Correlation with Baseline: 0.61
Correlation with White: 0.53 112 Mean Child Individual Income Rank for Males with Parents at 25th Percentile
70
45 Degree Line
60
50
40
30
Mean Rank of Black Males at 25th Percentile 25th at BlackMales of Rank Mean
30 40 50 60 70
Mean Rank of White Males at 25th Percentile 113 Distribution of Black – White Gap in Individual Ranks Across Tracts
Male Children with Parents at 75th Percentile
.08
.06
.04
Density .02 Raw Fraction < 0: 15.2% Signal Fraction < 0: 1.9%
Mean Gap: 9.2 pctiles 0
-40 -20 0 20 40
White Minus Black Rank Given Parents at 75th Percentile 114 Mean Child Individual Income Rank for Males with Parents at 75th Percentile
90
80 45 Degree Line
70
60
50
40
30
Mean Rank of Black Males at 75th Percentile 75th at BlackMales of Rank Mean
20 20 30 40 50 60 70 80 90
Mean Rank of White Males at 75th Percentile 115 Distribution of Black – White Gap in Individual Ranks Across Tracts
Women Children with Parents at 25th Percentile
.08
.06
.04
Density .02 Raw Fraction < 0: 72.5% Signal Fraction < 0: 83.6%
Mean Gap: -3.0 pctiles 0
-40 -20 0 20 40
White Minus Black Rank Given Parents at 25th Percentile 116 Distribution of Black – White Gap in Individual Ranks Across Tracts
Women Children with Parents at 75th Percentile
.08
.06
.04
Density .02 Raw Fraction < 0: 60.8% Signal Fraction < 0: 69.0%
Mean Gap: -2.1 pctiles 0
-40 -20 0 20 40
White Minus Black Rank Given Parents at 75th Percentile 117 Black-White Gap in Incarceration Rate vs. Father Presence
Male Children with Parents at 25th Percentile - Poverty Share Less than 10% 12 White; Slope: -0.01 (0.003)
Black; Slope: -0.06 (0.007)
10 8
Diff: 7.9 6
Diff: 5.2
4
Percentage of Children Incarcerated Children of Percentage 2
20 40 60 80
Percentage of Black Children with Father Present 118 Tract Poverty Rates vs. Mean Child Individual Rank
Black and White Children
20 15 Diff. at p=25: -7.4
Diff. at p=75: -4.8
10 Tract Poverty (%) 2000 in Rates Poverty Tract
White (Intercept: 12.91; Slope: -0.05)
5 Black (Intercept: 21.63; Slope: -0.11) 20 40 60 80 100
Mean Child Individual Income Rank 119 Fraction of Kids with Father Present vs. Individual Income Rank
Black and White Children
70 60
Diff. at p=25: 27.8 Diff. at p=75: 26.9
50 40
White (Intercept: 69.20, Slope: -0.02)
Pct. of Children with Father Present (Parent p=25) (Parent Present with Children Father of Pct. Black (Intercept: 41.02, Slope: -0.00) 30
20 40 60 80 100
Mean Child Individual Income Rank 120 Disruptive Behavior, by Race and Gender
20 White Black
17.8
15 10
6.3
5 4.1
1.8
0 Percent Students with Disruptive Behavior in Gr. 10 Students with Gr. in Disruptive Behavior Percent
Boys Girls
121 Source: National Educational Longitudinal Study 1988 Childhood Exposure Effects on Probability of Being Incarcerated in 2010
White Males
0.8
0.6 0.4
Slope: -0.025
(0.004) 0.2
Coefficient on Incarceration Rate Incarceration on Coefficient δ: 0.094 0
5 10 15 20 25 30
Age of Child when Parents Move 122
Childhood Exposure Effects for Males on Income Rank at Age 24
1 .8
.6 Slope: -0.017 (0.001) Slope: -0.040 (0.002)
.4 Slope: -0.000
(0.009) .2
δ : 0.221
0 Coefficient on Predicted Rank Destination in Predicted on Coefficient
5 10 15 20 25 30
Age of Child when Parents Move 123 Top 5 and Bottom 5 CZs in Upward Mobility for Low-Income Black Men Among 100 Largest CZs by Black Population
Mean Individual Income Rank White Minus Black Commuting Zone Black Males (p=25) Individual Income Rank (p=25)
A. Top 5 CZs
Boston, MA 44.3 7.8 Lafayette, LA 44.0 11.6 Lake Charles, LA 43.1 11.1 Baton Rouge, LA 43.1 10.8 New York, NY 42.4 13.2
B. Bottom 5 CZs
Grand Rapids, MI 35.5 11.1 Cleveland, OH 35.2 12.6 Youngstown, OH 35.2 12.9 Tampa, FL 34.9 9.3 Cincinnati, OH 34.7 10.1 124