Downward Mobility

7/28/2019 Downward Mobility

1/36THE PEW CHARITABLE TRUSTS ECONOMIC MOBILITY PROJECT

Waking Up from theAmerican Dream

By Gregory Acs

Downward Mobilityfrom the Middle Class:


2/36

SEPTEMBER 2011

The Pew Charitable Trusts is driven by the power of knowledge to solve todays mostchallenging problems. Pew applies a rigorous, analytical approach to improve publicpolicy, inform the public and stimulate civic life.

By forging a broad and nonpartisan agreement on the facts, figures and trends relatedto mobility, the Economic Mobility Project is generating an active policy debate abouthow best to improve economic opportunity in the United States and to ensure that the

American Dream is kept alive for generations that follow.

TEAM MEMBERS

Susan K. Urahn, Managing Director, Pew Center on the StatesErin Currier, Project Manager, Economic Mobility ProjectLauren Wechsler, Senior Associate, Economic Mobility ProjectDaniel Colbert, Administrative Assistant, Economic Mobility Project

ACKNOWLEDGEMENTS

Gregory Acs authored this report in his previous role as Senior Fellow at the UrbanInstitute. Dr. Acs is currently Unit Chief for Income Security and Employment at theCongressional Budget Office. Scott Winship conducted extensive analyses for this reportin coordination with Dr. Acs in his previous role as Research Manager at the EconomicMobility Project. Dr. Winship is currently a Fellow at the Brookings Institution.

The Economic Mobility Project thanks all team members, Christopher Jencks, HarryHolzer, Sara McLanahan, Chris Wimer, Ianna Kachoris, Samantha Lasky, Laura Fahey,Kari Miller, Lori Metcalf, Michael Crowley and Cynthia Magnuson for providing valuablefeedback on the report. Design expertise was provided by Willie/Fetchko Graphic Design.Dr. Acs also thanks James Kaminski for his valuable research assistance.

This report is intended for educational and informational purposes.

For additional information on The Pew Charitable Trusts and the Economic MobilityProject, please visit www.economicmobility.org or email us at [email protected].

All Economic Mobility Project (EMP) materials are reviewed by and guided with input from the projects Advisory Board (see inside back cover). The views expressed in this report are those of the authors, and notnecessarily those of the institutions they represent or of EMPs Advisory Board.

September 2011 The Pew Charitable Trusts. All Rights Reserved.

901 E Street NW, 10th Floor 2005 Market Street, Suite 1700 Washington, DC 20004 Philadelphia, PA 19103


3/36

DOWNWARD MOBILITY FROM THE MIDDLE CLASS: WAKING UP FROM THE AMERICAN DREAMiii

Table of Contents

Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Data and Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

Figure 1: Defining the Middle Class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

How Common is Downward Mobility from the Middle? . . . . . . . . . . . . . . . . 7

Figure 2: Chances of Downward Mobility from the Middle . . . . . . . . . . . . . . . . . 7

What Drives Downward Mobility from the Middle? . . . . . . . . . . . . . . . . . . . . 8

Figure 3: Additional Chance of Falling From the Middle to the BottomAssociated with Individual Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Figure 4: Additional Chance of Falling 20 Percentiles Below ParentalRank Associated with Individual Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Figure 5: Additional Chance Real Income is 20 Percent or More BelowParents Income Associated with Individual Characteristics . . . . . . . . . . . . . . . . 10

How Does Downward Mobility from the Middle Differ AcrossDemographic Groups? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Figure 6: Intergenerational Downward Mobility by Race and Gender:Dropping Out of Middle-Class Status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

Figure 7: Intergenerational Downward Mobility by Race and Gender:Income Rank 20+ Percentiles Below Parents . . . . . . . . . . . . . . . . . . . . . . . . . . . 14


4/36

Figure 8: Intergenerational Downward Mobility by Race and Gender:Real Income is 20 Percent or More Below Parents Income . . . . . . . . . . . . . . . . 14

Explaining Racial and Gender Differences in Downward Mobilityfrom the Middle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Table 1: Characteristics of Men and Women Who Started in theMiddle Class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Figure 9: Black Mens Additional Chances of Downward Mobility VersusWhite Mens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Figure 10: White Womens Additional Chances of Downward Mobility Versus White Mens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Endnotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

THE PEW CHARITABLE TRUSTSiv


5/36

DOWNWARD MOBILITY FROM THE MIDDLE CLASS: WAKING UP FROM THE AMERICAN DREAM1

Executive SummaryThe idea that children will grow up to bebetter off than their parents is a centralcomponent of the American Dream, andsustains American optimism. However,Downward Mobility from the Middle Class:

Waking up from the American Dreamfindsthat a middle-class upbringing does notguarantee the same status over the courseof a lifetime.1 A third of Americans raisedin the middle classdefined here as thosebetween the 30th and 70th percentilesof the income distributionfall out of the middle as adults. The data also showdifferences in rates of downward mobility

from the middle based on both familybackground and personal characteristics.

The research for this report wasundertaken to answer critical questionsabout what accounts for downwardmobility from the middle class, and howthose factors influence people differentlydepending on their race and gender. Four

main findings were identified:

Marital status, education, test scoresand drug use have a strong influenceon whether a middle-class child loseseconomic ground as an adult.

In this era of two-worker families, bothmen and women who are divorced,widowed or separated are more likely tolose their middle-class status, as are never-married men and women.

Compared with married women,women who are divorced, widowedor separated are between 31 and 36percentage points more likely to falldown the economic ladder. In turn,never-married women are 16 to 19percentage points more likely to bedownwardly mobile than marriedwomen.

Men who are divorced, widowed orseparated are 13 percentage pointsmore likely to drop out of the middleclass than are married men, and menwho have never married are 6 to 10percentage points more likely to fallthan married men.

Men and women raised in middle-class

homes are generally more likely to fallout of the middle if they do not obtaineducation beyond high school.

Women with a high school diplomaor less who are raised in middle-classhomes are between 14 and 16


6/36

THE PEW CHARITABLE TRUSTS2

EXECUTIVE SUMMARY

percentage points more likely to bedownwardly mobile than womenwho get a college degree.

Men with no more than a high school

diploma are 7 to 15 percentagepoints more likely to be downwardlymobile than men with just somepostsecondary education but nobachelors degree.

A relatively low score on the ArmedForces Qualification Test (AFQT), whichmeasures reading comprehension, mathknowledge, arithmetic reasoning and wordknowledge, correlates with downwardmobility, as does the use of heroin or crackcocaine.

Race is a factor in who falls out of themiddle class, but only for men.

White, black and Hispanic womenare equally likely to experiencedownward mobility out of the middle

class, but 38 percent of black menfall out, compared with 21 percentof white men. Hispanic men alsoappear more likely than white mento fall from the middle as adults,but the difference is not statisticallysignificant.

There is a gender gap in downwardmobility from the middle, but it isdriven entirely by a disparity betweenwhite men and white women.

Only among whites are women moredownwardly mobile than men: Thirtypercent of white women fall out of

the middle class, compared with 21percent of white men. Black womenexperience less downward mobilitythan black men, and Hispanic men

and women have nearly identicalchances of falling from the middle.

Differences in average test scores arethe most important observable racialdifference in accounting for the largedownward mobility gap between blackmen and white men, but none of thefactors examined in the report shedslight on the gap between white menand white women.

Black men raised in middle-classfamilies are 17 percentage pointsmore likely to be downwardlymobile than are white men raised inthe middle. Taking into account arange of personal and backgroundcharacteristicssuch as fathersoccupational status, individualeducational attainment and maritalstatusreduces this gap, but stillleaves a sizable portion unexplained.However, taking into accountdifferences in AFQT scores betweenmiddle-class white and black menreduces the gap until it is statisticallyindistinguishable from zero.

On the other hand, even afteraccounting for personal andbackground differences, the gapbetween white men and whitewomen remains almost completelyunexplained.


7/36


In January 2009, the Economic MobilityProject (EMP) commissioned a publicopinion poll to assess Americansperceptions of their own economicmobility and opportunity and the

mobility prospects of future generations. When asked to define the AmericanDream, one of the more popular optionschosen was your children beingfinancially better off than you. 2 Indeed,the promise of each generation doingbetter than the one that came before it isa founding principle of our country andsustains American optimism.

Defining middle class as those betweenthe 30th and 70th percentiles of theincome distribution, this report finds

Introductionthat a third of Americans raised in themiddle class fall down the income ladderas adults. 3 Whats more, the data showdifferences in rates of downward mobilitybased on both family background and

personal characteristics.

What accounts for downward mobilityfrom the middle class, and how dothose factors influence people differentlydepending on their race and gender?

What might explain differences indownward mobility by race and gender?Using survey data that tracks Americans

from youth into adulthood, this reportexplores these questions to betterunderstand why the American Dreameludes some in the middle class.


8/36

percentiles. Using the 30 th percentileas the lower bound for middle-incomestatus has some appealit is around theincome level at which most individualsare ineligible for public-assistance

programs for low-income families and,as such, the life experiences of those justabove and just below the cutoff may bequalitatively different.

A familys income is adjusted for familysize by dividing it by the poverty line forthe family. The poverty line measures afamilys needs; it varies by family size and

composition. In these data, those raisedin middle-class families had income-to-needs ratios from 1.70 to 3.37 (the30 th to 70 th percentiles). This definitionof middle class roughly translates toincome from about $32,900 to $64,000in 2010 dollars for a family with twoadults and two children (see Figure 1).

This report draws from the NationalLongitudinal Survey of Youth (NLSY)1979 cohort, focusing on youth whowere age 14-17 in 1979 and who livedin their parents homes in 1979 and

1980.4

Their economic status was thenassessed in 2004 and 2006, when theywere between the ages of 39 and 44. 5 Thesample is divided into three racial groups:Non-Hispanic whites (including a smallnumber of people who are not white,black or Hispanic); non-Hispanic blacks;and Hispanics.

The middle-class group is defined asthose falling between the 30 th and 70 th percentiles of the family-size-adjustedincome distribution. This is a broaderdefinition of middle class than thatused in EMPs 2008 report, Getting Aheador Losing Ground: Economic Mobility in America, which defined middle incomeas falling between the 40 th and 60 th

Data and Denitions



9/36


to be downwardly mobile. On the otherhand, it does consider someone who dropsfrom the 69 th to the 49 th percentile to bedownwardly mobile, even though theperson in this example remains squarelyin the middle of the income distribution asan adult.

Both of these definitions are based onrelative mobilitywhere one standsrelative to ones peers in the incomedistribution, compared with where onesparents stood relative to their peers.However, since living standards haveimproved over time due to economicgrowth, even a youth who is worse off

Def ining the M iddle Cla ssThe Income Range of the M iddle Class has Sh ifted S ign if icantly Over the Past Generat ion

Figure 1

020,000

40,000

60,000

80,000

100,000

120,000

140,000

160,000

180,000

$200,000

Parents Generation Childrens Generation

$64,000AND ABOVE

MIDDLE CLASS

LESS THAN$32,900

$110,600AND ABOVE

MIDDLE CLASS

LESS THAN$53,900

NOTE: Income calculated for fa mi ly of four . Cut po ints def ining the middle class are based on income adjusted for fa mily size. A fami lys incom e is d ivided by the federal poverty l ine correspond ing to its s ize and co m pos ition (e .g ., number of adults andnumber of ch ildren). The dollar a mounts in the f igure are the result of multiplying the s ize-adjusted cut po ints by the federalpoverty l ine for a fa mi ly of four (two children and two adults) . The cut po int am ounts for each generat ion (m easured in 1979and in 2004 and 2006) are then adjusted to reflect purchas ing power in 2010, us ing the Personal Consu mpt ion Expend ituredeflator fro m the Bureau of Econo mic Analysis.

Measures of MobilityThree measures are used to assess thedownward mobility of youth raised inmiddle-class families. The first definesdownward mobility as the share of middle-class youth who fall below the30th percentile of the income distributionwhen they are 39- to 44-year-olds. 6 Thesecond measure considers adults to bedownwardly mobile if their income rank is20 or more percentiles below their parentsrank in 1978-79. Unlike the first measure,this measure does not consider someonewho drops from the 31 st percentile as achild to the 29 th percentile as an adult

DATA AND DEFIN ITIONS


10/36


size-adjusted income is more than 20percent below their parents income.These three measures are interrelatedbut capture different facets of mobility,

and considering all three gives a morecomplete picture of who is downwardlymobile from the middle and why.

DATA AND DEFIN ITIONS

relative to his peers than his parentswere might be materially better off than his parents. To address this issue,the analyses use a third measure that

assesses downward absolute mobility:adults who started out in the middleclass are considered downwardlymobile if their inflation- and family-


11/36


of their parents at a similar agethe thirdmeasure of downward mobility.

While the mobility findings do reflectwhat one would expect to see in terms

of mathematical chances of leaving themiddle class, not all middle-class childrenare equally likely to fall. As the remainderof this report shows, there are notabledifferences in downward mobility basedon both family background and individualchoices.

If growing up in the middle class did notconfer any advantages to children, thenone would expect that 30 percent of themwould end up in the bottom 30 percentof families as adults. That is to say, no

matter what their parents income was,everyonerich, poor or middle classwould have the same chance of being inthe bottom 30 percent. Thirty percent of adults raised in rich families would endup there, 30 percent of adults raised inpoor families, and 30 percent of thoseraised in middle-class families.

In fact, at least for adults raised inmiddle-class families, that is basicallywhat happened28 percent of adultswhose parents were in the middle classfell from the middle themselves. Manydid not fall far, however (remember thateven moving from the 31 st percentileto the 29 th is enough). Looking at thesecond measure of downward mobility, 28

percent of adults also fell 20 percentilesor more below their parents rank (thoughof course, they were not exactly the same28 percent who moved down by the firstmeasure). Finally, 19 percent of adults hadincome at least 20 percent lower than that

How Common is DownwardMobility from the Middle?

Chances of Downward Mob ility from the M iddle

Figure 2

0

5

10

15

20

25

30

Percent fallingout of middle-

class status(below 30thpercentile)

Percent falling20 or morepercentiles

below parentsrank

Percent withreal income

20% or morebelow parents

income

28 28

19


12/36


What Drives Downward Mobilityfrom the Middle?measures reading comprehension, mathknowledge, arithmetic reasoning andword knowledge. Studies have shownthat AFQT scores correlate well withperformance in the military, as well

as with adult wages. It is important tonote that beyond measuring a personshuman capital in the form of knowledgeor cognitive skills, AFQT scores likelyreflect a host of other factors that affecttest performance not included here, suchas motivation and self-confidence, whichcould also influence downward mobility. 10 Furthermore, AFQT scores are likely to be

influenced by a range of outside factors,such as school quality, which are notaccounted for in the models here becausemeasures are unavailable.

Figures 3 to 5 illustrate how theprobability of downward mobility isconnected to family background, choicecharacteristics and AFQT scores. 11

Because these factors may affect men andwomenand their mobilitydifferently,men and women are examined separately.Each chart shows the predicted likelihoodof downward mobility associated withhaving one characteristic rather thananotherafter statistically accounting for

This section investigates how a range of factors including several mobility driversexamined in past EMP research, suchas parental income and education andindividual education and family structure,

affect downward mobility from themiddle. 7 The first set of factors consistsof family background characteristics orthose characteristics that do not reflectthe youths own choices. These includewhether ones mother has a high schooldiploma, whether ones father works ina professional or managerial occupationand the income percentile rank into

which ones family falls.8

The secondset of characteristics includes choicecharacteristics, such as youths educationalattainment and marital status, which inpart reflect their own preferences as theymove into adulthood. 9 To explore theimpact of potentially destructive decisions,drug use is included.

Additionally, youths percentile scoreson the Armed Forces Qualification Test(AFQT) are included. Everyone in theNLSY-79 survey was asked to take the

AFQT, a standardized test administered bythe U.S. military to determine qualificationfor enlistment in the armed forces. It


13/36


Additional Chance of Fall ing Fro m the M iddle to the Botto m Associatedwith Ind ividual Character istics

Figure 3

Black(vs. white)

Motherwithout

highschooldiploma

Fathernot in

profes-sionaloccupation

1979income 20

percentileslower

H.S.or less

(vs. somecollege)

H.S.or less

(vs. collegegrad)

Has nevermarried

Is divorced Has usedcrack

Has usedheroin

AFQT 20percentiles

lower

-10

0

10

20

30

40

-20

Men Women

6.0

-2.7

3.0 4.46.9

1.3 1.33.7

13.19.5

7.5

16.3

10.0

17.613.0

35.8

14.5 12.8

6.1 6.4

26.2

-15.9 A d d i t i o n a l

P r o

b a

b i l i t y o

f D o w n w a r

d M o

b i l i t y

( P e r c e n

t a g e

P o

i n t s )

Additional Chance of Fall ing 20 Percent iles Below Parental Rank Assoc iatedwith Ind ividual Character istics

Figure 4

NOTE FOR FIGURES 3 AND 4: Dark bars ind icate the effect is stat istically sign if icant at p


14/36


WHAT DRIVES DOWNWARD MOBILITY FROM THE MIDDLE?

Compared with married men, men whoare divorced, widowed or separated are 13percentage points more likely to drop outof middle-class status, 8 percentage pointsmore likely to drop at least 20 percentilesbelow their parents economic positionand 9 percentage points more likely tohave income that is at least 20 percentbelow the level of their parents. 13 In part,this likely reflects the loss of a secondincome that often accompanies divorceor separation, though it is important toremember that incomes are adjusted forfamily size in these analyses.

More strikingly, compared with marriedwomen, women who are divorced,widowed or separated are 31 to 36

the impact of all the other factors. Manyfactors are strongly related to one another,making it difficult to tease out theirindividual importance. However, familystructure, education and AFQT scores areconsistently associated with downwardmobility for both men and women,whereas drug use increases downwardmobility among men. 12

In general, being divorcedand never having married areassociated with downwardmobility. The associationsfor women are stronger thanfor men, especially for beingdivorced.

Additional Chance Real Inco me is 20 Percent or More Below Parents Income Assoc iated w ith Ind ividual Character istics

Figure 5

NOTE: Dark bars ind icate the effect is stat istically sign if icant at p


15/36


having no more than a highschool education is stronglycorrelated with downward

mobility.Compared with those with no more thana high school education, women whoattended or completed college are 9 and16 percentage points, respectively, lesslikely to drop out of the middle class asadults. The gaps are the same9 and16 percentage pointswhen looking

at the likelihood of income falling 20percent below ones parents. Womenwith college degrees are 14 percentagepoints less likely to fall 20 percentilesor more below their parents levels thanthose with schooling at the high schoollevel or below. Among men, having acollege degree appears less important fordownward mobility than it is for women,

and having attended college at all is moreimportant.

Among men who were in themiddle class as youth, usingcrack cocaine or heroin isassociated with dramaticincreases in downward

mobility.Unsurprisingly, these analyses suggestthat one of the worst choices a middle-class male youth can make with respect tofuture mobility is to use hard drugs. Maleyouth who have tried crack cocaine are

percentage points more likely to bedownwardly mobile, depending on themeasure. Never-married women are 16to 19 percentage points more likely to

be downwardly mobile than marriedwomenalso a stronger association thanexists for men, who are 6 to 10 percentagepoints more likely to fall if they havenever married. 14 These gender differenceslikely are related to lower earnings amongwomen than men.

Middle-class youth with

lower AFQT percentilescores are significantly morelikely to be downwardlymobile than those withhigher scores.

This finding holds true across all threemeasures of mobility for both women and

men. Scoring 20 percentiles lower on the AFQT increases the chances that a manor woman raised in a middle-class familywill fall to lower-income status as an adultby about 6 percentage points. Similarly,a 20-percentile decline in AFQT scoresis associated with a 5 percentage-pointrise in the chance of falling 20 or morepercentiles below ones parents income

rank, and a 6 percentage-point rise in thechance that income will be 20 percent ormore below that of ones parents.

Among men and womenraised in middle-class homes,



16/36


impact on women are never statisticallydifferent from zerothey are impreciselymeasured, probably because of the smallnumber of female drug users who start

out in the middle class.16

The next section explores racial andgender mobility gaps, but because of theimprecision of the estimates, it mostlyconsiders how gaps are affected by groupsof factors, such as family background andchoice factors.

10 to 18 percentage points more likelyto experience downward mobility thanthose who have not, and the figures forheroin are 19 to 26 percentage points. 15

Among women, the effects of hard druguse on downward mobility are harder topin down. Figures 3 to 5 appear to showthat heroin use among women actuallyprotects against downward mobility,but that would be an inappropriateconclusion. The estimates for drugs



17/36


African Americansexperience more downwardmobility than whites.Figure 6 shows the likelihood that a

person raised in a middle-class family willfall into a lower-income category as anadult. The leftmost set of bars indicate thatblacks raised in middle-class families aresignificantly more likely (37 percent) thannon-Hispanic whites (25 percent) to dropbelow the 30 th income percentile as adults.Figure 7, showing the share of youthfalling 20 percentiles or more below their

parents income percentile, and Figure 8,showing the percent with income thatis 20 percent or more below the level of their parents, reveal the same pattern.Differences between non-Hispanic whitesand Hispanics are comparatively smalland not statistically significant. Althoughdifferences between blacks and Hispanicsappear bigger, these also fail to achieve

statistical significance.

The previous section demonstratedthe effects of various personal andbackground characteristics on downwardmobility from the middle for men andwomen separately. However, because

whites, blacks and Hispanics often lookdifferent along these and a number of other dimensions, as do men and women,they have different rates of downwardmobility, which is not conveyed byFigures 3 to 5. This section, therefore,turns to the question of downwardmobility differences by race and sex.

Across the three mobility measures,

three main findings are apparent. First,consistent with previous EMP research, African Americans experience significantlymore downward mobility than whites,regardless of the measure used. Second,the differences in downward mobility byrace are limited to differences betweenwhite and black men. Finally, theonly notable gender gap in downward

economic mobility is among whites.

How Does Downward Mobility

from the Middle Differ AcrossDemographic Groups?


18/36


HOW DOES DOWNWARD MOBILITY FROM THE MIDDLE DIFFER ACROSS DEMOGRAPHIC GROUPS?

Intergenerat ional Downward Mob ility by Race and Gender Dropp ing Out of M iddle-Class Status (Below 30th Percent ile)

Figure 6

05

1015202530354045

50 PERCENT

All Men Women

Note for F igures 6 to 8 : Tabulat ions from the NLSY-79. Sample co mpr ises ind ividuals who l ived in middle-class fa milies in1979-1980 who were between the ages of 14 and 17 in 1979 and 39-44 between 2004 and 2006 . Middle class is def ined asfami ly income-to-needs rat ios between the 30th and 70th percent iles of the income d istr ibut ion .

* Indicates that the d ifference co mpared to wh ites is stat istically sign if icant *p


19/36


Differences in downwardmobility by race are limitedto differences between

white and black men;women of different races areequally likely to experiencedownward mobility.The share of black men who fall out of middle-class status is nearly twice as highas the share of white men who do so (38percent versus 21 percent). Hispanic men

also appear more likely than white men todrop out of the middle as adults, but thedifference is not statistically significant.

Across the three measures of mobility,white, black and Hispanic womenexperience similar rates of downwardmobility from the middle, and thedifferences among them are not statisticallysignificant.

The gender gap indownward mobility fromthe middle is only present

among whites.Only among whites are women moredownwardly mobile than men: Thirtypercent versus 21 percent in Figure 6, anda gap equally sizable in Figures 7 and 8,all of which are statistically significant.In fact, black women consistently showless downward mobility than black men,

although the difference is never statisticallymeaningful. Hispanic men and womenhave nearly identical chances of fallingfrom the middle.

HOW DOES DOWNWARD MOBILITY FROM THE MIDDLE DIFFER ACROSS DEMOGRAPHIC GROUPS?


20/36


in backgrounds, characteristics and testscores investigated here.

The first step to explaining differencesin dropping from the middle by race

and gender is to see how the familyand personal characteristics of men andwomen of different races vary (see Table1). These comparisons suggest that, onnet, white men and women raised inmiddle-class families are advantagedrelative to their black and Hispaniccounterparts. The differences betweenwhite and black men, especially with

regard to their educational attainment,their likelihood of being married, theirtest scores and their fathers occupation,are particularly large. Differences betweenwhite and black women also are large,except that they have similar educationlevels. Hispanics tend to fall betweenwhites and blacks for both sexes, with theexceptions of mothers education and their

own education.

It is important to note that a factor canbe important for explaining downwardmobility in general without it beingimportant for explaining the black-whitegap among men or the male-female gap

among whites. For instance, having high AFQT scores might promote mobilityfor both men and women, but if menand women have similar test scores, thenit will not explain mobility differencesbetween them. The analyses in this sectionassess the extent to which differencesin downward mobility by race andgender are accounted for by observable

differences in individuals backgroundsand characteristics. Among men raisedin middle-income families, the gap indownward mobility between whites andblacks is substantially reduced whenthose observable differences are taken intoaccount, with differences in test scoresplaying a prominent role. Among whitesraised in middle-income families, the

male-female gap in downward mobilitycannot be accounted for by differences

Explaining Racial and Gender

Differences in Downward Mobilityfrom the Middle


21/36


Parent CharacteristicsMother has at least a high school diploma 68% 64% 42% 67% 59% 44%

Fathers occupation is professional/managerial 21% 3% 16% 23% 5% 9%

Average family rank in income distribution: 1979 50.5 47.7 50.2 50.9 46.5 49.1

Individual CharacteristicsHas high school diploma 93% 95% 90% 96% 98% 91%

Has a college degree 27% 15% 17% 24% 22% 18%

Married 65% 45% 63% 67% 48% 62%

Has never married 13% 35% 19% 6% 23% 13%

Is divorced/widowed/separated 23% 21% 18% 27% 29% 25%

10+ lifetime uses of marijuana 42% 40% 45% 27% 12% 15%Used cocaine 32% 29% 29% 22% 9% 16%

Used crack 7% 12% 7% 7% 6% 4%

Used heroin 2% 4% 4% 1% 2% 0%

Average AFQT percentile 46.4 21.7 36.3 45.5 27.5 32.8

NOTE: Tabulations from the NLSY-79. Sample comprises people who lived in middle-class families in 1979-1980 who werebetween the ages of 14 and 17 in 1979 and 39-44 between 2004 and 2006. Middle class is defined as family income-to-needs ratios between the 30th and 70th percentiles of the income distribution. Sample sizes: 386 white men, 113 black men,89 Hispanic men, 366 white women, 123 black women, 112 Hispanic women.

Hispanic women relative to men seemless impressive than their advantages onthe characteristics here would predict.The only real disadvantage womenstarting in the middle seem to haverelative to men is that they are morelikely to be divorced. 17

By using statistical models that estimatethe effect of each factor on downwardmobility differences, holding the otherfactors constant, it is possible to exploremore deeply the sources of groupdifferences in downward mobility.

These differences in family andpersonal characteristics by race andsex are consistent with the findingson downward mobility: The greatestdifferences in these characteristics arebetween white and black men, and this isthe only pair for whom intergenerational

downward mobility rates differsignificantly. Based on the characteristicshere, downward mobility differencesbetween white men and women areunexplained (because they look so similaracross the range of factors considered),and the mobility outcomes of black and

Characteristics of Men and Women Who Started in the Middle Class

MEN WOMENWHITE BLACK HISPANIC WHITE BLA CK HISPANIC

Table 1

EXPLAINING RACIAL AND GENDER DIFFERENCES IN DOWNWARD MOBILITY FROM THE MIDDLE


22/36


characteristicson which black and whitemen might differ. Those differences mightin turn explain the average downwardmobility gap between the two groups.Investigating one set of factors at a timeis a way to see whether black and whitemen who are identical in terms of all thefactors considered still differ in terms of their downward mobility. If adding a set of factors to the analysis reduces the black-white mobility gap, then differences inthose factors help explain what is drivingdisparities in downward mobility from themiddle. 18

Differences in averagetest scores are the mostimportant observable racialdifference in accounting for the downward mobility gapbetween black and whitemen.Figure 9 shows how the gap indownward mobility between black andwhite men changes when taking intoaccount the factors explored abovefamily background and individual


Figure 9

Black Men s Add itional Chance s of Downward Mob ility Ver s us White Men s

0 5 10 15 20

Out of middle-class status(below 30th percentile)

Rank is 20 or more percentilesbelow parents rank

Real income is 20% or morebelow parents income

15.2%***13.9%***

10.9%**

7.7%

15.3%***

15.4%***

10.9%**

6.9%

17.3%***15.9%***

11.3%**

6.5%

49 percent

55 percent

62 percent

Race alone Race and Family BackgroundRace, Family Background and Choices Race, Family Background, Choices and Test Scores

After controlling for a varietyof background and choice

characteristics, the black-white

mobility gap is reduced by:

Note : Tabulat ions from the NLSY-79. Sample co mpr ises ind ividuals who l ived in middle-class fa mi lies in 1979-1980 who werebetween the ages of 14 and 17 in 1979 and 39-44 between 2004 and 2006 . Middle class is def ined as fa mi ly income-to-needsratios between the 30th and 70th percent iles of the incom e d istr ibut ion .

* Indicates that the d ifference co mpared to wh ites is stat istically sign if icant - * p < 0 .10, ** p < 0 .05, *** p < 0.01 .


23/36



The steadily shrinking bars in each panelof Figure 9 indicate that after taking

AFQT scores into account, along with thefamily background and choice factors,the black-white downward mobilitygap among men is smaller by 49 to 62percent, depending on the mobilitymeasure. The fact that the gaps are nolonger statistically distinguishable fromzero means that these factors may explainthe entire gap. And AFQT scores are thesingle biggest predictor of black-whitedifferences in downward mobility from

the middle class among these factors.19

Differences in downwardmobility between whitemen and women are mostlyunrelated to the factorsconsidered in this report.

Figure 10 shows how the downwardmobility gaps between white men andwomen change as different sets of factorsare statistically controlled. In starkcontrast to the black-white male gapsin Figure 9, the white male-female gapsin Figure 10 do not decline by much,even when the full set of factors includedin this report are controlled, and they

remain statistically significant.20

Given that the factors considered canaccount for a substantial portion of theracial gaps among men, it is striking howunrelated white gender gaps are to the

As illustrated in Figure 9, the findingsare consistent across the three measuresof downward mobility: Even aftercontrolling for differences in familybackgrounds, locations and choices,black men raised in middle-class familiesstill are significantly more likely to bedownwardly mobile than white menraised in middle-class families. Thetop-most bar in Figure 9, for example,shows the black-white difference in theprobability that a man who was in themiddle class as a youth falls out of themiddle in adulthood. Before takingother characteristics into account, blackmen raised in middle-class familiesare 17 percentage points more likelyto be downwardly mobile than whitemen (the gap shown in Figure 6). Thesecond bar in the top-most set showswhat happens to the gap when controlsfor family background characteristicsare added to the analysis. After takingdifferences in family background intoaccount, the black-white difference fallsto 16 percentage points. Accounting forchoices reduces the gap to 11 percentagepoints. In other words, all of these factorscombined provide part of the explanationfor the different rates of downwardmobility between black and white men,but a sizable unexplained gap remains.

Adding AFQT scores to the model,however, reduces black-white differencesin downward mobility so that the tworates are statistically indistinguishable.


24/36


might account for much of the downwardmobility gaps between white men andwhite women. Whatever is behind thegender gap among whites, it is likely to bea different set of factors than those behindthe racial gap among men.

White Wo men s Add itional Chances of Downward Mob ility VersusWhite Men s

Figure 10

Note : Tabulat ions from the NLSY-79. Sample co mpr ises ind ividuals who l ived in middle-class fa mi lies in 1979-1980 who werebetween the ages of 14 and 17 in 1979 and 39-44 between 2004 and 2006 . Middle class is def ined as fa mi ly income-to-needsratios between the 30th and 70th percent iles of the incom e d istr ibut ion .

* Indicates that the d ifference co mpared to wh ites is stat istically sign if icant - * p < 0 .10, ** p < 0 .05, *** p < 0.01 .

Out of middle-class status(below 30th percentile)

Rank is 20 or more percentilesbelow parents rank

Real income Is 20% or morebelow parents' income

0

10

15

5

20 PERCENT

9.2% ***9.2%*** 8.6%*** 8.3% *** 8.8%***9.4% ***

8.0%*** 7.8% ***6.5%***7.1%

***

6.0% *** 5.9% ***

Gender alone

Gender, Family Background and Choices

Gender and Family Background

Gender, Family Background, Choices and Test Scores

Controlling for:


diverse factors considered. A reasonablespeculation is that differences in labormarket factors (occupations chosen, hoursworked, years of experience at a givenage and hourly pay for a given job) andin family structure (single parenthood)


25/36


adulthood, double the number of whitemen who do so. In contrast, there is nota notable gap in downward mobilitybetween white and Hispanic men, norbetween women of different races.

The findings presented here raise severalimportant questions for future research.Because AFQT scores explain a largeportion of the black-white difference indownward intergenerational mobility, itis important to better understand why

AFQT scores differ so substantially,even among youth raised in middle-

class families. The scope of factors that AFQT could be reflecting, beyond apersons human capital potential oracademic ability, also needs to be betterunderstood. Further, future researchshould explore why racial and ethnicdifferences in downward intergenerationalmobility are confined to men and notwomen.

Finally, this report leaves open thequestion of why white women are moredownwardly mobile than white men.

A key element of the American Dream isthat each generation will exceed the livingstandards and economic position of theone that came before it. At the very least,parentsespecially in the middle class

want to ensure that their own economicposition will transfer to their children.Nonetheless, consistent with previouswork by EMP, this report demonstratesthat about one quarter of children raisedin middle-class families are downwardlymobile as adults, a finding that persistsacross three definitions of downwardmobility. The educational attainment,

family structure and test scores of menand women appear to be strong drivers of downward economic mobility for initiallymiddle-class Americans.

However, a more notable and troublingfinding from this and other research isthe stark contrast in downward mobilityrates between whites and blacks, and

in particular between white and blackmen. This report shows that nearly 40percent of black men raised in middle-class families fall from the middle in

Conclusion


26/36


CONCLUSION

gender-specific factors are important formobility. Even when confining analysesto black and white men, up to half of thedownward mobility gap is unaccounted

for by the factors considered here. Thesefindings highlight the importance of continued research into the drivers of downward mobility, how they mightdiffer across groups and how they mightvary from the drivers of upward mobility.

The striking lack of correspondencebetween the factors affecting racialgaps in mobility among men and thefactors affecting gender gaps among

whites highlights the complicatednature of economic mobility. The factthat black and Hispanic women alsoappear advantaged relative to theirmale counterparts, yet do not have lessdownward mobility, also suggests that


27/36


Appendix

MEN

Black 0.059809 0.062674 0.074435(.0544132) (.0526263) (.0500612)

Hispanic 0.044624 0.048453 0.032809(.0537959) (.0531297) (.0480597)

Mother Graduated High School -0.0299 -0.05833 -0.04565(.0451668) (.0456984) (.0400189)

Mother without High School Diploma -0.0751 -0.12599 -0.05345(.0878942) (.0814461) (.0754947)

Father Professional/Manager -0.06873 -0.06071 -0.00816

(.0421091) (.0438483) (.0386112)Father not in Professional Occupation -0.02842 -0.03764 -0.01793

(.0493095) (.0453886) (.0413799)

1979 Family Income Rank -0.00066 0.007614 0.002942(.0014985) (.0014479) (.00128)

Lived in Northeastern City -0.09145 -0.08432 0.001673(.1036279) (.089274) (.089885)

North Central City -0.02504 -0.05115 0.035946(.0890715) (.0840082) (.0788161)

North Central, Not in City 0.018907 -0.00611 0.016841(.0589251) (.060387) (.0492161)

Southern City 0.021115 0.016758 -0.01443(.1240681) (.1004041) (.0920716)

South, Not in City 0.000882 0.022583 0.049417(.0614843) (.0640848) (.0535433)

Western City -0.00951 -0.03782 -0.1106(.0967347) (.096651) (.0550827)

West, Not in City 0.038201 -0.02295 0.018956(.066724) (.0662019) (.0555794)

Modeling Results for Figures 3 to 5 (Explaining the Causesof Individual Downward Mobility)OLS Regression Coefficients, with Standard Errors in Parentheses

Out of middle-classstatus (below 30th

percentile)

Rank is 20 or morepercentiles below

parents rank

Real income is 20%or more below

parents income


28/36


APPENDIX

Black -0.02727 0.00519 -0.01004(.0542782) (.0528899) (.049182)

Hispanic -0.06009 -0.02238 -0.03045(.0535681) (.0543364) (.049527)

Mother Graduated High School -0.04422 -0.0131 -0.01244(.0429833) (.0420451) (.0390143)

Mother without High School Diploma 0.015582 0.09067 0.106586(.1224497) (.1364204) (.1152161)

Father Professional/Manager -0.01253 -0.03222 -0.02376(.049132) (.047679) (.0418255)

Father not in Professional Occupation -0.02573 -0.00275 -0.04103(.0607058) (.054946) (.0499059)

WOMEN


percentile)


parents rank


parents income

Region or City Missing -0.1355 -0.27603 -0.17912(.116787) (.0673124) (.0593596)

Some College -0.13059 -0.15454 -0.07394

(.0422372) (.0429476) (.035866)College Graduate -0.07472 -0.13204 -0.0501

(.0453155) (.0459726) (.035381)

Never Married 0.099885 0.095535 0.060695(.0484097) (.0490901) (.040473)

Divorced, Widowed, Sep. 0.129583 0.084559 0.094421(.0491784) (.0467579) (.0433222)

Used Mari juana 10+ Times -0.01031 -0.02079 0.026179(.0450417) (.0431814) (.0403333)

Used Crack Cocaine 0.144844 0.18075 0.106491(.0851293) (.0824937) (.0812122)

Used Powder Cocaine 0.069461 0.040906 0.047555(.0518451) (.0493978) (.046725)

Used Heroin 0.262137 0.1932 0.194766(.1122981) (.1268499) (.1423163)

Drug Use Missing 0.224225 0.07533 0.172945(.0843952) (.0793352) (.080532)

AFQT Percentile Score -0.00306 -0.00248 -0.00189(.0007888) (.0007831) (.0006588)

AFQT Score Missing 0.237882 0.164023 0.160922(.1653492) (.1324256) (.1408019)

Constant 0.3739 0.019583 0.047166

(.1064742) (.0994021) (.0873717)


percentile)


parents rank


parents income


29/36


APPENDIX

1979 Family Income Rank -0.00185 0.010339 0.004266(.0015568) (.001467) (.0013474)

Lived in Northeastern City 0.087857 0.082886 0.055236(.1722767) (.1603668) (.1523797)

North Central City -0.12664 -0.01691 0.013123(.1107818) (.1084472) (.1090926)

North Central, Not in City 0.005828 0.061238 0.00255(.0638172) (.0605467) (.0558175)

Southern City -0.04909 0.033201 -0.07437(.0863909) (.0936863) (.0773622)

South, Not in City 0.015218 0.054196 0.030711(.0665053) (.061652) (.0585146)

Western City -0.04404 -0.06957 -0.01571(.1292623) (.136347) (.1340688)

West, Not in City -0.03904 -0.03133 -0.01412(.0790885) (.0704638) (.0667849)

Region or City Missing -0.00306 0.113051 -0.19495(.1732173) (.1449294) (.1241349)

Some College -0.0946 -0.03057 -0.08569(.0482181) (.0472551) (.0432292)

College Graduate -0.16338 -0.13748 -0.16134(.0480604) (.0484432) (.0391828)

Never Married 0.175752 0.164823 0.185539(.0672245) (.0676648) (.0647703)

Divorced, Widowed, Sep. 0.357515 0.356422 0.305625

(.0487021) (.0475358) (.0462921)Used Mari juana 10+ Times -0.0439 -0.0151 -0.03489

(.0560645) (.0536151) (.0503491)

Used Crack Cocaine 0.128372 0.123807 -0.00413(.0886663) (.0931046) (.0821524)

Used Powder Cocaine 0.083478 0.067403 0.058228(.0631894) (.0629498) (.0563851)

Used Heroin -0.15941 -0.11473 -0.12654(.2341077) (.1669656) (.217134)

Drug Use Missing 0.072953 0.115473 0.156656(.0984604) (.1065179) (.0993843)

AFQT Percentile Score -0.0032 -0.00234 -0.00211(.0009806) (.0009827) (.0008756)

AFQT Score Missing 0.071975 -0.15629 -0.03984(.1164945) (.0655385) (.0607658)

Constant 0.517406 -0.224 0.052567(.1177215) (.0993324) (.0911937)


percentile)


parents rank


parents income


30/36


in America. Washington, DC: Economic MobilityProject, The Pew Charitable Trusts. http://www.economicmobility.org/assets/pdfs/PEW_EMP_GETTING_AHEAD_FULL.pdf.

Mazumder, Bhashkar. 2008. UpwardIntergenerational Economic Mobility in the United

States. Washington, DC: Economic MobilityProject, The Pew Charitable Trusts. http://www.economicmobility.org/assets/pdfs/PEW_EMP_UPWARD_INTERGENERATIONAL.pdf

McLanahan, Sara and Gary Sandefur. 1994. GrowingUp With a Single Parent: What Hurts, What Helps. Cambridge, MA: Harvard University Press.

Neal, Derek and William Johnson. 1996. The Role of Premarket Factors in Black-White Wage Differences.University of Chicago Press. Journal of PoliticalEconomy 104(55): 860-895.

Steele, Claude M. 2010. Whistling Vivaldi: And Other Clues to How Stereotypes Affect Us.New York, NY:

W.W. Norton and Co.

Economic Mobility Project. 2009. Findings froma National Survey and Focus Groups on EconomicMobility. Washington, DC: The Pew CharitableTrusts. http://www.economicmobility.org/poll2009.

Fischer, Claude S., Michael Hout, Martn Snchez Jankowski, Samuel R. Lucas, Ann Swidler, and Kim

Voss. 1996. Inequality by Design: Cracking the BellCurve Myth.Princeton, NJ: Princeton University Press.

Haskins, Ron, Harry Holzer, and Robert Lerman.2009. Promoting Economic Mobility by IncreasingPostsecondary Education. Washington, DC:Economic Mobility Project, The Pew CharitableTrusts. http://www.economicmobility.org/assets/pdfs/ PEW_EMP_POSTSECONDARY_ED.pdf.

Hertz, Tom. 2006. Understanding Mobility in America. Washington, DC: Center for AmericanProgress. http://www.americanprogress.org/kf/hertz_mobility_analysis.pdf

Isaacs, Julia, Isabel Sawhill, and Ron Haskins. 2008.Getting Ahead or Losing Ground: Economic Mobility

References


31/36


6 Again, this varies from Isaacs, Sawhill, and Haskins,2008, which considers youth who drop into thebottom income quintile (below the 20 th percentile) tobe downwardly mobile.

7 See Isaacs, Sawhill and Haskins, 2008; Haskins,Holzer and Lerman, 2009; and DeLeire and Lopoo,

2010.

8 That youth raised in married-parent families farebetter on a host of socioeconomic outcomes as adultsthan youth in single-parent families, even after takingincome differences into account, is well-establishedin the literature (McLanahan and Sandefur 1994).Here, however, the population of interest is selectedon incomeonly those youth raised in middle-classfamilies are included. As such, the youth from single-parent families who are nonetheless middle class might

have intangible personal and family assets that allowtheir families to attain middle-class status and mightguard against downward mobility. In contrast, youthfrom two-parent families that do not achieve higher-income status (i.e., they are middle class) might haveintangible personal and family liabilities that mightcontribute to downward mobility. This selection intothe middle-class sample contributes to a perversefinding when youth family structure is included in theregression models: being raised in a two-parent familycontributes to downward mobility, particularly amongwhites. Similar selection-driven findings are reportedby Mazumder (2008) and Hertz (2004). Rather thaninclude a measure of family structure in this descriptivemodel and implying that blacks are protected againstdownward mobility because they are more likely tocome from single-mother families, the models hereexclude family structure. Excluding family structurehas negligible effects on the other included variables.

1 Middle class is a social construct that reflectsoccupational status, education and income amongother factors. For ease of explanation in this report,however, the term middle class is used solely as adescription of income status.

2 Economic Mobility Project, 2009. When asked to

rank various definitions of the American Dream ona scale of 1 to 10, with 10 describing the AmericanDream perfectly, respondents ranked your childrenbeing better off financially than you as 7.8 out of 10.

3 See Endnote 1.

4 The NLSY-79 began with a cohort of more than12,000 youth ages 14-21 in 1979, re-interviewingthem annually through 1994 and biennially sincethen, gathering data on employment, income, familyformation, schooling and risk-taking behaviors. In theearly waves of the survey, data on family backgroundand parents were also collected. The sample of youthfrom middle-class families who appear in the NLSYas adults in 2004/2006 comprises 1,189 observations.The number of whites, blacks and Hispanics are 752,236, and 201, respectively. There are 588 men and601 women; among whites there are 386 men and 366women; among blacks, there are 113 men and 123women; and among Hispanics, there are 89 men and112 women. Attrition from the NLSY for youth raisedin middle-class families does not substantially alter theweighted characteristics of our analysis sample.

5 Two years of data are used at each point in time toaverage out variation in annual income. Because theNLSY asks about income received in the previous year,the analyses compare parents income in 1978 and1979 with their adult childrens income in 2003 and2005.

Endnotes


32/36

factors are inconsistently or weakly associated withdownward mobility, or are imprecisely estimated. Seethe Appendix for full regression estimates.

12 The results in this section are based on linearprobability models that control for whether onesmother has a high school diploma, whether onesfather is a professional or manager, ones parentsincome rank in 1979 and ones AFQT percentile score.They also include a full set of dummies for central-city residence and region; dummies for race/ethnicity,educational attainment, marital status, and drug use;and dummies for missing data on mothers education,fathers occupation, region/central-city residence, druguse, and AFQT scores. Figures 3 to 5 do not showvariables that are not consistently substantively orstatistically significant across the different models. Fulldetails and results of the modeling are presented in the

Appendix.

13 These differences are statistically significant ata level of p


33/36


ENDNOTES

most important factor behind black-white downwardmobility differences. No other factor considered inthis report had a statistically meaningful effector asubstantively large effecton any of the black-whitemobility gaps among men, except that the gap in thelikelihood of falling 20 percentiles would have beeneven larger if not for the fact that the parental familyincome rank of black men was lower on averagethan that of white men. Even more strikingly, thecollective effect of the family background factors andof the choice factors was never statistically significant.Between 27 percent and 47 percent of the black-white mobility gap among men is explained by black-white differences in the means of the various factorsincluded in the report. Of course, AFQT scores reflectfamily background and other influences to some

extent, so it is not technically accurate to say thatdifferences in cognitive skills, scholastic achievementor knowledge explain the gap.

20 Nor does estimating separate models for whitemen and women identify a single factor for whichmale-female average differences account for any partof any of the three mobility gaps.

differences between women and men. For example,years of work experience are not explored here,but doing so would likely go some way towardreconciling these findings, since women are morelikely than men to spend time out of the labor forceto raise children.

18 Technically, the analyses are based on stepwiseOLS regression models that pool black and whitemen, in which groups of independent variables areiteratively added to an initial model that includes onlyan indicator variable for race.

19 Using a statistical technique called a Oaxacadecomposition allows an explanation of theracial gap in downward mobility in terms of bothracial differences in the means of the variousfactors themselves (for example, differences incollege graduation rates) and racial differences inthe associationbetween background factors anddownward mobility (for example, racial differences inhow important having a college degree is for avoidingdownward mobility). Consistent with the results inFigure 9, AFQT score differences appear to be the


34/36


35/36

Pro ject PrincipalsRichard Burkhauser, Ph.D., American Enterprise Institute

Marvin Kosters, Ph.D., American Enterprise Institute

Ron Haskins, Ph.D., The Brookings Institution

Stuart Butler, Ph.D., The Heritage Foundation

William Beach, The Heritage Foundation

Ray Boshara, The New America Foundation

Reid Cramer, The New America Foundation

Harry Holzer, Ph.D., The Urban Institute

Eugene Steuerle, Ph.D., The Urban InstituteSheila Zedlewski, The Urban Institute

Pro ject AdvisorsDavid Ellwood, Ph.D., Dean, John F. Kennedy School of Government, Harvard University

Christopher Jencks, M. Ed., John F. Kennedy School of Government, Harvard University

Susan Mayer, Ph.D., Harris School, University of Chicago

Bhashkar Mazumder, Ph.D., Federal Reserve Bank of ChicagoSara McLanahan, Ph.D., Center for Research on Child Wellbeing, Princeton University

Ronald Mincy, Ph.D., School of Social Work, Columbia University

Timothy M. Smeeding, Ph.D., Institute for Research on Poverty,University of Wisconsin-Madison

Eric Wanner, Ph.D., The Russell Sage Foundation


36/36

9 0 1 E S T R E E T , N W , 1 0 T H " " , U 7 - / " ] {

Date post:	03-Apr-2018
Category:	Documents
Upload:	esty-dyah-imaniar
View:	221 times
Download:	0 times

Downward Mobility

Documents