Policies to Address Poverty in America — The Hamilton Project (June 2014)
Summary
Policies to Address Poverty in America, a June 2014 volume from The Hamilton Project at Brookings, edited by Melissa S. Kearney and Benjamin H. Harris. It collects fourteen antipoverty policy proposals organized in four sections: promoting early childhood development, supporting disadvantaged youth, building skills, and improving safety net and work support. Proposal 12, Encouraging Work Sharing to Reduce Unemployment, is by Katharine G. Abraham and Susan N. Houseman. The introduction reports that 30.4 million adults and 16.1 million children lived in poverty in 2012 under the official Census count, and summarizes data on poverty by race, family structure and work status. The volume ends with a list of the proposals and their authors.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
Full text
JUNE 2014
Policies to Address Poverty in America
Edited by Melissa S. Kearney and Benjamin H. Harris
ACKNOWLEDGMENTS
The Hamilton Project wishes to thank members of its Advisory Council for their valuable
insights and contributions to this book. In particular, the Project is grateful to Roger C.
Altman, Robert Greenstein, Peter Orszag, Meeghan Prunty Edelstein, Robert E. Rubin,
and Leslie B. Samuels for helpful discussions and comments on drafts. The contents
of this volume and the individual papers do not necessarily represent the views of
individual Advisory Council members.
The policy memos contained in this volume have benefitted from the expert feedback
provided by participants at The Hamilton Project anti-poverty authors’ conference held
in March 2014 at the Brookings Institution. We are very grateful to all who participated
in that meeting.
The editors acknowledge the outstanding work of The Hamilton Project staff on this
volume. Karen Anderson provided expert guidance on all aspects of production,
including providing feedback on multiple drafts of each of the individual papers. Elisa
Jácome and Lucie Parker provided outstanding research assistance and steadfast
dedication to the production of this volume. It would not have been possible without
them. We also gratefully acknowledge the contributions of Joseph Sullivan, Chanel
Dority, Laura Howell, Brian Goggin, Peggah Khorrami, and Allen Sirolly. And finally, we
are grateful to David Dreyer, Susan Kellam, Alison Hope, and Fred Hviid.
MISSION STATEMENT
The Hamilton Project seeks to advance America’s promise of opportunity, prosperity,
and growth.
We believe that today’s increasingly competitive global economy demands public policy
ideas commensurate with the challenges of the 21st Century. The Project’s economic
strategy reflects a judgment that long-term prosperity is best achieved by fostering
economic growth and broad participation in that growth, by enhancing individual
economic security, and by embracing a role for effective government in making needed
public investments.
Our strategy calls for combining public investment, a secure social safety net, and
fiscal discipline. In that framework, the Project puts forward innovative proposals from
leading economic thinkers — based on credible evidence and experience, not ideology
or doctrine — to introduce new and effective policy options into the national debate.
The Project is named after Alexander Hamilton, the nation’s first Treasury Secretary, who
laid the foundation for the modern American economy. Hamilton stood for sound fiscal
policy, believed that broad-based opportunity for advancement would drive American
economic growth, and recognized that “prudent aids and encouragements on the part of
government” are necessary to enhance and guide market forces. The guiding principles
of the Project remain consistent with these views.
Policies to Address Poverty in America
EDITED BY:
Melissa S. Kearney and Benjamin H. Harris
JUNE 2014
2 Policies to Address Poverty in America
Table of Contents
INTRODUCTION 5
SECTION 1. PROMOTING EARLY CHILDHOOD DEVELOPMENT
PROPOSAL 1. EXPANDING PRESCHOOL ACCESS FOR DISADVANTAGED CHILDREN 19
PROPOSAL 2. ADDRESSING THE PARENTING DIVIDE TO PROMOTE EARLY
CHILDHOOD DEVELOPMENT FOR DISADVANTAGED CHILDREN 29
PROPOSAL 3. REDUCING UNINTENDED PREGNANCIES FOR LOW-INCOME WOMEN 37
SECTION 2. SUPPORTING DISADVANTAGED YOUTH
PROPOSAL 4. DESIGNING EFFECTIVE MENTORING PROGRAMS FOR DISADVANTAGED YOUTH 47
PROPOSAL 5. EXPANDING SUMMER EMPLOYMENT OPPORTUNITIES FOR LOW-INCOME YOUTH 55
PROPOSAL 6. ADDRESSING THE ACADEMIC BARRIERS TO HIGHER EDUCATION 67
SECTION 3. BUILDING SKILLS
PROPOSAL 7. EXPANDING APPRENTICESHIP OPPORTUNITIES IN THE UNITED STATES 79
PROPOSAL 8. IMPROVING EMPLOYMENT OUTCOMES FOR DISADVANTAGED STUDENTS 87
PROPOSAL 9. PROVIDING DISADVANTAGED WORKERS WITH SKILLS TO SUCCEED IN THE LABOR MARKET 97
SECTION 4. IMPROVING SAFETY NET AND WORK SUPPORT
PROPOSAL 10. SUPPORTING LOW-INCOME WORKERS THROUGH REFUNDABLE CHILD-CARE CREDITS 109
PROPOSAL 11. BUILDING ON THE SUCCESS OF THE EARNED INCOME TAX CREDIT 119
PROPOSAL 12. ENCOURAGING WORK SHARING TO REDUCE UNEMPLOYMENT 129
PROPOSAL 13. DESIGNING THOUGHTFUL MINIMUM WAGE POLICY AT THE STATE AND LOCAL LEVELS 137
PROPOSAL 14. SMARTER, BETTER, FASTER: THE POTENTIAL FOR PREDICTIVE ANALYTICS
AND RAPID-CYCLE EVALUATION TO IMPROVE PROGRAM DEVELOPMENT AND OUTCOMES 147
AUTHORS 157
ENDNOTES 163
REFERENCES 171
The Hamilton Project • Brookings 3
4 Policies to Address Poverty in America
Introduction
Melissa S. Kearney, Benjamin H. Harris, and Karen L. Anderson
The Hamilton Project
WHY POVERTY BELONGS ON THE NATIONAL POLICY Project work, 29.6 percent of families live within 150 percent
AGENDA of the poverty line; nearly half live within 250 percent of the
Millions of people live in poverty in this country. They threshold (Kearney et al. 2013). Many individuals and families
suffer not only material deprivation, but also the hardships weave in and out of poverty, even if they are not classified as
and diminished life prospects that come with being poor. poor under the annual income measure. From 2009 to 2011,
Childhood poverty often means growing up without the about 90 million individuals—31.6 percent of Americans—
advantages of a stable home, high-quality schools, or were episodically poor (poor for two or more consecutive
consistent nutrition. Adults in poverty are often hampered months during a thirty-six-month period) (Edwards 2014).
by inadequate skills and education, leading to limited wages In the aftermath of the Great Recession, some disadvantaged
and job opportunities. And the high costs of housing, health workers struggle to obtain the necessary training for
care, and other necessities often mean that people must fruitful employment, while others grapple with long-term
choose between basic needs, sometimes forgoing essentials unemployment at unprecedented rates. Long-term challenges
like meals or medicine. While by some measures the poor remain with us: too many of our nation’s youth drop out of
suffer less material deprivation than their counterparts of a high school, too many of our children are born into unstable
half century ago—almost all households now have access to home environments, and too many of our young adults are
basic necessities like electricity and running water, as well out of school and out of work. This threatens our nation
as consumer goods like televisions and computers—the with the prospect of a permanent class of individuals who
social and economic costs of poverty remain as real as ever are unable to contribute productively to and benefit from a
and threaten to undermine the nation’s social fabric and thriving economy.
economic future.
Furthermore, research demonstrates that poverty leads to
Fifteen percent of Americans—30.4 million adults and 16.1 substantial and sustained neurobiological stressors that can
million children—lived in poverty in 2012, according to the inhibit intellectual and emotional development and sound
official Census poverty count.1 This share rises to 16.0 percent decision making. For children in particular, poverty means
when adjustments for costs and benefits are accounted living with the stress that comes from insufficient nutritional
for under the more comprehensive Supplemental Poverty intake, living in the presence of violence in their community
Measure (SPM). Yet even these counts, as high as they are, or household, and not having a secure place to sleep at night.
understate our nation’s experience with poverty. For every These challenges make it harder for children to learn and thrive
person classified as poor, many more hover just above the in school, which, in turn, leads to problems that cumulate over
threshold. As has been highlighted in earlier Hamilton childhood and into adulthood. The concern is that children
The Hamilton Project • Brookings 5
Introduction
born into deprivation will live their lives stuck in a perpetual of Americans, they constitute 32.4 percent of the episodically
poverty trap. poor and 42.4 percent of the chronically poor (those who are
poor for thirty-six consecutive months) (Edwards 2014).
Improving the economic well-being of less-advantaged
individuals has been a central focus of The Hamilton Project for Poverty is not concentrated among racial and ethnic
many years, which has resulted in numerous discussion papers, minorities, but minorities are disproportionately likely to be
including proposals to expand the wage subsidies for workers,2 poor. Whites make up 74.5 percent of the episodically poor
reform and strengthen the food stamp program,3 provide tax and 62.8 percent of the chronically poor; their corresponding
relief for working families,4 reform unemployment insurance,5 population share is 80.1 percent. African Americans
expand access to higher education,6 as well as a proposal to comprise 18.1 percent of the episodically poor, 31.0 percent of
develop a better measurement of poverty,7 among others. This the chronically poor, and 12.6 percent of the total population
volume builds on this focus and these existing proposals. (Edwards 2014). Hispanics constitute 17.1 percent of the
total population and 25.6 percent of individuals in poverty
Poverty is a complex, multifaceted problem that can be (DeNavas-Walt, Proctor, and Smith 2013).
overcome only through a comprehensive set of innovative
policies and effective reforms. Tackling poverty requires Those with steady employment, not surprisingly, are much less
a national commitment toward building human capital, likely to be poor, but work is no assurance that individuals can
harnessing the economic power of that investment, and escape poverty. While only 2.9 percent of full-time, year-round
providing a safety net when jobs are scarce or individuals are workers live in poverty, 7.3 percent of all workers do not earn
simply not intellectually or physically capable of economic self- more than the poverty threshold (DeNavas-Walt, Proctor, and
sufficiency. It means a commitment to addressing the causes Smith 2013). Those who find stable employment often work for
and consequences of poverty throughout the life course. wages too low to enable them to rise above the poverty line.
Many other workers struggle to find full-time jobs.
In recognition of these challenges, The Hamilton Project
has commissioned fourteen innovative, evidence-based Family structure, along with work status, is also an important
antipoverty proposals. These proposals are authored by a determinant of poverty rates. Individuals living in married-
diverse set of leading scholars, each tackling a specific aspect couple families make up 64.0 percent of the total population,
of the poverty crisis. The papers are organized into four broad but account for 47.8 percent of the episodically poor and 25.7
categories: (1) promoting early childhood development, (2) percent of the chronically poor. Though individuals living
supporting disadvantaged youth, (3) building skills, and in female-headed households make up only 14.9 percent
(4) improving safety net and work support. The proposals of the total population, they constitute 25.0 percent of the
put forward in this volume are forward-looking and, if episodically poor and 42.8 percent of the chronically poor
implemented, would have important beneficial impacts on the (Edwards 2014). Among the 7.1 million families with income
future well-being of America’s next generation. below the federal poverty level, 69.7 percent are headed by a
single parent (60.4 percent are single female parents) and 30.3
WHO IS POOR IN AMERICA? percent are headed by a married couple (Kearney et al. 2013).
The face of poverty in America is diverse, and includes
In summary, poverty affects a diverse population of individuals,
individuals of all races and ethnicities, ages, and family types.
with varying geographic, racial, age, employment, and family
Poverty is found across all fifty states and in Washington, DC.
characteristics. Poverty is not a static condition; many people
In 2012, every state had a poverty rate of at least 10 percent,
cycle in and out of poverty. For many Americans, it is a lifelong
ranging from a high of 24.2 percent in Mississippi to a low of
threat: two-thirds of Americans will live in poverty for at least
10.0 percent in New Hampshire (Bishaw 2013). While poverty
a year at some point in their lives (Rank and Hirschl 1999).
is present in every major metropolitan area in the country,
There is no silver bullet policy lever to combat poverty, but
it also resides in rural counties and suburbs. In 2012, 14.5
there are effective ways to intervene at all points in the life
percent of Americans living inside metropolitan areas were
course and hammer away at the root causes of poverty and its
classified as poor, as were 17.7 percent of Americans living
consequences of economic disadvantage.
outside metropolitan areas (DeNavas-Walt, Proctor, and
Smith 2013). PROMOTING EARLY CHILDHOOD DEVELOPMENT
Some groups of people are more likely than others to Achievement gaps between children from low- and high-
experience the hardships of poverty, however. Children are income families appear early in life and then persist through
especially at risk, with poverty rates that are nearly twice that high school and afterwards. For example, by age four, children
for elderly Americans. Though children make up 25.2 percent in the highest income quintile score, on average, near the 70th
6 Policies to Address Poverty in America
percentile on tests of literacy and mathematics, compared to Earlier policy efforts focused on marriage promotion yielded
children in the lowest-income quintile who score near the 30th disappointing findings. As a result, poverty scholars are
percentile (Waldfogel and Washbrook 2011). Scholars and turning to an emphasis on delaying pregnancy and preventing
policymakers have increasingly come to appreciate the role unplanned pregnancies, with the goal of increasing the rate
of noncognitive skills as well, highlighting the importance of at which children are born to mothers and fathers who have
socioemotional traits such as self-esteem and self-control that planned for those births and are in a better position to care for
develop early in life (Heckman, Stixrud, and Urzua 2006). their children. The policy memo by Isabel Sawhill and Joanna
Venator addresses this issue and puts forward a proposal to
Early childhood interventions can play an important role in promote greater knowledge and choices about contraception
addressing poverty in America. These interventions need to among women and their health-care providers.
be broad in their focus, and need to address issues of early
childhood schooling and high-quality child care, as well as SUPPORTING DISADVANTAGED YOUTH
addressing family circumstance and parenting practices. Disadvantaged youth seemingly face barriers at every turn.
The work of Nobel laureate James Heckman and colleagues They all too often struggle in school, commit crimes and
has emphasized that early childhood interventions can have are victims of violent crimes, have few positive adult role
significant long-term impacts on educational and economic models in their lives, and lack sufficient skills—academic and
attainment (see, for example, Knudsen et al. 2006). These behavioral—to succeed in the workforce.
findings have been highlighted in earlier work by The Hamilton
Project.8 In this volume, Elizabeth U. Cascio and Diane The rate at which disadvantaged youth drop out of high
Whitmore Schanzenbach contribute a policy memo offering school is one concrete measure of how our nation’s poor
a thoughtful consideration of early childhood education and youth struggle to move up the economic ladder. According
proposing a framework for states to improve their educational to recent estimates, nearly four in ten eighth-grade students
investment in young children by expanding access to high- from families in the lowest income quartile did not eventually
quality preschool. graduate from high school (Ingels, Owings, and Kaufman
2002). In school districts located in our country’s fifty largest
The home environment is also a crucial input into early cities, only 53 percent of students graduated from high
childhood experiences. On this dimension, too, poor children school (Swanson 2009). These dropout rates are particularly
are increasingly at a disadvantage. Numerous studies have worrisome given the limited earnings and job prospects for
shown that higher-educated, higher-income parents spend high school dropouts in today’s economy. The consequences
more time with their children, and more time in educational of low educational attainment and lack of labor market skills
activities in particular (Guryan, Hurst, and Kearney 2008; are too severe to ignore; thus, finding effective ways to foster
Kalil, Ryan, and Corey 2012). The policy memo by Ariel Kalil in the academic skills and socioemotional development of
this volume proposes a new federal initiative to study effective disadvantaged youth through their teenage years must be a
early childhood interventions in the home environment. priority in our nation’s multipronged attack on poverty.
Better understanding of these programs can ultimately
lead to smarter, more-innovative, and more-accountable In their policy memo, Amy Ellen Schwartz and Jacob Leos-Urbel
developmental programs for children and families. cite an emerging body of research suggesting that, in addition
to the immediate benefits of a summer job and the wages
In terms of family structure, it is important to acknowledge associated with that employment, summer youth employment
that poverty rates are five times as high among children living programs can improve educational outcomes, strengthen social
with single mothers compared to children in two-parent and emotional development, and decrease crime rates. Building
households. This has led to concern among scholars about on evidence that summer employment can be a very positive
the rise in single motherhood and its associated consequences driver of adolescent development, Schwartz and Leos-Urbel
for poverty. It has also led to concern about the rate at which propose a nationwide summer youth employment program,
lower socioeconomic groups are moving away from marriage aimed at helping low-income youth to build human capital and
and the implications that has for the intergenerational so transition to a productive adulthood.
transmission of poverty. For instance, only 9 percent of births
to college-educated women are outside marriage (virtually Mentorship, too, can play a critical role in positive youth
unchanged from a generation ago). In sharp contrast, 57 development. In his memo, Phillip B. Levine notes that
percent of first births to women with high school diplomas or upwards of 9 million children have no caring adults in their
less are nonmarital (Shattuck and Kreider 2013). lives; he cites credible evidence that effective mentoring
programs can help propel young people up the economic
The Hamilton Project • Brookings 7
Introduction
ladder (Bruce and Bridgeland 2014; Cavell et al. 2009). He increase in their wages has not kept up with those with more-
establishes a framework for evaluating mentorship programs, advanced education.
calling for higher levels of private and non-profit-sector
investment in youth mentorship. A second, related trend is what labor economists have referred
to as a polarization of job opportunities in the United States.
The policy memo by Bridget Terry Long addresses the issue As David Autor explained in his earlier Hamilton Project
of underpreparation for college. Long notes that only 32 paper, the U.S. labor market has witnessed expanding job
percent of students leave high school at least minimally opportunities in high-skilled, high-wage occupations on
prepared for four-year college, and the proportion is much the one end, and low-skilled, low-wage occupations on the
smaller for African American and Hispanic students—20 and other.13 Employment prospects for middle-skilled workers in
16 percent, respectively (Greene and Foster 2003). Moreover, white-collar occupations—clerical, administrative, and sales
only 59 percent of low-income students who met a minimum positions—have weakened, as have those for middle-skilled
standard of being academically qualified for college completed workers in blue-collar occupations—production, craft, and
a bachelor’s degree within eight years, in contrast to 89 percent operative positions. These trends have been experienced by
of high-income students (Adelman 2006). This low level of other economies around the world, suggesting that there are
preparation threatens college completion: only 9 percent of global economic forces that have led to a restructuring of the
students from the bottom income quartile who enter college labor market.
actually complete a bachelor’s degree by age twenty-five
(Bailey and Dynarski 2011). Long proposes to reform the The magnitude of this challenge and its stark implications
remediation system in this country to better support young, for poverty in America can only be addressed with a massive
underprepared students in their transition to college. commitment to skill-upgrading. To date, however, our nation’s
commitment to investment in skills has lagged behind that of
In addition to tackling the three issues highlighted here, other countries. As Sheena McConnell, Irma Perez-Johnson,
strengthening our country’s K–12 education system is of and Jillian Berk point out in this volume, the United States does
utmost importance. Multiple papers previously published by not currently invest heavily in vocational training compared
The Hamilton Project have addressed this issue, and so we do with other countries. Whereas the United States spends less
not include papers on education reform in this volume.9 than 0.05 percent of its GDP on vocational training, other
industrialized nations invest up to ten times as much. In their
BUILDING SKILLS policy memo, McConnell, Perez-Johnson, and Berk propose
Skill development and job creation are critical components of strengthening vocational training for disadvantaged adult
our nation’s fight against poverty. It is increasingly difficult workers to boost employment and reduce poverty.
for individuals to be economically secure in today’s global
economy with limited skills and education. Recognizing As Robert I. Lerman points out in this volume, the United
the paramount role of adequate skill and job creation in States also lags far behind our competitors in apprenticeship
our national economy, The Hamilton Project has devoted investment. While apprenticeships offer a productivity-
enhancing approach to reducing inequality and expanding
considerable attention to these topics in years past, with
opportunity, Lerman notes that the numbers in the United
papers on using data to improve workforce training,10 creating
States have declined in recent years to levels about one-tenth
more-effective education and workforce development systems
of those in Australia, Canada, or Great Britain. Lerman puts
in the states, 11 and improving worker advancement in the low-
forth a proposal to better encourage apprenticeship training
wage labor market.12
and put the United States on a par with other countries with
Stagnant wage growth for low-skilled workers is a persistent regard to training. On a related topic, Harry J. Holzer in his
economic threat. For four decades, high-skilled workers have policy memo observes that the courses pursued by many low-
seen their wages increase while less-skilled workers have income college students do not equip them with the skills
seen their economic positions erode. High school graduates demanded by the labor market. Holzer’s proposal focuses
and those with less than a high school diploma saw their real on educational reform to incentivize public colleges and
wages fall through the late 1970s and 1980s and rebound a bit universities to better tailor their curricula to improve labor
in the early 1990s, only to remain stagnant since then (Autor, market outcomes for graduates. Clearly, there is significant
Katz, and Kearney 2008). In contrast, since the mid-1970s, opportunity to improve our system of education and training
those with the highest levels of education—more than sixteen to better equip America’s workforce with the skills that are
years—have seen their wages rise steadily. Those with a college demanded and rewarded in today’s global economy.
degree or some college have seen some improvement, but the
8 Policies to Address Poverty in America
IMPROVING SAFETY NET AND WORK SUPPORT and wage-support policies like the federal minimum wage.
A strong safety net is crucial to fighting poverty in America. Arindrajit Dube proposes a framework for designing effective
Without programs designed to lift the poorest households out minimum wage policies at the state and local levels to better
of poverty, roughly twice as many Americans would live below compensate workers in high-cost areas in a way that recognizes
the poverty line today. As revealed by the SPM (see footnote and minimizes potential negative employment effects.
1), including government programs in the calculation of Katharine G. Abraham and Susan N. Houseman contend that
poverty halves the share of Americans classified as poor from the unemployment insurance program could be even more
31 percent to 16 percent (Fox et al. 2013). Evidence further effective if it facilitated work-sharing arrangements, such that
suggests that the safety net is becoming even more effective at employers would be less inclined to reduce their workforce
fighting poverty: in 1967, government benefits cut poverty by during cyclical downturns. They find that if U.S. employers
only about one-quarter. had work-sharing usage at European levels, as many as one
The safety net has become especially effective at fighting in eight of the roughly 8 million jobs lost during the Great
poverty among the elderly. Programs like Social Security, Recession could have been saved (Abraham and Houseman
Medicare, and Supplemental Security Income—making up forthcoming). In their policy memo in this volume, Abraham
36.1 percent of the federal budget in 2012—have helped drive and Houseman propose reform of the U.S. network of work-
elderly poverty down to less than 10 percent and so promote a sharing programs to reduce unemployment, especially during
dignified and healthy retirement for America’s oldest citizens economic downturns.
(Danziger and Danziger 2005). In many ways, the social safety In addition to the three policy memos described above, two
net for elderly Americans can be considered a great success. other papers in this volume discuss proposals for supporting
The two largest safety-net programs today, in terms of low-income families. Recognizing that child-care costs can
expenditure outlays, are the Earned Income Tax Credit discourage work and take up valuable resources for low-income
(EITC) and the Supplemental Nutritional Assistance Program families, James P. Ziliak proposes expanding and reforming
(SNAP). Poverty scholars generally regard these programs to the tax credit for child care to make work pay for working
be effective. SNAP is the quintessential safety-net program parents. Finally, Scott Cody and Andrew Asher propose
and has proven to be responsive to weak economic conditions improving the administration of safety-net programs at all
in exactly the way a true safety-net program should be. levels of government by harnessing the power of predictive
When economic conditions weaken, SNAP caseloads rise; analytics and rapid-cycle evaluation. If adopted, the proposal
when economic conditions improve, SNAP caseloads fall. in their paper would make social safety-net programs more
Furthermore, researchers have documented the long-term cost-effective, while also guiding program administrators in
health and economic benefits of this food assistance program their quest to better support American families.
to low-income children and individuals (Almond, Hoynes, CONCLUSION
and Schanzenbach 2011; Hoynes, Schanzenbach, and Almond
2012). A recent Hamilton Project discussion paper by Diane Poverty remains one of America’s most important policy
Whitmore Schanzenbach proposed reforms to strengthen challenges. On any given day, 46.5 million Americans,
SNAP to make the nutritional benefits even greater.14 including 16.1 million children, endure the hardships of
poverty. Millions more hover with great vulnerability just
The EITC has been shown to encourage work among single above the poverty line. Still more may be able to meet their
mothers and to lead to long-term improvements in the well- current basic needs, only to find themselves living in poverty
being of families and children (Dahl and Lochner 2012; Evans in the future. The persistent threat of poverty represents a
and Garthwaite 2014; Hoynes, Miller, and Simon forthcoming). failure of our economic system to provide all children with
As noted by Hilary Hoynes in this volume, the EITC also has the support they need to acquire human capital and to provide
immediate and significant impacts on poverty, raising 6.5 able-bodied working-age Americans sufficient opportunities
million Americans out of poverty in 2012 alone (CBPP 2014). for stable and well-paid employment.
Hoynes’ policy memo in this volume proposes to build on this
success by raising the EITC benefits for one-child families. No single policy will cure poverty, and this volume
recognizes the multidimensional nature of the problem. In
Another set of programs and policies aimed at working this collection of fourteen policy memos, national experts
Americans are not classified as safety-net programs, but are put forth individual evidence-based proposals, each designed
instead considered to be work support for those in the labor to address a specific aspect of poverty. Proposals range from
force. These programs include unemployment insurance aiding the development of the youngest individuals, to
The Hamilton Project • Brookings 9
Introduction
supporting disadvantaged teens, to improving our national Public policies have a played a significant role in mitigating
system of training and education. Importantly, the various the devastations of poverty. Fully twice as many Americans
policy memos call on a variety of implementing agencies; would be impoverished if not for public safety-net programs.
this is an acknowledgement of the reality that alleviating During the Great Recession countercyclical antipoverty
poverty requires commitments by governments at all levels, programs like SNAP and unemployment insurance served
in addition to the private sector and nongovernmental to support millions of American families in need, and not
agencies. Poverty is indeed a nationwide problem that only eased the pain of the recession, but also contributed to
requires a nationwide solution. the recovery. Policymakers continue to rely on American
innovation to improve these programs—applying technology
This volume does not consider the full range of antipoverty and knowledge to the administration and evaluation of public
policies. Readers may note an absence of policies relating programs in an effort to improve their effectiveness and
to fighting homelessness or reforming disability insurance reduce their cost. But there is still much to do.
programs. There are no policies directly relating to asset
accumulation, such as those to support homeownership With commitment, focus, and hard-headed compassion,
or to increase savings. Nor do we address the issue of K–12 policymakers and concerned individuals can make a sustained
education—a major concern for those at the lower end of the difference and bring down the stubbornly high rates of poverty
income distribution. Some of these topics have been addressed in the United States. The proposals included in this volume are
by prior Hamilton Project discussion papers; others will be put forward with the goal of making economic prosperity a
addressed in future Hamilton Project work. more broadly shared promise for all who live in our wealthy
nation. In this spirit, we offer fourteen new policy proposals to
help address and reduce poverty in America.
10 Policies to Address Poverty in America
The Hamilton Project • Brookings 11
Introduction
Summary of Proposals
Paper title Proposal description
Section 1. Promoting Early Childhood Development
Expanding Preschool Access for Proposes a framework calling for the establishment of a high-quality program in areas
Disadvantaged Children where preschool programs do not exist, improved preschool quality in those states
and localities with subpar programs, and expanded access in areas where high-quality
Elizabeth U. Cascio and Diane Whitmore
programs already exist.
Schanzenbach
Addressing the Parenting Divide to Proposes a new federal task force supporting the collection of evidence to develop
Promote Early Childhood Development more-effective parenting interventions and to promote improved child development in
for Disadvantaged Children early years.
Ariel Kalil
Reducing Unintended Pregnancies for Proposes to combat unintended pregnancies through a social marketing campaign to
Low-Income Women encourage more young women to use long-acting reversible contraceptives (LARCs).
Isabel Sawhill and Joanna Venator
Section 2. Supporting Disadvantaged Youth
Designing Effective Mentoring Programs Proposes expanding community-based mentoring programs, such as the Big Brothers
for Disadvantaged Youth Big Sisters program, in accordance with a set of best practices.
Phillip B. Levine
Expanding Summer Employment Proposes distribution of federal grants to states for municipalities to provide summer
Opportunities for Low-Income Youth employment to disadvantaged youth, first through a pilot program and then through a
nationwide expansion.
Amy Ellen Schwartz and Jacob Leos-Urbel
Addressing the Academic Barriers to Proposes improving placement in college remediation classes, providing better college
Higher Education remediation services, and adopting measures to prevent the need for remediation.
Bridget Terry Long
12 Policies to Address Poverty in America
Targeted population Who implements? Potential outcomes
Disadvantaged preschool- State and local governments Promote expansion of cost-effective, high-quality public preschool
aged children, especially for low-income children to reduce the income-based gap in
those who currently have school readiness and improve school outcomes for disadvantaged
limited preschool access preschool children. Costs would depend on the existing preschool
options in each state.
Low-income families with U.S. Department of Health Collect evidence on successful parenting interventions for young
young children between the and Human Services’ children through rigorous experiments, and develop new interventions
ages of 0 and 5 Administration for Children that are lower cost and better matched to families’ needs.
and Families
Low-income, unmarried U.S. Department of Health Expand awareness so more low-income women use a LARC
women between the ages and Human Services’ or another method of contraception, and thereby reduce the
of 15 and 30 Office of Population Affairs, number of unintended pregnancies to lower the number of
in conjunction with state children born into poverty.
governments
Disadvantaged youth who Nongovernmental Improve educational outcomes for disadvantaged youth and raise
have no or few adult role organizations—including lifetime earnings by approximately $7,500. Prior program costs have
models in their lives nonprofits, foundations, averaged approximately $1,600 per child.
charitable organizations, and
others—as well as private-
sector entities, and the
federal government in some
circumstances
Low-income youth between U.S. Department of Labor, Expand summer job programs for disadvantaged youth to increase
the ages of 16 and 19 who state and local governments, school attendance, improve educational outcomes, and reduce violent
are enrolled in, or have and community-based behavior and crime. Cost would be roughly $2,000 per participant.
recently graduated from, organizations
high school
Disadvantaged, School districts, community Reduce the need for college-level remediation and better match
academically colleges, university systems, underprepared students with effective resources and supports to
underprepared students in and state and federal equip them with the skills they need to succeed in college and in the
high school and college governments workforce. Reforms would likely result in higher educational outlays in
the short run, but would lead to cost savings for students, institutions,
and taxpayers in the long run.
The Hamilton Project • Brookings 13
Introduction
Summary of Proposals
Paper title Proposal description
Section 3. Building Skills
Expanding Apprenticeship Opportunities Proposes a series of targeted federal and state-level initiatives to expand access to
in the United States registered apprenticeship programs by creating marketing initiatives, building on existing
youth apprenticeship programs, extending the use of federal subsidies, and designating
Robert I. Lerman
occupational standards.
Improving Employment Outcomes for Proposes the creation of financial incentives for public colleges to offer classes in high-
Disadvantaged Students return fields and for employers to offer more training to their employees.
Harry J. Holzer
Providing Disadvantaged Workers with Proposes increased funding for training programs targeted to low-skill workers through
Skills to Succeed in the Labor Market the Workforce Investment Act (WIA) Adult program and a series of reforms to training
programming offered by state and local workforce boards.
Sheena McConnell, Irma Perez-Johnson,
and Jillian Berk
Section 4. Improving Safety Net and Work Support
Supporting Low-Income Workers through Proposes converting the federal Child and Dependent Care Credit from a
Refundable Child-Care Credits nonrefundable tax credit to a refundable one, capping eligibility at $70,000 and making
the credit a progressive function of income, the age of the child, and utilization of
James P. Ziliak
licensed care facilities.
Building on the Success of the Earned Proposes expanding the Earned Income Tax Credit (EITC) by raising the benefits for
Income Tax Credit families with one child to be on par with the benefits for families with two children.
Hilary Hoynes
Encouraging Work Sharing to Reduce Proposes that the federal government subsidize state work-sharing payments during
Unemployment economic downturns, make work sharing a requirement for state unemployment
insurance systems, change federal requirements to modify provisions of state work-
Katharine G. Abraham and Susan N.
sharing plans that may discourage employer participation, and provide states with
Houseman
adequate funding to administer work-sharing programs.
Designing Thoughtful Minimum Wage Proposes that states and localities consider median wages and local costs when
Policy at the State and Local Levels setting minimum wages, index the minimum wage for inflation, and engage in regional
wage setting.
Arindrajit Dube
Smarter, Better, Faster: The Potential Proposes that federal, state, and local agencies conduct thorough needs assessments to
for Predictive Analytics and Rapid- identify where predictive analytics and rapid-cycle evaluation can improve service delivery.
Cycle Evaluation to Improve Program
Development and Outcomes
Scott Cody and Andrew Asher
14 Policies to Address Poverty in America
Targeted population Who implements? Potential outcomes
At-risk youth and middle- U.S. Department of Labor’s Expand apprenticeship opportunities for both youth and adults to
skill adults in low-wage jobs Office of Apprenticeship, U.S. improve human capital and raise earnings by an average of nearly
Department of Commerce, $78,000 over two and a half years after leaving the program. Costs
state governments, and would vary, but successful programs have been implemented at
Career Academies $5,500 per apprentice.
Disadvantaged youth who Public colleges and university Incent colleges to reform their curricula to generate better labor
possess at least some systems, state governments, market outcomes, with wage gains of up to 30%, for disadvantaged
level of basic academic and the federal government students. Estimates suggest that $2 billion in expenditures could fund
preparation for higher occupational training for up to 2 million individuals.
education
Low-skilled adult workers Federal government through Improve labor market outcomes—including earnings increases of
with limited workforce congressional legislation, and between $300 and $900 per quarter—for inexperienced, low-skilled
experience state and local workforce adult workers. Increasing WIA funding to generate these benefits
boards would require direct outlays from the federal government.
Low- and middle-income Federal government through Increase tax-based subsidies for center-based, quality child care
working families with less congressional legislation for low-income families to increase their labor force participation,
than $70,000 in income disposable income, and usage of higher-quality child care. Cost would
and with children ages 12 depend on a host of factors, but the proposal could be revenue-
and under neutral or better.
Low-income families with Federal government through Strengthen work incentives for low-income one-child families. Raise
one child who qualify for congressional legislation after-tax income by about $1,000 for one-child EITC beneficiaries,
the EITC leading to improved health and children’s cognitive skills, and raising
410,000 people—including 131,000 children—out of poverty. Annual
cost would be roughly $9 billion before accounting for extra tax
revenue from higher levels of work.
American workers who Federal government through Increase employers’ usage of work sharing rather than layoffs during
would otherwise become congressional legislation and cyclical downturns, which could have saved as many as 1 in 8 of the
unemployed during a U.S. Department of Labor, jobs that were lost during the Great Recession. Work sharing may
cyclical economic downturn and state governments impose costs by reducing the pace of structural adjustment during
economic downturns, but if the program is well designed, this effect
should not be substantial relative to the policy’s benefits.
Low-wage workers State and local governments Raise the earnings of low-wage workers with minimal negative impacts
on employment.
Agencies at all levels Agencies that administer Provide more effective services for individuals living in poverty by
of government that run social service programs targeting services appropriately and by identifying effective program
programs targeting across all levels of improvements. Initial investments in analytical capabilities may be
individuals living in poverty government offset by long-term savings.
The Hamilton Project • Brookings 15
16 Policies to Address Poverty in America
Section 1.
Promoting Early Childhood Development
The Hamilton Project • Brookings 17
18 Policies to Address Poverty in America
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 1: Expanding Preschool Access
for Disadvantaged Children
Elizabeth U. Cascio
Dartmouth College
Diane Whitmore Schanzenbach
Northwestern University
Introduction shown in figure 1-1, only about 10 percent of four-year-old
children nationwide participate in Head Start, a rate that has
Poverty has little association with the cognitive abilities of stayed roughly constant for the past twenty years. Essentially
nine-month-old children (Fryer and Levitt 2013).1 By the start all the growth in public preschool enrollment over time has
of kindergarten, however, not only do poor children perform come from the expansion of state-funded programs, which
significantly worse on tests of cognitive ability than children grew from four states in 1980 to forty states today.
from higher-income families, but teachers also report that
these children have much more difficulty paying attention and Even so, many state programs have weak standards, as shown
exhibit more behavioral problems (Duncan and Magnuson in figure 1-2. During the 2011–12 school year, only 9 percent
2011).2 The poverty gap in school readiness appears to be of all four-year-olds nationwide—roughly 31 percent of
growing as income inequality widens (Reardon 2011). those enrolled in state-funded preschools—were enrolled in
programs that met at least eight common quality benchmarks
THE POLICY LANDSCAPE related to curriculum, teacher education, class size, and
support services.4 The average Head Start program meets only
One popular proposal to narrow this gap is to expand formal
five of these benchmarks (Espinosa 2002).
educational opportunities to poor children under the age
of five. Stark gaps in preschool participation by family In this context, President Obama proposed to expand access
socioeconomic status mirror the achievement gaps described to preschool education while simultaneously leveling up
above. The most recent data available show that only about preschool quality nationwide (Office of the Press Secretary
50 percent of four-year-old children in families in the lowest 2013). The White House proposal would provide block grants
income quintile are enrolled in preschool. Among families to states to offer free preschool education to four-year-old
in the top income quintile, on the other hand, the preschool children from low- and moderate-income families, provided
enrollment rate of four-year-olds is considerably higher, at 76 that these preschool programs score highly on the quality
percent. Nearly all (88 percent) of preschool participants in standards checklist presented on the vertical axis in figure 1-2.5
the lowest-income families are enrolled in public programs.3 State and local governments are not waiting for federal action.
Most notably, New York City mayor Bill DeBlasio campaigned
Poor children can currently attend preschool for free through
on the promise of funding universal pre-kindergarten (pre-K),
two programs: the federally funded Head Start program,
and in March 2014 New York governor Andrew Cuomo and
which targets children in families with incomes less than 130
the state legislature agreed to a five-year, $1.5 billion plan to
percent of the federal poverty level; and state-funded public
offer high-quality full-day pre-K—not just in New York City,
programs, which may also serve middle-class children. As
but across the state.
The Hamilton Project • Brookings 19
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 1: Expanding Preschool Access for Disadvantaged Children
FIGURE 1-1.
Percent of Four-Year-Olds Enrolled in Public Preschool Programs and Number of States
Funding Preschool Programs, 1965–2011
45 45
40 40
35 35
Percent of four-year-olds
30 30
Number of states
25 25
20 20
15 15
10 10
5 5
0 0
1965 1970 1975 1980 1985 1990 1995 2000 2005 2010
Public preschool enrollment rate Head Start enrollment rate Number of states funding preschool
Sources: Barnett et al. 2012; The Inter-university Consortium for Political and Social Research (ICPSR) n.d.; Martin et al. 2013; National Bureau of Economic Research (NBER) n.d.; Office of
Head Start (OHS) various years; National Center for Health Statistics (NCHS) 2005; authors’ calculations.
Note: Data on the public preschool enrollment rate come from the Current Population Survey, October supplement. For 1968–1992, data are derived from ICPSR (n.d.). For 1993–2011,
data are derived from NBER (n.d.). The Head Start enrollment rate is the Head Start enrollment of four-year-olds (calculated as total national Head Start enrollment multiplied by the share of
enrollment comprising four-year-olds) in a given year divided by the number of children born in the United States four years prior. Data on Head Start enrollment come from the OHS (various
years). Data on the number of children born for 1990–2007 (corresponding to the number of children age four for 1994-2011) come from Martin and colleagues (2013). Data on the number of
children born for 1974–1989 (corresponding to the number of children age four for 1978-1993) come from NCHS (2005). Data on the number of states funding preschool come from Barnett
and colleagues (2012).
Evidence on the impacts of early education is broadly
supportive of policy efforts in early education. The research on
The Challenge
early education has shown it improves participants’ outcomes Given that there are several ways to expand preschool access,
across a variety of dimensions: higher school attendance rates, the policy challenge is to design an expansion program that
fewer failing grades, less grade retention, a higher likelihood of is cost-effective. Cost-effectiveness requires that policymakers
graduating from high school, and less involvement in criminal consider the likely benefits of a particular intervention in a
activity. Improvements in these areas account for many of given setting.
the economic benefits of preschool programs. However,
important questions remain regarding access—the benefits A useful organizing framework for the policy evidence is
versus the costs of expanding public preschool options beyond to consider the quality of a possible preschool intervention
lower-income children—and exactly how quality would be against the quality of the environment in which a child
best defined from a policy perspective. This policy memo is would otherwise be placed. A preschool program with
directed primarily toward state and local policymakers who a developmentally appropriate curriculum, nurturing
want to strengthen the public preschool options in their area student–teacher interactions, and parental support might be
while considering budgetary trade-offs. beneficial in preparing disadvantaged children for school,
but less beneficial for children from an already otherwise
enriched environment. Even a lower-quality preschool
program can have an impact on children from the most
disadvantaged backgrounds.
20 Policies to Address Poverty in America
Elizabeth U. Cascio, Diane Whitmore Schanzenbach
FIGURE 1-2.
Relationship between Quality and Access in State-Funded Preschool Programs, 2011–12
School Year
12
10
Score on quality standards checklist
8
6
4
2
0
0 20 40 60 80 100
Percent of four-year-olds in state-funded preschool
Source: Barnett et al. 2012.
Note: Bubble size represents the number of children born in the state four years prior. The dashed line represents the regression fit, weighting by this figure; the unweighted fit is substantively
similar. The quality standards checklist gives equal weight to each of ten factors: (1) program has comprehensive early learning standards; (2) teachers are required to have a bachelor’s degree;
(3) teachers are required to have specialized training in preschool; (4) assistant teachers are required to have a Child Development Associates (CDA) degree (or equivalent); (5) teachers are
required to attend at least fifteen hours per year of in-service; (6) the maximum class size is twenty students; (7) staff to child ratios are 1:10 or better; (8) program offers vision, hearing, health,
and one support service; (9) program offers at least one meal; (10) program offers site visits.
This organizing framework is illustrated graphically in Perry (along with other high-quality, targeted preschool
figure 1-3. On the horizontal axis is an index measure of a interventions, such as the Abecedarian and Nurse-Family
child’s socioeconomic status, which can be thought of as a Partnership) provides excellent evidence because it was a
combination of family income, educational attainment of randomized controlled experiment that collected follow-up
the adults in the home, and so on. On the vertical axis is the data on participants for decades. Early findings from Perry
quality of the child’s learning environment. Considering home showed initial increases in IQ scores for the treatment group,
inputs alone, as shown by the purple line, there is a positive although these gains faded to zero by the time participants
relationship between the child’s socioeconomic status and the reached age ten (Gramlich 1986; Schweinhart et al. 2005).
quality of the child’s learning environment.6 Despite no difference in measured IQ by late childhood, the
Perry treatment students performed statistically significantly
One line of evidence on the longer-run impacts of preschool better in school: they were absent fewer days, were less likely
participation derives from programs of the first variety— to have been assigned to special education, had fewer failing
programs that are very high quality and serve very grades and higher high school grade point averages, were
disadvantaged populations. Arguably the most famous of more likely to graduate from high school, and generally
these is the Perry Preschool program, drawn in light green in reported more-positive attitudes toward schooling. These
figure 1-3. Perry Preschool was a two-year intervention in the improvements persisted into adulthood, when the treatment
early 1960s involving half-day school attendance and weekly group was statistically significantly more likely to be employed
home visits for extremely disadvantaged three- and four-year- and less likely either to have been arrested or to have received
old African American children living in Ypsilanti, Michigan. transfer payments such as cash welfare or Supplemental
The Hamilton Project • Brookings 21
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 1: Expanding Preschool Access for Disadvantaged Children
FIGURE 1-3.
Framework for Considering the Impact of Preschool, Historic Context
Quality of learning environment
Perry preschool
Head Start (historic) Home inputs
(no preschool)
Socioeconomic status
Nutrition Assistance Program benefits (formerly known as and are more likely to complete high school and attend college
the Food Stamp Program).7 (Deming 2009; Garces, Thomas, and Currie 2002).8
Considering the improvements in long-term outcomes from One criticism of Head Start is that it is low quality on average,
a monetary standpoint, every $1.00 spent on the program and exhibits variable quality across locations. While it is
translated into $8.00 worth of benefits (Heckman et al. 2010). The considered lower quality than the Perry program, figure 1-3
high rate of return to Perry Preschool may represent an upper illustrates that Head Start is nonetheless a higher-quality
bound on the return to preschool investment today, because (as environment than what the participant would experience in
illustrated in figure 1-3) it represented such a large increase in the absence of the program, either at home or in the type of
the quality of the participants’ learning environments. child care that is typically available to low-income parents
(Currie 2001). Since Head Start represents a less dramatic
Another line of evidence derives from Head Start, the long- increase in the quality of a child’s environment than Perry
standing federal preschool program. Head Start is considered Preschool, its long-term impacts are more muted but still
to be lower quality than Perry Preschool, and although it is positive.
targeted to low-income children, it serves a large number
of children who are not subject to such extreme levels of Figure 1-1 shows that more children across the income
disadvantage. As represented by the blue line in figure 1-3, the distribution are attending preschool today than ever before.
long-term Head Start evidence spans cohorts of preschool- However, preschool quality varies across socioeconomic status,
age children between 1968 and 1990, a period of expansion as illustrated in the conceptual diagram in figure 1-4. Against
in other preschool opportunities for low-income children (see the backdrop of increasing preschool enrollment, the first
figure 1-1). Although experimental evidence is not available randomized evaluation of Head Start was conducted in 2002;
from this period, there are several careful quasi-experimental the results sharply differ from the earlier quasi-experimental
studies that demonstrate impressive impacts of Head Start research. While four-year-old Head Start participants in the
on both short- and long-term outcomes. For example, Head Head Start Impact Study saw faster improvements in language
Start has been shown to have a substantial positive effect and literacy skills over the course of their Head Start year, these
on vocabulary test scores during elementary school and to relative gains were gone by the end of kindergarten; by the end
cause a child to be less likely to repeat a grade (Currie and of third grade, there remained only suggestive evidence of a
Thomas 1995; Deming 2009). While test score gains fade to a positive impact of Head Start on reading scores. Furthermore,
fraction of their initial levels by ages eleven to fourteen, there in no follow-up year did the Impact Study treatment students
is evidence that some Head Start participants are less likely to outperform the control students in math skills, grade
have ever been charged with a crime or to be a teenage parent, retention, or teacher reports of student behavior (Puma et al.
22 Policies to Address Poverty in America
Elizabeth U. Cascio, Diane Whitmore Schanzenbach
FIGURE 1-4.
Framework for Considering the Impact of Preschool, Current Policy Context
Preschool today
(public and private)
Quality of learning environment
High-quality public program (Oklahoma, Georgia)
Home inputs
(no preschool)
Socioeconomic status
2012). While it is possible that the prior nonexperimental Head on programs in two states—Georgia and Oklahoma—that
Start research yielded upward-biased estimates, it may also be meet essentially all of the same standards as the Tennessee
the case that the continued growth in state-funded programs program but serve much higher shares of the four-year-old
and in maternal employment (and use of other nonparental population (see box 1-1).
child care) has diminished Head Start’s potential impact. In
other words, Head Start may not represent the same increase The introduction of a high-quality, universal preschool program
in the quality of a child’s environment today as it did in the is illustrated in figure 1-4 by the light green dashed line. In this
past when there were fewer preschool alternatives. Indeed, framework, enrolling in the high-quality public preschool
the majority (roughly 60 percent) of children in the Head improves the quality of the learning environment experienced
Start Impact Study control group attended some other formal by low–socioeconomic status children, albeit by less than
education or child-care setting (Puma et al. 2012).9 the full distance from no preschool, because many of these
children would be enrolled in some preschool program even
A recent experimental evaluation of the state-funded pre-K in the absence of the new, high-quality option. Yet for higher–
program in Tennessee—where preschool or center-based socioeconomic status children the improvement in learning
child-care participation rates at age four in the control group environment represented by the introduction of high-quality
were lower (27 percent) and program quality was higher—has preschool is smaller, and in some cases may even be negative. 11
yielded results that are slightly more positive.10 The Tennessee
program, which was primarily targeted toward youth from The empirical results of the high-quality programs in Oklahoma
low-income households, yielded higher scores for participants and Georgia line up well with the conceptual framework
on tests of literacy, language, and math at the end of the pre-K illustrated in figure 1-4. By comparing children just old enough
year; participants were rated by their kindergarten teachers as to enter preschool to those who just miss the entry age cutoff
being more ready for school (Lipsey et al. 2013a). While the (a regression discontinuity approach), studies have found that
difference in measured cognitive abilities of the treatment and the Oklahoma preschool program raises short-term test scores
control groups disappeared by the end of kindergarten, former (Gormley and Gayer 2005; Wong et al. 2008).12 Where reported,
pre-K participants were much less likely to have been retained effect sizes for disadvantaged students (minorities and low-
in kindergarten and had slightly stronger school attendance income children) are in the range of those found in the Tennessee
records subsequent to the pre-K year (Lipsey et al. 2013b). study (Gormley and Gayer 2005). Subsequent analyses find that
the positive impacts of the Georgia and Oklahoma preschool
As was the case with Head Start, the only evidence on programs on disadvantaged children are still measurable
longer-term outcomes of state-funded preschool programs is when the students reach fourth and eighth grades (Cascio and
nonexperimental. Much of this research has to date focused Schanzenbach 2013; Fitzpatrick 2008). Students in Georgia and
The Hamilton Project • Brookings 23
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 1: Expanding Preschool Access for Disadvantaged Children
Oklahoma who are more advantaged, however, do not display Unfortunately, though quality and access matter considerably
sustained test score improvements from access to high-quality, for the cost of operating a pre-K program, we have limited
universal preschool. policy evidence to address questions about their impacts
on potential benefits. For example, there is limited evidence
The lack of test score impacts for more-advantaged students in of short-term benefits from higher-access, lower-quality
Georgia and Oklahoma, and the similarity of initial impacts programs. Likewise, while the regression discontinuity
in these states and in Tennessee, suggest that a universal, design has now been applied in multiple states to estimate the
high-quality program may yield no academic gains above short-term cognitive impacts of preschool, and effect sizes do
and beyond a targeted one, though it comes at an additional not appear to be strongly related to quality (see Wong et al.
cost.13 Consequently, one might wonder what the optimal mix 2008), state-specific estimates are somewhat uncertain, and
should be between quality and access. For example, could some states differ along other dimensions—most importantly in
of the gains from high-quality targeted programs, like that in terms of how nonparticipants spent the year in the absence
Tennessee, be achieved for disadvantaged students at a similar of preschool.
cost in higher-access, lower-quality programs, such as through
positive spillovers from the presence of higher-income children?
BOX 1-1.
Case Study on Universal Pre-Kindergarten in Georgia and in Oklahoma
Georgia was the first state to offer free pre-K for all four-year-olds. Georgia’s program, which began in fall 1995, is funded
by state lottery proceeds and serves the four-year-old population through a combination of half-day and full-day programs
operated out of both public schools and private centers. In fall 1998 Oklahoma became the second state to offer universal
public pre-K. Oklahoma’s pre-K program differs from Georgia’s in several respects: it is funded through not just state, but
also local and federal tax revenues; it operates almost exclusively out of public schools; and it serves a higher share of the
four-year-old population (74 percent to Georgia’s 59 percent, according to the most recent estimates). These differences aside,
both programs meet most common quality benchmarks, scoring high (8 or 9) on the National Institute for Early Education
Research scale (figure 1-3).
There is a growing body of evidence on the impacts of these programs on children’s readiness for kindergarten. When tested
at age five, children who attended Oklahoma pre-K for a full academic year outperformed their counterparts who just missed
being able to attend the program given their birthdays (Gormley and Gayer 2005; Wong et al. 2008). Comparable estimates
do not yet exist for Georgia, but Fitzpatrick (2008) found that cohorts of children eligible to attend Georgia’s pre-K program
(those aged four in fall 1995 and later) performed better on tests in fourth grade than did ineligible cohorts, both in absolute
terms and relative to cohorts of children aged four before and after fall 1995 in other states. However, the positive impacts
of the Georgia program on fourth-grade test scores were confined to disadvantaged children. Using a similar approach to
estimate the test score impacts of both the Georgia and Oklahoma programs, Cascio and Schanzenbach (2013) similarly find
a positive impact on fourth-grade scores for children from lower-income families. They also find a positive impact on eighth-
grade test scores for lower-income children, but it is smaller than the impact on fourth-grade scores.
The apparent successes of the Georgia and Oklahoma programs in improving children’s school readiness have fueled
recent calls for government-funded preschool expansion. However, we think that research findings do not necessarily
support universal programs in all scenarios. The impacts on test scores are largest for economically disadvantaged children,
particularly in later grades. This pattern of findings is sensible given that children from higher-income families will have more
and better options for school enrollment at age four (figure 1-4). Indeed, evidence suggests that for every ten children from
higher-income families who enrolled in the Georgia or Oklahoma programs, four or five would otherwise have been enrolled
in a private preschool program. There is little research evidence to suggest that children from higher-income families or the
families themselves benefit in any way beyond saving on child-care expenses (Cascio and Schanzenbach 2013). While worthy,
the goal of reducing the child-care costs for middle-class families could potentially be achieved in a lower-cost way. Thus,
given the policy evidence, only if state or local budget conditions permit would we recommend consideration of a widely
accessible program, and even then we urge policymakers to learn as much as possible about the alternatives to the proposed
program for any newly targeted children.
24 Policies to Address Poverty in America
Elizabeth U. Cascio, Diane Whitmore Schanzenbach
THE ROLE OF SUBSTITUTION that rates individual programs within a state along a variety of
It is challenging to design a state preschool program—even dimensions, most of them having to do with input measures.
one targeted toward low-income children—that does not While such measures are only rough proxies for the classroom
induce a lot of switching from another preschool to the public environment, they do provide important information to
program. The largest impact per unit cost comes from moving families deciding among various preschool options.
low-income children from attending no preschool to attending A 2013 review of the evidence by a panel of experts for the
some preschool. Many low-income children would otherwise Society for Research in Child Development concluded that
attend another program such as Head Start or center-based the most important aspects of quality in preschool education
care; the additional educational impact of attending a high- are stimulating and supportive interactions between teachers
quality state preschool program will be more muted for these and students, and effective use of a developmentally focused,
children. intensive curriculum (Yoshikawa et al. 2013). There are
As a high-quality program becomes less targeted toward promising methods to identify the programs and classrooms
low-income children and enrolls more middle-income that perform well on these measures, such as classroom
children, the share of new enrollees who otherwise would observations using the Classroom Assessment Scoring System
have attended preschool grows. The additional educational (CLASS), which measures the degree to which teachers interact
impact of switching from a high-quality, private preschool to with their students in a manner that stimulates learning in an
a high-quality, public preschool is likely to be close to zero. emotionally supportive environment. Recent work by Sabol
The number of switchers and the cost of the program can and colleagues (2013) has shown that preschool classroom
be limited somewhat by charging tuition to higher-income observations of the interactions between teachers and students
families who enroll in the program. using CLASS are more predictive of test-score gains than are
other inputs measures such as teacher education or class size.
ENSURING HIGH QUALITY A drawback of this approach is that it is relatively costly to
implement.
While “high quality” is a concept easily understood in theory,
it is more difficult to measure and enforce in practice. One way
to judge a state’s overall preschool quality is to use the criteria
established by the National Institute for Early Education
A New Approach
Research (NIEER); NIEER measures how many of ten Since the impact of preschool expansions hinges on both the
benchmarks regarding the level of inputs are met by a state’s level of quality of the preschool program and on how much
preschool policy. (This index is represented on the vertical preschool improves the quality the child’s experience relative
axis of figure 1-2.) There are drawbacks to this approach to what the child would be doing otherwise, policymakers
because these benchmarks are only rough proxies for the must carefully consider the existing context in order to design
classroom practices that are thought to make a high-quality and implement an effective preschool program.
program. For example, a state’s policy meets two benchmarks
if it has a class-size cap of twenty and a maximum student– NO PROGRAM: START A HIGH-QUALITY,
teacher ratio of 10:1. The state policy meets three more TARGETED PROGRAM
benchmarks based on the training level of teachers: one if the In states where there is currently no public preschool, the
head teacher is required to hold a bachelor’s degree, a second evidence suggests a targeted high-quality program may yield
if the teacher is required to have specialized pre-K training, a strong return. Therefore, a better investment may be in a
and a third if assistant teachers are required to hold at least smaller, higher-quality program rather than in a larger, low-
an Associate degree in child development. While on average quality program, especially if there are substantial numbers
these characteristics may be positively associated with higher- of low-income children who are not currently enrolled in
quality programs, they are not necessarily the causal pathway a preschool program. If substantial numbers of children
to a high-quality classroom experience. are already enrolled in Head Start, switching into a higher-
quality state program may still improve children’s educational
For example, it may not actually improve preschool classrooms
outcomes. Though we expect these gains to be lower for Head
to replace teachers who have no bachelor’s degrees but years
Start children than for children who would not otherwise have
of experience with teachers who have bachelor’s degrees but
attended any preschool, there is evidence that Head Start has
no experience. Thus, changing a policy to meet the NIEER
shifted its emphasis toward children ages three and under
benchmark may not actually result in an improved classroom
as state-funded preschool programs have expanded (Bassok
experience for preschool children. In a similar spirit, many
2012). Some children newly enrolled in state programs may
states have adopted Quality Rating and Improvement Systems
The Hamilton Project • Brookings 25
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 1: Expanding Preschool Access for Disadvantaged Children
then be attending Head Start at age three instead of age four, WINNERS AND LOSERS
and thus be receiving two years of government-funded early As described above, the largest gains will be expected when
education instead of just one. For states with strong Head Start low-income children are moved from no preschool to a high-
programs, it would be useful to work closely with the existing quality preschool. From an academic perspective, the gains
Head Start program to ensure the highest possible return on will be expected to be smaller (or even zero) for higher-income
the overall public investment. children who switch into a public program from a comparable
EXISTING LOWER-QUALITY PROGRAM: private preschool program. Nonetheless, for reasons that
IMPROVE QUALITY include the importance of peer interaction and political
popularity, the best policy may be a universal program.
In states with programs that score poorly on quality
measures—such as California, Florida, Ohio, and Texas— The social benefits to enhancing public preschool options may
the best plan may be to increase the quality of the program far outweigh the costs of investing in both the expansion costs
before expanding access to more students. Adopting state and quality improvements. For example, beyond any academic
standards in line with the NIEER quality benchmarks may benefits the available evidence suggests that high-quality
be a first step to increasing quality. For example, requiring preschool can have longer-term benefits for society through
head teachers to have a bachelor’s degree, providing health reductions in crime, teenage pregnancy, and dependence
screening and referrals, introducing site visits to monitor on public assistance. Narrowing the early educational gap
quality, and requiring a student-teacher ratio of 10:1 or higher between low-income and higher-income children is an
are all markers of quality used by NIEER. It is important to important step toward reducing income inequality over time.
note that these quality benchmarks are only rough proxies
for the learning environment experienced by the child. As a
result, meeting more of the NIEER quality benchmarks may Questions and Concerns
not substantially improve a child’s classroom experience.
Should we take money away from Head Start to invest in
In other words, meeting the benchmarks might be
state preschool instead?
necessary—but perhaps not sufficient—to achieve a high-
quality program. Another promising approach would be to The existing evidence on preschool is all drawn relative to
ensure that preschools have implemented a developmentally a baseline with the existing Head Start program. We don’t
focused, intensive curriculum with integrated, in-classroom know whether the impacts would be similar if resources were
professional development as recommended in the Society for shifted from Head Start to state programs. State expansions of
Research in Child Development report (Yoshikawa et al. 2013). preschool programs would be better combined with a national
effort to improve the effectiveness of Head Start. Gordon and
EXISTING HIGHER-QUALITY PROGRAM:
Mead (2014) outline policies to improve Head Start.
EXPAND ACCESS
In states with existing high-quality programs that reach What are the benefits of expanding the program to three-
only a small share of four-year-olds, efforts should focus year-olds?
on expanding access to the programs. It is important to
There are several reasons to think that expanding a preschool
understand that while the state’s cost of expanding access is
program to the most disadvantaged three-year-olds would
the same for all children, the potential educational impacts
have a larger impact on learning than expanding a preschool
of the expansion will depend on what the newly enrolled
program to the more advantaged four-year-olds for the same
children would have been doing otherwise. To the extent
price. For example, the strong results found in the Perry
that new enrollees are moving from lower-quality Head
Preschool program described in the text were from a two-
Start programs, day care, or no preschool, the impacts would
year intervention starting at age three. The Head Start Impact
be expected to be larger. However, we would expect the
Study finds positive short-term impacts on achievement levels
education impacts on new enrollees switching from high-
of three-year-olds; similar to the impact findings for four-
quality private preschools to be more muted. Some of this
year-olds, however, the impacts are substantially diminished
substitution (and cost) could be offset by charging tuition to
by third grade.
higher-income families. Nonetheless, there are documented
benefits of program expansion even when a high fraction of
children switch from private to public preschool. For example,
public preschool expansions decrease families’ out-of-pocket
spending on child care.
26 Policies to Address Poverty in America
Elizabeth U. Cascio, Diane Whitmore Schanzenbach
Do we get a large gain from expanding from a half-day to a
full-day program?
Conclusion
By the time they reach kindergarten, disadvantaged children
There is limited evidence on the impact of moving from a already show an achievement gap relative to their higher-
half- to a full-day preschool program. There appears to be income peers. In an attempt to level the playing field for low-
a persistent, positive impact of full-day preschool; because income youth, some have called on policymakers to invest in
evidence of that impact is largely drawn from an intervention early childhood education by expanding high-quality preschool
that also increased the length of the school year, it needs to access to a greater number of American families. Indeed,
be interpreted with some caution (Robin, Frede, and Barnett research has shown that expanding access to high-quality
2006). There is also evidence from the Head Start Impact preschool programs can be a cost-effective way to narrow the
Study that full-day programs have a larger impact on cognitive achievement gap and help low-income children build skills.
skills than half-day programs (Walters 2014). Experts in this However, the impact of the program depends critically on a
area caution that the impact does not come from additional child’s education in the absence of the intervention. Children
time alone, but stress the importance of ensuring that the with ample developmental and educational support—such as
curriculum and instruction is aligned to make the most of the those enrolled in private preschool programs—will benefit
extra time. far less from expanded access to preschool relative to those
Besides academic achievement, what other areas are without access to high-quality preschool.
affected by preschool programs? In this policy memo we provide guidelines for policymakers
There are a variety of outcomes that have been shown to be seeking to maximize the impact of investment in early
positively impacted by preschool. Children have had better childhood education. Our framework calls for the
school outcomes across a variety of dimensions: higher establishment of a high-quality program in areas where
attendance rates, fewer failing grades, less grade retention, programs do not exist, improved preschool quality in those
a higher likelihood of graduating from high school, and less states and localities with subpar programs, and expanded
involvement in criminal activity. Improvements in these access in areas where high-quality programs already exist. The
areas account for many of the economic benefits of preschool available evidence suggests that expansion of early education
programs. programs along these lines will lead to improved educational
outcomes for disadvantaged children, in addition to a host
Do we expect a large impact on mothers’ employment? of other social benefits such as lower crime, reduced teenage
pregnancy, and a lessened reliance on the social safety net.
Another benefit to free preschool that is often mentioned
is that it may enable more mothers to become employed by
reducing the opportunity cost to working. Nonetheless, the
best estimates are that this impact will be relatively small.
For example, if free preschool reduces the cost of child care
by around $5,000 per year, and if a mother with a high school
diploma or less would earn about $25,000 per year, then
preschool reduces the cost of working by about 20 percent.
Based on labor supply estimates, this would imply a relatively
modest 0.8 to 1.6 percentage-point increase in labor supply.
The Hamilton Project • Brookings 27
28 Policies to Address Poverty in America
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 2: Addressing the Parenting Divide
to Promote Early Childhood Development
for Disadvantaged Children
Ariel Kalil
University of Chicago
Introduction Kalil, Ryan, and Corey (2012) further show that highly educated
parents not only spend more time with their children than do
Growing income inequality over the past three decades has less-educated parents, but that they spend that time differently.
created a social divide with stagnated incomes for families at Specifically, highly educated mothers shift the composition of
the bottom of the distribution and sharply increased earnings their time as their child grows in ways that adapt to different
for those at the top (Atkinson, Piketty, and Saez 2011). As the developmental stages. When children are in preschool, for
economic destinies of affluent and poor American families example, college-educated mothers focus their time on
have diverged, so too has the educational performance of the reading and problem solving. This is precisely when time spent
children in these families (Reardon 2011). Socioeconomic in learning activities best prepares children for school entry.
gaps in children’s cognition and behavior open up early in During the middle school years, college-educated parents
life and remain largely constant through the school years shift their attention to the management of their children’s
(Duncan and Magnuson 2011). However, rising inequality in lives outside the home—precisely the ages when parental
income is not the sole cause of the divergence in children’s management is a key, developmentally appropriate input. Non-
achievement and behavior (Duncan et al. 2013). Parents do college-educated parents do not match their time investments
more than spend money on children’s development—they to children’s developmental stages in this fashion. Indeed, based
also promote child development by spending time with their on mothers’ patterns of time use across a variety of activities,
children in cognitively enriching activities and by providing researchers now posit that highly educated parents, more so
emotional support and consistent discipline. The “parenting than less-educated parents, view time with their children as
divide” between economically advantaged and disadvantaged an investment behavior with which to increase their children’s
children is large and appears to be growing over time along future human capital (Guryan, Hurst, and Kearney 2008). As
these dimensions (Altintas 2012; Hurst 2010; Reeves and highly educated parents increasingly adopt these patterns of
Howard 2013). investing in their children, the destinies of the children of
college-educated parents may diverge even farther from those
Consider the parenting time divide between economically of their less-advantaged peers.
advantaged and disadvantaged households. National time
diaries show that mothers with a college education or greater The disparities in parental time investment are important
spend roughly 4.5 more hours each week directly interacting because time with children is shown to have direct and causal
with their children than do mothers with a high school effects on children’s cognitive test scores (Villena-Rodán and
diploma or less (Guryan, Hurst, and Kearney 2008). This Ríos-Aguilar 2011). Price (2010) finds that an additional year
relationship is especially noteworthy because higher-educated of daily mother–child reading increases children’s reading test
parents also spend more time working outside the home. scores in the early school grades by 41 percent of a standard
The Hamilton Project • Brookings 29
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 2: Addressing the Parenting Divide to Promote Early Childhood Development for Disadvantaged Children
deviation from average. By comparison, the Perry Preschool Parents are children’s first teachers and, to equalize the playing
program, which is widely upheld as a model, has effect sizes on field, governments need to invest in parents so that they can
arithmetic achievement at age fourteen equal to 34 percent of better invest in their children. Gaps in children’s skills could
a standard deviation, but at a cost of $20,500 (in 2013 dollars) be narrowed if less-advantaged parents adopted the parenting
for each participant (Schweinhart et al. 2005). practices of their more-advantaged peers, and many parenting
interventions aim to do just that. Unfortunately, large-scale
To bridge the parenting divide and improve the life chances parenting interventions in the United States yield modest
of economically disadvantaged young children, I propose results at best and do not often change children’s cognitive
that the Administration for Children and Families (ACF) at or behavioral skills in the long run (Furstenberg 2011). An
the Department of Health and Human Services be tasked evidence and innovation agenda that helps policymakers
with building the knowledge base to create an Early Years identify and invest in what works is crucial for supporting
Family Policy to promote more-effective parenting and child parents’ engagement with their children.
development in low-income families, especially for children
from birth to age five. One leading family intervention for low-income children—
the Nurse-Family Partnership program—is being targeted
This policy memo outlines action steps that the ACF can take for substantial expansion by the federal government. The
to develop an evidence and innovation agenda to support program provides weekly in-home visits by trained nurses
parents in helping their children reach their full potential. to low-income, first-time mothers, starting before the child
In particular, I argue for three major evidence-based is born and continuing through the child’s second birthday.
innovations: (a) increasing participation in existing programs One mission of the program is to improve children’s
so that they can achieve their intended goals, (b) modifying or health and development by helping young, economically
adapting existing programs to test new approaches that might disadvantaged mothers provide more-competent care. Some
be more cost-effective and/or cost less, and (c) developing experimental evaluations of the program show that it reduces
new interventions that are lower cost and better matched to child maltreatment. In one study, mothers who received nurse
families’ needs. For all three innovations, I advocate building visits during their pregnancy and the child’s infancy had only
on new knowledge from the field of behavioral science, given 0.29 substantiated reports of child abuse and neglect at some
its potential for helping identify ways of changing behavior point before the child’s fifteenth birthday. Mothers who did
that are more cost-effective. These efforts have the potential not receive nurse visits, in contrast, had on average 0.54 such
to deliver smarter, more-innovative, and more-accountable reports (Olds et al. 1997). These results are noteworthy because
programs for children and families. This commitment child maltreatment is costly not only for the individuals
necessarily demands experimentation and testing with affected, but also for society (Zaveri, Burwick, and Maher
an eye toward developing new Early Years Family Policy 2014). The program also yields long-run benefits for some
interventions that can be offered cost-effectively and at scale. children. By age nineteen, females in the treatment group had
fewer arrests and convictions; a subset of these young women
had fewer children and less Medicaid use than their control
The Challenge group counterparts (Eckenrode et al. 2010). Despite the notable
CURRENT POLICY APPROACHES AND LIMITATIONS impacts, there were no overall long-term treatment effects
on high school graduation; economic productivity; number
At present, social policy for fostering the skills of young of sexual partners, use of birth control, and teen pregnancy
children largely focuses on education intervention by or childbearing; and use of welfare, Supplemental Nutrition
improving young children’s access to preschool programs Assistance Program (SNAP), or Medicaid. Furthermore, there
and increasing the quality of their primary and secondary were no long-run impacts of the program on males (Eckenrode
schools. Model early childhood intervention programs and et al. 2010).
other school-based efforts can narrow the gap between low-
income children and their middle-class counterparts (Chetty In short, this touted program appears to have made
et al. 2011; Currie 2001; Deming 2009). However, even though only modest improvements in parenting and the home
such interventions have demonstrated long-term success environment. When the children were about preschool age,
(albeit for the relatively few children who have participated the experimental evaluation revealed no overall treatment
in them), family background remains an important correlate differences in the HOME Inventory score (a measure of the
of children’s educational achievement and attainment (Bailey cognitive stimulation and emotional support provided to
and Dynarski 2011; Belley and Lochner 2007; Reardon 2011). the child in the home). Even among the small subsample of
highly disadvantaged mothers, the impacts on parenting were
30 Policies to Address Poverty in America
Ariel Kalil
modest and for the most part not statistically significant (Olds, OBSTACLES TO PROGRAM PARTICIPATION:
Henderson, and Kitzman 1994). Results from other large-scale INSIGHTS FROM BEHAVIORAL SCIENCE
randomized trials evaluating the impact of early intervention Perspectives from behavioral economics show that basic
programs designed to promote positive parenting and more- human psychology often puts up roadblocks on the path
enriched home environments (e.g., Parents as Teachers) have between expressed intentions and actual behavior (Fudenberg
also shown few statistically significant effects for low-income and Levine 2006; Laibson 1997; Thaler 1991). Optimal
families (Wagner, Spiker, and Linn 2002). Yet the average behavior requires self-control. When surveyed about weight
cost to serve a family for forty-five weeks in a home-visiting loss or low savings rates, for example, many individuals report
program is about $6,500; the Nurse-Family Partnership that they would like to lose weight or save more but lack the
program is on average even more expensive and can cost up to willpower to do so (Thaler and Benartzi 2004). Parenting
almost $14,000 for each parent participant (Zaveri, Burwick, offers many examples of often difficult and sometimes even
and Maher 2014). unpleasant demands whose rewards are uncertain and for
Unfortunately, many large-scale parenting interventions have which the payoff may not be enjoyed until many years later.
limited impacts, in part because of high rates of attrition, It is also difficult to change habits that have been developed
low take-up, and lack of engagement. In some home-visiting and reinforced over time: parenting behaviors are correlated
programs, more than half of enrolled families drop out across generations and shaped by the beliefs and preferences
early, with attrition rates generally ranging from 35 percent of influential relatives and neighbors in our social networks
to 50 percent (Wagner, Spiker, and Linn 2002). Early Head (Duncan et al. 2005). Successful parenting programs will
Start, another major early childhood intervention program, require unlearning a set of parenting practices and beliefs
also lacks strong participation (Love et al. 2005). Designed that may be deeply rooted in one’s family origin, culture, and
to provide child care from birth through age three, Early community (Wagner, Spiker, and Linn 2002). Rowe (2008), for
Head Start delivers home visits, parenting education, and example, reports evidence that low-income parents, compared
family support. An experimental evaluation of the program to their higher-income counterparts, respond less often to
showed that almost half of the families left the program before their young children’s utterances, based in part on their beliefs
their child was thirty months old, and more than one-third that adults cannot “make” babies talk.
dropped out before they had been enrolled for eighteen of the
thirty-six months. Only 16 percent of the sample participated Parents tend to want what is best for their children, but many
for the full duration of the program. Assessments of Early parents are not getting the most out of the programs they are
Head Start’s qualitative dimensions were no better: program participating in, either because they are not participating fully
administrators rated only 37 percent of families in the full in the programs or because the programs are not giving parents
sample as consistently “highly engaged,” rated 32 percent as the tools they need for optimal parent–child interactions.
“inconsistently engaged,” and rated 25 percent as “engaged at Programs should help interested parents make decisions that
a low level” or “not at all” (ACF 2002). These problems stand in are aligned with their professed intentions and goals. This
the way of long-term behavior changes for low-income parents would involve the redesign of programs and services to help
and their children. parents get the most out of what these programs are offering.
It should be noted that it is custom for most large-scale impact The challenge is to figure out how to make use of these
evaluations to measure effects on all children who were offered insights effectively to improve programs and policies.
a space to participate in the program (known as an “intent- Fortunately, there is compelling experimental evidence on
to-treat” measure). When intent-to-treat results are converted this point from interventions designed to promote health
to effects for children who actually participated (known as and financial behavior. In these arenas, programs designed
“treatment-on-the-treated”), early childhood intervention on principles from behavioral science have proven effective
programs appear to have larger effects. For example, Ludwig for weight loss, smoking cessation, financial savings, and
and Phillips (2008) find that the benefits to Head Start are health behavior, among other outcomes (for examples see
substantially higher when the intent-to-treat results are Ashraf, Karlan, and Yin 2006; Charness and Gneezy 2005;
converted to treatment-on-the-treated results. Little is known Kamenica 2012; Milkman et al. 2011; and Stockwell et al.
about the effects for participants in a program like Early Head 2012). Elements common to many of these interventions
Start who completed at least, say, half of the program. This include commitment devices, which work by formalizing a
underscores an emphasis going forward on increasing take-up pledge to do something or achieve an objective; incentives,
rates and engagement. which work by offering financial or nonfinancial rewards or
recognition for changing behavior; and planning prompts,
The Hamilton Project • Brookings 31
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 2: Addressing the Parenting Divide to Promote Early Childhood Development for Disadvantaged Children
which provide reminders designed to overcome problems of processes. Although conventional wisdom attributes lack
forgetfulness and procrastination. Many of the ideas in the of participation and engagement to parents’ stress and
behavioral economics toolkit are low cost, light touch, and complicated lives, as Wagner, Spiker, and Linn (2002) argue,
highly scalable. To date, however, these insights have had little there are few empirical data to support these assumptions.
impact on the way we design parenting interventions.
We should find a way to deliver parenting programs effectively
Cognitive behavioral science offers a complementary despite parents’ challenging life circumstances. If not, we
perspective on parent engagement by highlighting the problem will continue to produce apologetic reports documenting
of cognitive scarcity among low-income parents stemming disappointingly weak effects and will eventually lose the
from their past and current exposure to toxic stress (Mani et political and public will to spend taxpayer dollars on such
al. 2013). One potentially important source of income-based efforts.
differences in parenting is the repercussions of the daily
stressors of low-income parents’ lives that place cognitive and On the second front, we need to design and experiment with
emotional demands on parents’ attention and self-control. new strategies for making parenting programs more efficient
These financial strains leave little room to follow through and more effective, drawing on new insights from behavioral
on decisions that can affect their children’s future (Mani et science. Specifically, we need to draw on insights that lead
al. 2013; Mullainathan and Shafir 2013; Shah, Mullainathan, to promising new avenues to improve take-up, retention,
and Shafir 2012). Accordingly, the possibilities for purposeful, engagement, and impact of early childhood, parenting, and
goal-directed parenting are greatly diminished. related public health interventions.
Some promising new approaches are focused on parents’ The main barrier to scaling-up parenting interventions
executive function skills, key components of which include nationwide is the currently limited understanding of the
impulse control, working memory, and mental flexibility. key ingredients of successful programs. Public support for
Experiences of trauma and stress make focus, memory, and government-funded home visiting programs is weak (The
mindful attention and decision-making difficult (Shonkoff Pew Charitable Trusts 2014), and efforts in this arena are
2012). Although experimental evidence is currently lacking, hampered by the idea that family policy is an intrusion in the
some promising programs for low-income parents are using private sphere of family life. We do not debate, however, that
coaching, multimedia, and computer games that have been children should have regular vision and hearing screenings
specifically designed to create ways for adults to improve throughout their school years. But unlike receiving a regular
memory, focus, attention, impulse control, organization, schedule of such screenings, we have no consensus about what
problem-solving, and multitasking skills (Babcock 2014). families should be required to do to help children achieve
Mindfulness meditation training, mind–body exercises (e.g., their full potential. Moreover, whereas hearing screenings
relaxation breathing practice), and brain games are tools are considered the best way to identify hearing deficiencies in
that may increase the quality of parent–child interactions, order to prevent or minimize effects on educational progress,
and likely better mental health and health outcomes to boot we do not have a screening to identify risk factors or effective
(Davidson et al. 2003). parenting behavior to prevent children from, say, dropping
out of high school.
EARLY YEARS FAMILY POLICY AGENDA
A New Approach
I propose that the President of the United States task an
This policy memo proposes that policymakers become agency, most likely the ACF at the Department of Health and
better informed on effective interventions that can motivate Human Services, with filling knowledge gaps that impede the
and support parents to do the things that parent–child development of an Early Years Family Policy agenda.
programs are intended to encourage. Although the lack of
participation and engagement has long vexed researchers and An Early Years Family Policy agenda at the ACF should be
program administrators, the standard model for parenting consistent with the evidence and innovation agenda proposed
interventions has changed little over time. To achieve last year by the Office of Management and Budget (OMB)
success and scale-up, and to be cost-effective, we need to in the Executive Office of the President (OMB 2013). The
make progress on two related fronts. First, we need to better chief component of this effort is strengthening agencies’
understand parental motivation to participate in programs. abilities to continually improve program performance by
Attrition and engagement require explicit empirical attention; applying existing evidence about what works, generating new
programs should be designed in a way that can model these knowledge, and using experimentation and innovation to test
new approaches to program delivery.
32 Policies to Address Poverty in America
Ariel Kalil
DESIGN AND EVALUATION OF EVIDENCE-BASED I next offer some examples—also summarized in table 2-1—of
PROGRAMS the kinds of research trials and evaluations of new approaches
Specifically, the ACF should design and evaluate rigorous to changing parent behavior that the ACF should help fund,
experiments, using randomized control trials where possible, design, and evaluate.
to test the efficacy of new interventions and design refinements Home-visiting programs
to existing interventions. Given the evidence outlined in
this paper, research findings from the social and behavioral An experiment that my colleagues at the University of Chicago
sciences can be harnessed to implement low-cost approaches and I are currently designing will test a behaviorally informed
to improving program results. The goal should be to develop intervention intended to increase the frequency with which
new interventions (or adaptations to existing interventions) low-income parents engage in educational play with their
that use the cutting-edge tools of behavioral economics and children. This study will randomly assign about 500 parents of
new insights from neuroscience that guide current thinking preschool-age children to a treatment and control condition.
about executive function and mindfulness. The treatment combines information about the importance
of educational playtime, a commitment to spend the time,
The centerpiece of this proposal is a new research competition recognition for spending the time, and planning prompts.
sponsored by the ACF at a level of $10 million annually for Parents in the treatment and control group will be given
five years. With these funds, the ACF will hold peer-reviewed electronic tablets to take home for six weeks; these tablets
competitions to select grantees who are willing to embed will be preloaded with educational apps and games, and will
innovative randomized control trials into existing programs. record the amount of time parents spend using them with
This approach avoids reinventing the wheel, and focuses their preschool-age children. The experiment will test whether
instead on innovations in program design and delivery that the suite of behaviorally informed nudges and incentives
increase parental engagement and impact. In addition, I significantly increases the time parents in the treatment group
propose that $1 million of the competition funds each year spend with their children. This is the first study of its kind that
be targeted to developing new interventions that are lower we know of, and thus there is great scope for funding similar
cost and better matched to families’ needs. Priority for these types of studies with different parents or caregivers in the low-
funds each year should be targeted to grantees proposing the income population.
use of affordable technology as a tool to promote parental
engagement and participation in programs. (I expand on this A second example highlights innovations in home visiting
idea below.) with a program being developed and evaluated by Bierman
and colleagues (2013). This study is testing the REDI Parenting
To facilitate the efforts of this new evidence and innovation program, a home-visiting program designed to complement
agenda, and for relatively minimal cost, agencies can form the Head Start classroom program by enhancing the school
partnerships with academic experts, including using externally readiness of economically disadvantaged preschoolers.
funded Intergovernmental Personnel Act assignments, to Each month parents receive a REDI activity club box at the
receive conceptual advice on cutting-edge research findings home visit, containing learning materials for them to use,
that should inform how policies are designed, and to receive books for them to read, and games for them to play with
technical support on designing, evaluating, and iterating their children. The books have explicit questions embedded
experimental field studies. to support parents’ interactive book reading; this element of
the intervention draws from behavioral insights. That is, the
Upon the successful completion of these activities, the ACF program removes the seemingly trivial barriers to engaging
can make recommendations for expanding efforts with a parents in this type of parent–child interaction by devising
proven track record, identify gaps in knowledge, and design questions and prompting children to respond. Evidence
a roadmap to achieve new knowledge. These efforts not suggests that providing this home-visiting intervention has led
only would elevate attention to parenting and the home to sustained effects through third grade. In contrast, impacts
environment, but also would create a plan for coordination faded out for children who participated in the classroom
with efforts to expand preschool opportunities for low-income without the home-based intervention.
children. Following this plan of action will help to ensure
that children arrive at preschool as prepared for learning Technology-based initiatives
as possible, and will increase the chance that the quality of
The ACF should also prioritize the design and evaluation
parenting and the home environment are sufficiently strong to
of new strategies that make use of affordable technology as
prevent fade-out of high-quality preschool experiences.
a tool to promote parental engagement and participation in
programs. Advances in technology not only could address
The Hamilton Project • Brookings 33
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 2: Addressing the Parenting Divide to Promote Early Childhood Development for Disadvantaged Children
TABLE 2-1.
Summary of New Parenting Interventions
Program Type of program Intervention Sample description Results
Educational Play Behaviorally informed Provides electronic 500 parents of To be determined
Intervention, University intervention tablets loaded with preschool-age children
of Chicago, ongoing educational apps and
games that record
amount of time parents
spend using them with
pre–K children
Head Start REDI Home-visiting Provides twice-monthly 356 four-year-old Sustained impacts
Program, The program designed to home visits in pre–K children in 44 Head on vocabulary, literacy
Pennsylvania State complement Head Start and kindergarten Start classrooms skills, and
University, 2003 social behaviors
through third grade
InfantNet, Lane County Web-based parenting Provides mothers of 40 infants and their Positive impacts on
Oregon, 2006–2008 intervention and remote infants with computer, mothers with income at parental mental health
coaching program webcam, Internet or below 185 percent and children’s social
connection, and of U.S. Poverty Income behavior; parents
technical/training Guidelines used over 90 percent
support for six months of material
Momba, Yale University, Interactive web-based Provides low-income First-time, low-income To be determined
Connecticut, ongoing smartphone application mothers with access mothers
to social network of
pregnant and new
mothers
Sources: Baggett et al. 2010; Bierman et al. 2013; Seger 2012.
barriers to effectiveness, but also could open up new avenues touch and lower-cost than the traditional in-person service
for programs to make an impact. Given the ever-decreasing delivery model, may be suitable for many families.
costs of hardware and the low marginal costs of software,
using technology to improve on existing approaches, as well Prototypes of such programs have begun to emerge from the
as to develop new approaches, is a promising strategy from research world. For example, Baggett and colleagues (2010)
a cost–benefit perspective. One example of such an approach created InfantNet—a Web-based parenting intervention and
might be an interactive parenting coaching program that remote coaching program for low-income single mothers of
mimics home-visiting programs. To envision the potential infants—which was originally designed to provide parent
merits of such a novel approach, consider the idea that many support services to families in rural areas. The program
parenting interventions rely on a model where one delivery provided mothers of infants with a computer, webcam,
method fits all, and that these interventions require a serious Internet connection, and technical training/support for
commitment of time. A technology-based approach in which six months. In a pilot sample of forty caregivers, mothers
educational materials were preloaded on a digital device completed eleven online sessions that included modeling
or were downloadable from the Internet could reduce a videotapes, computerized videotaping of actual parent–
program’s dependence on home visits. Parents would not have infant interactions, and weekly phone calls with a coach who
to depend on face-to-face meetings to stay current with the monitored the parents’ use of the materials and reviewed
program and, provided they have access to the Internet, could the parent–infant interaction video in consultation with the
make use of social media platforms to develop partnerships parent. The results suggested that parents used more than 90
with other parents. Such an approach, which is both lighter- percent of the materials and found them useful and easy to
34 Policies to Address Poverty in America
Ariel Kalil
understand. The intervention also had a positive impact on Visiting, and the Nurse-Family Partnership. However, for the
parental mental health and children’s social behavior. reasons I have outlined in this proposal, it seems wise in an
era of scarce government resources to devote some funding
As another relevant example, researchers at the Yale Child to develop and evaluate new approaches that are potentially
Study Center are in the process of creating an interactive more cost-effective to improve parenting and promote child
Web-based smartphone application modeled after successful development, rather than focusing evaluation and knowledge-
social networking tools. The app will create a virtual network building efforts exclusively on status quo approaches.
of first-time low-income mothers to connect them to one
another, mental health services, and parenting support; it will
also incorporate rewards for participation (Seger 2012). Questions and Concerns
These nascent efforts are incorporating insights from What programs besides home-visiting programs would
behavioral science and advances in technology (and benefit from behavioral insights and technology?
sometimes both). They have shown promising results, albeit
almost exclusively at the pilot or proof-of-concept stage, and The emphasis in this proposal has been on changing parenting
merit more testing and investment. behavior, and this naturally lends itself to a discussion of
home-visiting programs. The insights from innovative
COSTS AND BENEFITS approaches to research and evaluation can be applied to any
Researchers have estimated that some parent-training program that interacts with parents. For instance, key goals
programs delivered by home visitors return $1.80 for every of the Head Start preschool program are to engage parents in
$1.00 invested, especially for the highest-risk families (Aos et the classroom and to conduct outreach to improve parental
al. 2004). They are nevertheless costly. It seems reasonable to support of children’s learning at home. These parent-directed
expect at least that great of a return on investment if existing efforts could be enhanced with new knowledge from the R&D
programs can be made more efficient and cost-effective, or if efforts proposed here. New knowledge from behaviorally
new programs can be designed with the same goal. To support informed or technology-enhanced efforts could also be
this effort, the ACF should prioritize high-quality, low- applied in child welfare programs, Head Start, Early Head
cost evaluations and rapid, iterative experimentation. Such Start, and early intervention.
approaches can follow the lead of those in the private sector
Would the behavioral insight-informed approaches for
that use frequent, low-cost experimentation to test strategies
parents also enlighten the work of other early childhood
to improve results and return on investment.
care providers?
To put the proposed $10 million in annual research and
The emphasis in this proposal has been on parents, and I
development spending in context, it is useful to compare it to
have argued that this approach is necessary to improve the
the commitment President Obama has made to expand home
life chances of low-income children. But this proposal may
visitation to additional low-income children. Specifically, the
not be sufficient. Young children are exposed to multiple
Affordable Care Act of 2010 included $1.5 billion over five
types of nonparental caregivers and teachers. There is
years for states to operate the Maternal, Infant, and Early
every reason to think that behavioral insight–informed
Childhood Home Visiting program. The administration’s
approaches could yield important positive benefits for other
proposed fiscal year 2014 budget adds $6 million to the $400
early childhood caregivers. For instance, teachers in early
million allocation for that program. It also proposes that
childhood education programs serving low-income children
Congress ensure the continuation of the program beyond 2014
often suffer from stress and job burnout, in part due to the
by investing $15 billion in funding for the program from 2015
challenges of dealing with the stress and trauma experienced
through 2025.
by the children under their care. Tools that make the job
The Department of Health and Human Services is spending of these caregivers easier, whether based on technology or
additional funds on a five-year national evaluation of the a mindfulness intervention, and that help teachers focus,
Maternal, Infant, and Early Childhood Home Visiting problem-solve, and multitask, hold great potential for
program, as mandated by the Affordable Care Act improving caregivers’ efforts and interactions with young
(Michalopolous et al. 2013). The national evaluation study is a children. For example, Landry and colleagues (2009) show
large-scale (with a sample size of 5,000), in-depth, expensive, how technology and its capability for providing immediate
multiyear effort. It will yield results on the short-term personalized feedback significantly improves teachers’
impact on family outcomes of four different types of existing ability to plan their behavior and makes their interactions
home-visiting programs, including Early Head Start–Home with preschool children more efficient and effective.
The Hamilton Project • Brookings 35
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 2: Addressing the Parenting Divide to Promote Early Childhood Development for Disadvantaged Children
“Light-touch, low-cost” sounds good in theory, but how are
your innovative programs going to serve families where
Conclusion
parents suffer from serious mental health problems or are In sum, the United States has made little progress toward
otherwise in extremely stressful circumstances? narrowing the achievement gap between advantaged and
disadvantaged children. Parenting interventions have had
Some parents will always need intensive services and will limited success, in large part because participation retention
require an ongoing personal relationship with a home visitor and/or the quality of engagement in such programs is low.
or social worker. However, there is another group that does not I propose the development of an evidence and innovation
need or desire such an intense relationship. The problem right agenda to support parents to meet their goals of helping
now is that we don’t have a very good estimate of how large children reach their full potential. New knowledge from the
either of these two groups is or what their preferences are for field of behavioral science has great potential for helping
the different ways in which they could interact with programs. identify ways of changing behavior that are more cost-
Moreover, most existing programs take a one-size-fits-all effective. The challenge is to figure out how to make use of
approach, which is likely inefficient for both groups of parents. these insights effectively to improve programs and policies
Innovation in program design and delivery is likely to yield for low-income parents and children. The ACF should devote
benefits to a broad share of the targeted parent population. substantial additional resources to creating and promoting
In absence of a federally funded intervention, is there an Early Years Family Policy agenda focused on new and
anything that community groups can do to bridge the improved ways to support parenting and child development
parenting divide? in low-income families with young children. Such an agenda
has the potential to deliver smarter, more-innovative, and
Yes. Research that builds more-useful evidence can and more-accountable programs for children and families. This
should occur at multiple levels—from federal down to local commitment necessarily demands experimentation and
efforts. Local programs are often more nimble and flexible and testing with an eye toward developing new interventions that
thus could potentially more easily move toward the behavioral can be offered cost-effectively and at scale.
science–informed experimentation approach I have outlined
here. Owing to this flexibility, community organizations may
also be well positioned to adopt a framework of continuous
quality improvement. In addition, experimentation at the local
level is critical for understanding how program innovations
interact with local contexts, specific populations, and different
types of practitioners.
36 Policies to Address Poverty in America
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 3: Reducing Unintended Pregnancies
for Low-Income Women
Isabel Sawhill
The Brookings Institution
Joanna Venator
The Brookings Institution
Introduction U.S. Department of Health and Human Services’ Office of
Population Affairs (OPA) use Title X monies to fund states
Children born to young, unmarried mothers in the United for the purpose of launching a social marketing campaign to
States face an elevated risk of poverty. More than half of educate women about the safety, effectiveness, and convenience
births last year to women under the age of thirty were outside of long-acting reversible contraceptives, or LARCs. These state-
of marriage. In 2012, single mothers headed nearly 25 percent run campaigns would target the population of women most
of families, compared to 13 percent in 1970 (U.S. Census vulnerable to births outside of marriage: low-income women
Bureau 2012). In that same year, 47 percent of children living between the ages of fifteen and thirty.
in single-mother families lived below the federal poverty level,
more than four times the 11 percent poverty rate for children
living with their married parents (U.S. Census Bureau 2013). The Challenge
Children of single mothers fare less well in school and in
THE GROWTH OF SINGLE-PARENT FAMILIES
life than children of married parents (see McLanahan and
Sandefur 1994; Waldfogel, Craigie, and Brooks-Gunn 2010). Since about 1980, the growth of single-parent families has been
For these reasons, addressing the situation into which driven almost entirely by an increase in childbearing outside
children are born needs to be a key component in our nation’s of marriage, often the result of people sliding into relationships
fight against poverty. and having an unplanned baby.1 As seen in figure 3-1, this
growth has been concentrated among less-educated women.
Most single mothers claim that their pregnancy was unwanted
or mistimed. Because births to unmarried mothers are largely The result is a growing class divide in family-formation
unintended births, we believe that the most realistic approach patterns. Combined with growing gaps in income and in
to slowing the growth of single-parent families is to help education, this widening divide in family structure threatens
women delay childbearing until both parents are ready to social mobility (Sawhill 2012; Sawhill and Venator 2014).
raise a child and prepared to make a long-term commitment
to the other parent. Doing so will improve child well-being Pregnancies and births to unmarried women are largely
and reduce child poverty rates. unplanned. Approximately half of all pregnancies in the United
States are reported by the mother as unintended, and that
To that end, we propose a social marketing campaign number increases to 70 percent among single women under
designed to improve knowledge and attitudes about ways to thirty (Zolna and Lindberg 2012).2 Unintended pregnancy
prevent unintended pregnancies so that women can make rates are highest for women that are the least economically
better-informed decisions. Specifically, we propose that the advantaged, as seen in figure 3-2. In particular, unintended
The Hamilton Project • Brookings 37
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 3: Reducing Unintended Pregnancies for Low-Income Women
FIGURE 3-1.
Percent of Births to Unmarried Mothers by Education, 1970–2012
80
Percent of births that are to unmarried mothers
70
60
50
40
30
20
10
0
1970 1976 1982 1988 1994 2000 2006 2012
Less than a HS diploma HS diploma Some college College degree or more
Sources: Centers for Disease Control and Prevention 2014; authors’ calculations.
pregnancy rates for poor women (women with incomes at or become teen parents, and are less likely to graduate from high
below 100 percent of the federal poverty level) and low-income school or college (McLanahan and Sandefur 1994; Waldfogel,
women (women with incomes between 100 percent and 199 Craigie, and Brooks-Gunn 2010).
percent of the federal poverty level) are more than triple the
rate for women with incomes at or above 200 percent of the Because unintended births are concentrated among low-
federal poverty level. income unmarried women, reducing the number of these
pregnancies would decrease the number of children born to
DELAYING CHILDBIRTH AS AN ANTIPOVERTY poor single mothers. A recent paper, based on a simulation
STRATEGY with a variety of data sources, suggests that eliminating all
unwanted (but not mistimed) births would lower the share of
Delaying births is no guarantee that poverty will be reduced.
children born into poverty by 2 percentage points and increase
As noted above, most of the increase in unwed childbearing
the percentage of children born to college-educated mothers
is occurring among less-educated women. Given their
by 4 percentage points (Karpilow et al. 2013).
disadvantages, they might be poor regardless of whether
or not they postponed childbearing.3 For this reason, it is
important to combine our proposal with measures to improve
the educational and labor-market opportunities of less- A New Approach
advantaged women. But we believe that delaying pregnancy is If a large proportion of less-advantaged young adults are
a crucial step toward improvements in child well-being and in having children as the result of unplanned pregnancies, then
lowered child poverty rates. one way to reduce child poverty is to prevent unintended
pregnancies and births. Encouraging more young women
Children born to young, unmarried mothers are more
to use effective forms of birth control, especially LARCs,
likely to fare worse on many dimensions, including school
can help accomplish that goal. The first step in this process
achievement, social and emotional development, health, and
is to increase awareness among young women about the
success in the labor market. These children are at greater risk of
availability, convenience, safety, and effectiveness of these
parental abuse and neglect (especially from live-in boyfriends
contraceptive devices through a social marketing campaign.
who are not the children’s biological fathers), are more likely to
38 Policies to Address Poverty in America
Isabel Sawhill, Joanna Venator
FIGURE 3-2.
Unintended Pregnancy Rates among Unmarried Women in their Twenties
250
Number of unintended pregnancies
200
150
per 1,000 women
100
50
0
White African Hispanic Did not Graduated Had some Less than 100%–199% 200%
American graduate high college or 100% or more
high school school or more
equivalent
Race/Ethnicity Educational attainment Income relative to federal
poverty level (percent)
Source: Zolna and Lindberg 2012.
To be effective, this initiative must be combined with efforts to patch were twenty times more likely to get pregnant than were
ensure that health providers are well-informed and prepared women who used a LARC (Secura et al. 2010). A LARC is
to provide LARCs, and that there are fewer barriers to roughly forty times more effective than a condom. The greater
affordable health care. More community health centers and effectiveness of LARCs compared to condoms or the Pill has
the expansion of Medicaid to all states as called for in the less to do with their ability to prevent a pregnancy—assuming
Affordable Care Act would help to ensure that providers could full compliance with a method—and much more to do with
accommodate the demands of a social marketing campaign. the fact that they change the default from being protected only
The Affordable Care Act—with its contraceptive mandate, when the method is used consistently and correctly, to always
subsidized premiums, Medicaid expansion, and investment being protected, regardless of what the user does. They are
in community health centers—has the potential to transform also easy to use and reversible. Once a woman and her partner
the health-care landscape. However, there will likely be decide that they want a baby, they can choose to remove the
some groups left uncovered and gaps in coverage for others, device with a quick return to the clinic.
especially in states that have so far rejected the Medicaid
expansion. In the meantime, our proposal deals with a THE ROLE OF SOCIAL MARKETING CAMPAIGNS
problem that will exist regardless of any successful expansion Health behaviors—particularly risky ones like smoking,
of health-insurance coverage. unhealthy eating, or unprotected sex—are influenced by social
norms and individual motivation. Social marketing campaigns
THE EFFECTIVENESS OF LARCS
identify these norms and the behaviors that need to be changed,
The class of contraceptive devices referred to as LARCs and create messages tailored to reach those people engaging in
includes implants and intrauterine devices (IUDs).4 These risky behaviors. An effective, well-communicated message can
have very low failure rates (<1 percent), far lower than the influence behavior in a positive way.
two most commonly used forms of contraception: condoms
(18 percent) and the Pill (9 percent). According to a study in Campaigns focused on health behavior have proved effective
the St. Louis area that gave women free contraception and in the past. For example, the American Legacy Foundation’s
counseling on the efficacy of different contraceptive methods, Truth campaign, aimed at reducing smoking among teens,
women who used the Pill, a transdermal ring, or a hormonal has been credited with changing attitudes about tobacco and
The Hamilton Project • Brookings 39
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 3: Reducing Unintended Pregnancies for Low-Income Women
BOX 3-1.
Prevention First Colorado
Colorado implemented the Prevention First Colorado campaign in the Denver area in 2009. The first part of the campaign
involved placing health educators in a few clinics in Denver who are responsible for contraceptive education, counseling,
and patient follow-up. Doctors in these clinics typically have limited time to spend with patients so these educators allow
for more one-on-one time for patients and more-extensive follow-up to reduce the number of patients who use birth control
inconsistently. Health educators also automatically sign up patients who are starting a new contraceptive method for a three-
month follow-up appointment in order to help them maintain consistency in use. The second part of the campaign is a public
education effort, which uses direct mail, bus ads, posters, Web sites, print ads, brochures, and community presentations about
the benefits and availability of contraception. The Prevention First Colorado campaign specifically focuses on encouraging
the use of the LARCs and uses messages like, “Life is full of surprises, pregnancy shouldn’t be one of them” to encourage
young women to go to clinics run by Women’s Health. As this campaign is still under way, evaluations of the effectiveness
of the campaign are not yet available.
reducing the number of teens who smoke by 22 percent over not using contraception given by women in a government
three years (Farrelly et al. 2005). Campaigns about sexual survey was, “I didn’t think I could get pregnant” (Mosher
behaviors have been less common and, until recently, have and Jones 2010). Other evidence suggests that many young
typically focused on condom use and HIV awareness. On people who have had unprotected sex and not gotten pregnant
average, these campaigns increased positive sexual behaviors infer (incorrectly) that they cannot or will not get pregnant
among the target population (e.g., men using a condom) by as from subsequent sexual encounters (Frohwirth, Moore, and
much as 6 percentage points (Evans, Silber-Ashley, and Gard Maniaci 2013). Focus group research in Colorado further
2007; Sawhill, Thomas, and Monea 2010). While 6 percent may suggests that many women are in denial about the risks of
sound small, given the broad reach of such campaigns, their pregnancy (Prevention First Colorado 2009).
cost-effectiveness is high. One approach of social marketing
campaigns is to embed messages in popular television shows. The second objective is to educate young women on
An analysis of MTV’s 16 and Pregnant suggests that the contraceptive options and dispel myths surrounding
message broadcast by the show (that is, the difficult reality of contraception, especially with regard to LARCs. Despite their
becoming a teen mother) led to roughly a 6 percent reduction effectiveness, only about 9 percent of women on contraception
in teen births between June 2009 and the end of 2010 (Kearney use IUDs (Finer, Jerman, and Kavanaugh 2012). Among
and Levine 2014). sexually active women aged twenty to twenty-four, about 3
percent use IUDs as their primary form of contraception, 27
A social marketing campaign targeting unintended percent use the Pill, 7 percent use another hormonal method
pregnancy would aim to produce continuous protection (e.g., patch, injectable, or contraceptive ring), and 15 percent
against pregnancy (through LARCs) since the main cause of rely on condoms; 42 percent of sexually active women in this
unintended pregnancies, almost as important as nonuse, is age group report using no contraception (Jones, Mosher, and
inconsistent use. More than half (52 percent) of unintended Daniels 2012).
pregnancies are due to nonuse of contraception, 43 percent are
due to inconsistent or incorrect use, and only 5 percent are due Young women also seem to lack knowledge about the range
to method failure (Gold et al. 2009). of birth control options available to them. One-fourth of
young adults have never heard of IUDs and more than half
Within the goal of encouraging more-consistent use of have never heard of the implant (Kaye, Suellentrop, and Sloup
contraception, the campaign would be designed around four 2009). Even when LARCs are readily available, women do not
objectives, drawing in part on lessons learned from past or always take advantage of them because of spurious concerns
ongoing campaigns in Colorado and Iowa with similar goals about side effects spread through word of mouth. For example,
(see boxes 3-1 and 3-2). a third of young adults still mistakenly believe that IUDs often
cause infections, partially because of the continued fallout
The first objective is to educate young women about the risks from Dalkon Shield’s faulty design in the 1970s (ibid.).
of pregnancy and to motivate them to protect against an
unplanned pregnancy. The most commonly cited reason for
40 Policies to Address Poverty in America
Isabel Sawhill, Joanna Venator
BOX 3-2.
Avoid the Stork
The Avoid the Stork campaign in Iowa, launched in early 2010, targeted women ages eighteen to thirty through television
ads, billboards, print and Web ads, college events, and giveaway promotions. The campaign used humor and created a brand
around the concept of avoiding unintended pregnancy: the mascot was a large, awkward stork who would interrupt a person’s
life to represent the consequences of a pregnancy. Development of the campaign took approximately a year, including time
to pilot test the ads among a subsample of college students. By the end of the campaign in 2011, over 70 percent of surveyed
women reported having seen or heard of the campaign; Iowa has seen a 4 percentage point decline in unintended pregnancies
between 2009 and 2011. (This box is based on Pederson 2012.)5
However, the latest research suggests that LARCs are safe for to talk to trained professionals about birth control. These
women of all ages, including adolescents and both pre- and trained professionals would advise women on all aspects of
post-childbearing women (Espey and Ogburn 2011; Peterson sexual health, including the importance of continued use of
and Curtis 2005; Tolaymat and Kaunitz 2007). Some women condoms and regular STD testing.
experience negative side effects, such as perforation and
infection; the likelihood of those two issues arising from an A fourth objective of the campaign is to make sure that once
IUD today, however, is less than 0.1 percent (Hubacher et a woman is motivated to use a LARC, she will be able to easily
al. 2001; Stoddard, McNicholas, and Peipert 2011). Implants find a clinic or health-care provider who has a supply of LARCs
have similarly been found to be efficacious and safe (Darney on hand and whose staff is trained to provide the appropriate
et al. 2009). Changing the message about contraception to counseling and care. Unfortunately, many physicians are
encompass more than just condom use or the Pill is important, not up-to-date or trained in how to provide LARCs to their
and campaigns in Colorado and Iowa have already started patients (Dehlendrof et al. 2010; Harper et al. 2008; Madden
to enlighten young women through social marketing and et al. 2010). Both the Colorado and Iowa campaigns provided
educational counseling. training to all clinic staff, and not just to physicians. The
University of California, San Francisco Bixby Center for
The third objective is to convince women that LARCs are not Global Reproductive Health is conducting a major study (2014)
just safe and effective, but also a low-maintenance and hassle- testing the effects of improved training for family-planning
free form of contraception, well-suited to women with busy clinicians on access to and use of LARCs. Their randomized
lives. The primary problem for some women is not access to trial has been underway since 2008 and the results are not
contraception, but rather their ability to use it consistently—to yet available, but concern about provider knowledge and
always use a condom in the heat of the moment, to remember training is widespread among those in the field. Although the
to take a Pill, or to get their prescription refilled so that there focus here is on the social marketing campaign, we strongly
are no gaps in protection. When asked why they were not recommend that any campaign be combined with efforts to
using contraception, many women who had an unintended make sure that providers are well-prepared when clients show
pregnancy reply, “I simply wasn’t thinking” (Edin et al. 2007). up. Expanding on this effort in detail is outside the scope of
Focus group research in Colorado showed that many women the current proposal.
often simply forget to take the Pill (Prevention First Colorado
2009). A social marketing campaign needs to persuade women IMPLEMENTATION DETAILS
that LARCs are the “no worry” and “no hassle” way to ensure For states looking to follow the models set by Iowa and
that they are effectively protected against an unplanned Colorado, the first step would be to secure funding for a social
pregnancy. marketing campaign. Both of these programs were created
through a private–public partnership, but past campaigns
It should be noted that these campaigns would not be
(such as the Don’t Kid Yourself campaign in the 1990s; see
advocating that women use LARCs as their sole method
box 3-3) have been federally funded under Title X (Weinreich
of birth control. Rather, the campaigns would emphasize
1999). We propose that the OPA set aside $100 million per
LARCs’ efficacy in reducing pregnancy while also counselling
year ($500 million over five years) under Title X specifically
that they do not protect against sexually transmitted
for states that intend to create social marketing campaigns to
diseases (STDs). One of the benefits of the campaign model
combat unintended pregnancy. Public–private partnerships
we are proposing is that it encourages women to go to clinics
would be encouraged as well. The deputy assistant secretary
The Hamilton Project • Brookings 41
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 3: Reducing Unintended Pregnancies for Low-Income Women
BOX 3-3.
Don’t Kid Yourself
In 1996, six states (Colorado, Montana, North Dakota, South Dakota, Utah, and Wyoming) implemented Don’t Kid Yourself,
a campaign with the goal of reducing unintended pregnancy among low-income women between the ages of eighteen and
twenty-four. They used radio ads, newspaper ads, posters, and drink coasters in bars, clubs, and coffee shops to spread messages
encouraging the use of birth control and providing information about how to get birth control at family-planning clinics.
Their pilot program in two cities was a success, but when they expanded regionally to fifty-five cities in all six states, there
was a much lower exposure rate—only 15 percent of the target population reported being exposed to the campaign. However,
the message was somewhat successful among those it reached. Three-fourths of those exposed initiated conversations with
significant others about birth control due to the campaign, and more than 55 percent of those exposed reported calling a
family planning clinic for more information. (This box is based on Weinreich 1999.)
for Population Affairs would award funds on a competitive events. Colorado used coasters in bars. Both campaigns also
basis, with eligibility criteria adapted from current Title X used television ads, billboards, print ads, and mailings.
guidelines. These criteria include the size and needs of the
community, the number of low-income women served by a We propose that the federal government work with a private
grant, the capacity of the applicant to carry out their proposal consulting firm or nongovernmental organization to develop
given community resources and staffing, adequacy of the the default brand and message for the campaign; we believe
applicant’s implementation plan given past research, and that providing this information to all grant applicants would
the relative availability of nonfederal resources within the be helpful in avoiding reinvention of the wheel each time.
community to be served. Some degree of market segmentation Iowa’s Avoid the Stork campaign worked with Worldwide
might be allowed, involving different target groups and Social Marketing to develop three different concepts that were
different messages, depending on what more-detailed research then tested with a subsample of their target demographic. They
showed about the needs in a particular state or area of the eventually settled on a humorous brand with a memorable
country. However, the OPA would provide a template based on mascot, but other campaigns, such as Don’t Kid Yourself
its research and the advice of a major marketing firm on the and Prevention First Colorado, used a more straightforward
best messages to use. This template could serve as the default message about the consequences of a surprise pregnancy.
in each case, but states could request deviation from the plan Clear metrics of success should be established in evaluating
based on their specific needs. the campaign; one such requirement for funding would be
During the first year of this initiative, the OPA would issue the willingness to submit to an independent evaluation of the
requests for campaign proposals from state governments. campaign’s success. Many past campaigns have focused on
States applying for grants would be encouraged to consult exposure to the ads, and not on changes in attitudes toward
widely with various stakeholders in the state and to evaluate LARCs or changes in behavior, such as the number of unintended
the specific needs of their state or region through surveys pregnancies or births averted. Some campaigns, such as the
and focus groups among the target population. For example, multistate intervention Don’t Kid Yourself in the 1990s, had very
Colorado conducted four focus groups and forty individual, poor exposure rates; however, Don’t Kid Yourself had positive
private interviews with low-income women between the effects on behavior among the 15 percent of the population that
ages of eighteen and twenty-four to better understand the it reached. Important metrics to evaluate are exposure to the
perceived barriers to consistent contraceptive use among that campaign, the number of women who switched contraception
specific population. Iowa conducted three statewide surveys, methods as a result of exposure to the campaign, the number
multiple focus groups, and in-depth interviews around the of women who contacted clinics advertised through the
state to understand how people viewed the issues surrounding campaign, attitudes toward LARCs, the number of pregnancies,
contraception. Focus groups not only would help explore the number of unintended pregnancies and/or pregnancies that
barriers to contraceptive use, but also would help to evaluate occurred among unmarried couples, the number of users of
the ways in which the target population gets information. For specific contraception methods, and the number of abortions
example, Iowa targeted community colleges because they were before, during, and after the campaign. Future campaigns can
able to draw large concentrations of twenty-somethings to learn from past campaigns’ successes and failures only if the
42 Policies to Address Poverty in America
Isabel Sawhill, Joanna Venator
evidence clearly relays who the campaigns reach and how they being able to stay in school or finding a stable partner before
affect those they reach. These metrics should be collected on a having children (Lichter and Graefe 2001; Ng and Kaye 2012).
state-by-state basis at the six-month, one-year, three-year, and Furthermore, the benefits to the children of being born to
five-year marks to capture both short- and long-term effects older parents in more-stable relationships are large.
of the campaign and any differences based on implementation
across states.
Questions and Concerns
COSTS AND BENEFITS
Do social marketing campaigns really work?
Compared to other antipoverty programs, social marketing
campaigns are very cost-effective. In fact, most evidence Some do and some do not. It is important that any campaign
suggests that they save money. Consider a $100 million be well-funded and well-designed to achieve a set of specific
annual investment that reaches one-fourth of unmarried objectives. In addition, there needs to be local buy-in, which is
women between ages fifteen and thirty in this country. why we recommend that states must make an active decision
Assume that 5 percent of these women shift to a LARC each to apply for grants and that the OPA evaluate applications
year as a result of the campaign, half of them from using a based, in part, on whether the state has sought and obtained
condom and half from using no contraception. The resulting local buy-in. In addition, the campaign will not be effective
reduction in unintended pregnancy each year would be unless funding for all forms of FDA-approved contraception is
about 160,000 averted pregnancies. Of the 40 percent (or available following the implementation of the Affordable Care
67,000) of unintended pregnancies carried to term, about Act in the states, and unless providers are trained to provide
half of these births (approximately 34,000) are to women all forms of contraception. With these caveats, as noted in the
living below the poverty line. Monea and Thomas (2010) text, campaigns can change the behavior of perhaps 5 percent
estimate a total taxpayer savings of $24,000 for each averted of the target population and avert a large number of births to
birth to a poor or low-income woman. Of the 34,000 averted poor women.
births in this scenario, about 10,500 would not occur at all,
resulting in savings of $253 million per year; the remaining Won’t these women be disadvantaged and their babies poor
births would be delayed on average by two years, resulting in no matter when they give birth?
additional savings of $280 million per year.6 This means that
By permitting women to complete more education, to gain
the savings to taxpayers would be over $500 million per year,
more work experience, and to form a stable two-parent family,
yielding a cost–benefit ratio of about five to one. If we loosen
the odds that any child will be born into poverty are reduced.
our assumptions to include all births to women eligible for
Moreover, women who defer childbearing until they want to
Medicaid-covered pregnancy costs (i.e., women below 200
be parents are likely to access more prenatal care, to be better
percent of the federal poverty level) rather than just births
parents, and to create better life prospects for the child.
to poor women (i.e., women below 100 percent of the federal
poverty level), the cost–benefit ratio increases to eight to one. Do these women who say they are having unintended
pregnancies really mean it?
Previous studies of costs and benefits have shown a similar
benefit-to-cost ratio for taxpayers. For example, Thomas (2012) Unintendedness is a continuum. There is no bright line
finds that a social marketing campaign costing $100 million between a birth that is planned and one that is unplanned.
per year will result in approximately a 4 percent reduction in Some women (and their partners) are clearly ambivalent or
unintended pregnancies, roughly a 2 percent reduction in the simply do not plan at all. That said, the only hard data we have
number of children born into poverty, and savings of $431 suggest that rates of unintended pregnancy are very high,
million to taxpayers per year. The taxpayer-savings figure especially among poor women. A large number will abort the
includes not only reduced Medicaid payouts for prenatal and pregnancy. On the other hand, the fact that so many say the
pregnancy care, but also an estimate of the cost to taxpayers pregnancy was unintended—and that mothers say this even
of publicly subsidized benefits (e.g., through the Temporary after they have bonded with their newborn infant—tends to
Assistance for Needy Families and the Earned Income Tax bias answers to this question downward, not upward.
Credit) for the children until the age of five (Monea and
Thomas 2011). Is it politically realistic for the government to fund a social
marketing campaign in such a contentious arena?
These calculations count only the public benefits of reducing
unintended pregnancies. There would be additional benefits Contraception is a politically contentious issue. Prior efforts to
for a mother of delaying a birth until she is ready, such as increase access to contraception have been met, at times, with
The Hamilton Project • Brookings 43
PROMOTING EARLY CHILDHOOD DEVELOPMENT
Proposal 3: Reducing Unintended Pregnancies for Low-Income Women
substantial political opposition; in some cases this opposition
has successfully derailed public programs. Still, many
Conclusion
programs have been implemented despite this opposition. Children born to young, unmarried parents are much more
In particular, we note the success in implementing social likely to grow up in poverty than are those born to older and/
market campaigns in Iowa and Colorado—two states that or married parents. Many of these children are born to women
fall in the middle of the political spectrum. Thus, while we who did not intend to get pregnant, and who state that the
acknowledge that political sentiment is a formidable obstacle pregnancy was either unwanted or mistimed. Reducing the
to universal take-up of social marketing campaigns aimed at number of children born to these mothers would significantly
contraception use, the successes in Iowa and Colorado suggest reduce the number of children born into poverty. Creating
that this barrier will not prove insurmountable in a wide greater awareness of the risks of pregnancy and how to reduce
swath of states. that risk will help women match their childbearing behaviors to
their intentions and make it easier for women to delay pregnancy
until they can give their child a stronger start in life. All the
evidence suggests that this proposal to launch social marketing
campaigns would reduce unintended pregnancies and births,
reduce child poverty, and save the government money in the
process. Family planning by itself will not eliminate child
poverty, but it is an important step in the process.
44 Policies to Address Poverty in America
Section 2.
Supporting Disadvantaged Youth
The Hamilton Project • Brookings 45
46 Policies to Address Poverty in America
SUPPORTING DISADVANTAGED YOUTH
Proposal 4: Designing Effective Mentoring
Programs for Disadvantaged Youth
Phillip B. Levine
Wellesley College
Introduction I propose that these programs be implemented in accordance
with a set of best practices and be rigorously evaluated in order
The need for mentoring programs is indisputable. Over 30 to determine the key components for program success with the
percent of children live in households headed by a single parent goal of designing the best possible interventions for improving
(or no parent), a rate that has doubled over the past forty-five the life outcomes of disadvantaged youth.
years (see figure 4-1). Six in ten African American children
live in households of this type, which actually reflects a slight
decline in recent years; this rate has been as high as two-thirds. The Challenge
Estimates indicate that upwards of 9 million children have no
A wide variety of programs aim to pair disadvantaged youth
caring adults in their lives (Bruce and Bridgeland 2014; Cavell
with role models in one-on-one relationships in hopes of
et al. 2009). This policy memo reviews the evidence of success
providing these youth with advice and guidance that they
from past and current mentoring programs and proposes ways
may not otherwise have. As noted at the outset, there is an
to move forward that could truly make a difference in the lives
immense need for mentors in this country given the number of
of young people by providing them with opportunities that
children who lack proper adult guidance (about 9 million), but
could propel them forward in life.1
determining how to establish an effective mentoring program
Although there are 5,000 mentoring programs in this country is not entirely straightforward. A major obstacle to moving
providing services to 3 million young people (Dubois et al. forward is sorting through the breadth of research on past and
2011)—with Big Brothers Big Sisters alone serving almost existing programs. This proposal does so and addresses this
200,000 children (Big Brothers Big Sisters of America central question: What can we learn about existing mentoring
2012)—many youth remain unserved. Before we propose programs to help design or modify them so that they alleviate
expanding mentoring programs to more youth, it is critical poverty among young people?
that we identify existing programs and the components of
First, the specific focus of individual programs is important
those programs that work best. This paper will do that, and
to consider in evaluating past research. Mentoring programs
then, based on the best available evidence, will argue that
come in many forms, some of which may satisfy a variety
community-based mentoring programs in the vein of the
of different goals but do not address poverty reduction
traditional Big Brothers Big Sisters model are most effective.
specifically. For instance, some programs assign mentors
I contend that community-based programs should receive
to victims of child abuse, where the goal of the program is
additional support of nongovernmental organizations
to limit the emotional damage done to the child. This may
(NGOs)—including nonprofits, foundations, and charitable
indirectly enable the child to be more successful in the labor
organizations—as well as private-sector entities. Moreover,
The Hamilton Project • Brookings 47
SUPPORTING DISADVANTAGED YOUTH
Proposal 4: Designing Effective Mentoring Programs for Disadvantaged Youth
FIGURE 4-1.
Percent of Children Living in Households Headed by a Single Parent or No Parent,
1968–2013
80
70
60
Percent of children
50
40
30
20
10
0
1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012
All children African American children
Sources: U.S. Census Bureau 2013; author’s calculations.
market, but that is not its specific focus. Other examples of Robert Lerman’s proposal in this series discusses these types
programs in this category include those that are directed at of programs in greater detail.
teen-pregnancy prevention, improvement of health status,
or reduction of recidivism among criminals. They may be One final restriction that I impose in examining previous
successful in their own dimensions and should be supported research is to focus on those evaluations that are conducted
accordingly, but my focus here is primarily on direct attempts within an experimental context. It is common in the literature
to improve economic well-being as at least one of the main to find examples of program evaluations that rely on what are
goals of the program. often labeled quasi-experiments. Typically, in these examples,
treatment and control groups are identified. The treatment
Second, my focus on alleviating poverty is a major filter in group represents those members who voluntarily participated
evaluating past evidence. The most direct way to improve in the program, however, and the control group is created
labor market success for a participant is to improve her as a set of other individuals who have similar demographic
educational outcomes. Several mentoring programs have that characteristics (e.g., age, race, family income). Selection bias
as an explicit goal, perhaps among many goals; these are the is an obvious concern in these evaluations: those who are
programs I consider. For instance, we have direct evidence that more motivated to succeed volunteer to participate, and this
children who get better grades, score higher on standardized differential level of motivation is not necessarily matched
tests, and are more likely to complete high school also do in the control group. These studies have an obvious bias in
better in the labor market. This policy memo concentrates on the direction of finding a positive effect of the program—a
those measures that can be directly translated into subsequent conclusion that may or may not be warranted. For this reason,
labor market success. I exclude these studies from my review.
Third, I do not consider training and career development The extensive resources provided by Child Trends, which
programs that include mentoring as just a minor aspect, such catalogs a large array of interventions with a multitude of
as Career Academies and Job Corps. The key component of program goals, are beneficial to this review (Child Trends
programs like these is vocational training. Mentoring services 2014). All of these interventions have been evaluated using
are included, but they are far from the focus of the programs. true experimental designs. In the Child Trends database,
48 Policies to Address Poverty in America
Phillip B. Levine
twenty-four interventions include some form of mentoring between ten and fourteen years old, largely economically
component, but most do not satisfy the four conditions disadvantaged, and almost exclusively living in single-parent
identified earlier.2 households. The results indicate that the youth who received
the mentoring treatment skipped school less often and felt
Various types of programs include a substantive mentoring more confident in their ability to complete schoolwork. Their
component with a key focus on improving educational grades also went up by 0.08 GPA points (on a 4-point scale).3
outcomes and subsequent labor market success. They can
be categorized in a number of dimensions, distinguishing The second evaluation run by Big Brothers Big Sisters
between those that are publicly or privately funded, those that addressed a school-based model of mentoring. In this
are school-based versus community-based, those that offer program, treatment-group youth, who were in Grades 4–9,
a comprehensive set of services that include mentoring, and were matched with volunteer mentors, and the pair met over
those that largely or exclusively focus on mentoring alone. the course of one school year, typically for one hour per week.
Most of these meetings ended when the school year came to a
The distinction between publicly and privately funded close. Academic support was often included in these meetings,
programs is obvious. School-based programs are those in but this was not the exclusive focus. Two-thirds of the students
which the mentor typically meets with the mentee after school; were receiving free or reduced-price lunch (indicating they
an important element of the meeting is frequently helping lived in lower-income households) and around half lived in
with school work. Because of the central nature of the school single-parent households. The results of this intervention
environment, these programs tend to meet less over the course were mixed. Some academic outcomes did improve, including
of a typical week and for fewer weeks per year dependent on the number of assignments completed and teacher ratings of
the school calendar and, particularly, with gaps during school overall academic performance. The impact on grades, however,
vacations and over the summer. Community-based programs was half the size of that in the community-based program
include longer meetings (perhaps on weekends) throughout (0.04 GPA points) and was not statistically significant.
the year and do not focus explicitly on academic support.
Two other school-based mentoring programs have been
Programs that offer more-comprehensive services along evaluated using an experimental design: the Student Mentoring
with mentoring can include aspects like financial incentives, Program (SMP; Bernstein et al. 2009), funded by the No Child
community service requirements, supplemental education, Left Behind Act, and the Study of Mentoring in the Learning
and the like. Given that mentoring is a sufficiently important Environment (SMILE; Karcher 2008). The design of both
component of these programs, I include these programs in programs included meetings between students and mentors
this review. for one hour per week over the course of the school year. In
practice, fewer meetings actually took place. SMP duration
In sum, my criteria narrow the focus to programs (1) that
are primarily targeted at improving economic outcomes, (2) averaged about one meeting per week over five or six months;
SMILE duration averaged only eight meetings over three
that include mentoring as a substantial component of the
months. In both programs meetings included discussions of
intervention, (3) that measure educational outcomes, and
academic activities, but were not limited to such discussions.
(4) that have been evaluated using an experimental design.
In SMP, most of the student participants were receiving free or
Evaluations are available for five past interventions that
reduced-price lunch, almost half were living in single-parent
satisfy these criteria. The features of these five programs are
households, and the majority were deemed academically at-
summarized below and in table 4-1.
risk. Most students in SMILE had family incomes under
Two of these evaluations were conducted by Big Brothers Big $20,000. The results from both programs were discouraging;
Sisters, which is the largest and best-known mentoring agency the interventions led to no significant improvement in any
in the country. It is a nonprofit organization that has been academic outcome. In attempting to reconcile the results from
matching volunteer mentors to disadvantaged youth for over the three student-based mentoring programs, Wheeler, Keller,
a century. More recently, it has conducted two evaluations and DuBois (2010) contend that the limited impact of SMP
of the programs that it runs. The first evaluation focused on and SMILE relative to that of Big Brothers Big Sisters (which
its community-based mentoring programs, which follow was not overwhelming in the first place) may be attributable
its original model (Tierney, Grossman, and Resch 1995). In to the fact that 17 percent of assigned student–mentor pairs
this evaluation, treatment group members were matched to never actually met in SMP and relatively few student–mentor
mentors who were members of the community, and the pair meetings took place in SMILE.
met a few times a month for an average of four hours per
meeting over the course of at least one year. The youth were
The Hamilton Project • Brookings 49
SUPPORTING DISADVANTAGED YOUTH
Proposal 4: Designing Effective Mentoring Programs for Disadvantaged Youth
TABLE 4-1.
Overview of Mentoring Programs Reviewed
Program Type of program Frequency Composition of Sample Impact on Cost per
and funding and duration sample size academic participant
of meetings outcomes per year (in
2013 dollars)
Big Brothers Community-based Two to four times Ages 10 to 14; 60% 959 Significant effects $1,530
Big Sisters mentoring per month for at boys; from single- on several
Community- (privately funded) least one year; parent, low-income measures,
Based Mentoring typical meeting households; with some including a 0.08
lasted four hours history of violence or increase in grade
substance abuse point average
(GPA)
Big Brothers School-based One-hour weekly Grades 4–9; 69% 1,139 Significant $1,177
Big Sisters mentoring meetings for one free or reduced-price effects on several
School-Based (privately funded) academic year lunch; close to 50/50 measures, such
Mentoring (under six months gender ratio; around as absenteeism
in practice) half in single-parent and assignments
households completed, but
no significant
effects on GPA
Department School-based One-hour weekly Grades 4–8; 85% free 2,573 No observable $1,522
of Education mentoring meetings for one or reduced-price lunch; impact
Student (publicly funded) academic year 44% in single-parent
Mentoring (under six months households; 60% at
Program (SMP) in practice) academic risk
Quantum Comprehensive Goal was 750 At-risk students 1,069 No observable $35,730
Opportunities program including hours/year, but entering Grade 9 impact
Program (QOP) substantive actual average was
mentoring 177 hours/year
component
(privately funded)
Study of School-based One-hour weekly Mainly Latino students 516 No observable No data
Mentoring in mentoring meetings for one between ten and impact available
the Learning (privately funded) academic year eighteen years of age;
Environment (under six months most with annual family
(SMILE) in practice) income under $20,000
Sources: Bernstein et al. 2009; Karcher 2008; Herrera et al. 2007; Schirm, Stuart, and McKie 2006; Tierney, Grossman, and Resch 1995; author’s calculations.
50 Policies to Address Poverty in America
Phillip B. Levine
Finally, the Quantum Opportunities Project provided based mentorship programs with a set of best practices in mind;
more-extensive services than the other programs, including it is useful to consider the components of those programs that
homework help, tutoring, life and family skills counseling would generate the greatest gains for program participants.
(including counseling on alcohol and drug abuse, sex, and One reason that community-based programs may have been
family planning), and a significant community service more successful than school-based programs is the nature
requirement, along with a substantive mentoring component and the extent of interaction between the mentor and mentee.
(Hahn, Leavitt, and Aaron 1994; Schirm, Stuart, and McKie These programs had more contact hours (typically three or
2006). In addition, students received financial incentives to four meetings per month lasting, on average, four hours per
encourage them to stay in the program. This program focused on meeting) over a longer period of time (about a year) than
at-risk students entering ninth grade. Of the five interventions school-based programs. This aspect likely contributed to its
reviewed here, the Quantum Opportunities Project is clearly success. School-based programs also focus directly, although
the most extensive, both in terms of services provided, program not exclusively, on academic support; community-based
length, and cost. An initial pilot of the intervention showed programs do not. Apparently, providing life guidance may be
positive results, including a 21 percent increase in high school more important than providing academic guidance.
graduation rates. The success of the pilot led to a larger-scale
evaluation, but the results could not be replicated, particularly One other aspect of program implementation that would
in terms of educational attainment. The follow-up study was likely be desirable is the demographic match between the
unable to find any effect in that dimension.4 mentor and mentee. Evidence from educational research and
evaluations of job-placement programs suggests that having
All of this evidence suggests that a traditional mentoring mentors that are of the same race and perhaps of the same
program of the community-based type, such as Big Brothers gender as the mentee is an important element of a successful
Big Sisters, is the approach most likely to be successful program (see Behncke, Frölich, and Lechner 2010; Dee 2004,
in improving subsequent labor market earnings among 2005). Interestingly, Big Brothers Big Sisters does not mandate
disadvantaged youth. School-based approaches have yielded matches by race, although it does by gender.
mixed results, at best. Several potential explanations could
explain this finding. First, their organization around the In terms of other program components, we do not have the
school imposes administrative hurdles that may lead to fewer luxury of additional experimental evidence to provide strong
and shorter meetings between mentors and mentees. Second, recommendations regarding the specific content that should
the emphasis on schoolwork, even if it is not exclusive, may be included in model mentoring programs. What we do have,
hinder the true benefit of a mentoring intervention, which is however, is the approach that Big Brothers Big Sisters used in
providing an adult voice of reason to adolescents who may its community-based programs that have been successfully
be lacking one. Conventional community-based approaches evaluated with positive results. Tierney, Grossman, and
also dominate a comprehensive approach that offers a number Resch (1995) document these program elements. I propose
of services, including a substantive mentoring component. that NGOs and private-sector entities consider the following
Perhaps it is no surprise based on the longevity of the program factors when promoting mentorship programs:
that Big Brothers Big Sisters is the type of intervention that 1. These programs should undertake thorough screening of
provides the clearest benefits to its participants. potential mentors. Tierney, Grossman, and Resch (1995)
report that Big Brothers Big Sisters uses background checks
to screen out those determined to “pose a safety risk, are
A New Approach unlikely to honor their time commitment or are unlikely to
I propose that NGOs and private-sector entities consider form positive relationships with the youth.” Only around
expanding mentoring programs of the community-based one-third of their volunteers met that test. Big Brothers
form. Having access to an adult, trusted voice of reason would Big Sisters rejected those whom it deemed inappropriate
likely be helpful to disadvantaged youth seeking to climb the and those who did not complete the necessary steps of the
economic ladder. Based on my discussion below regarding the screening process.
public and private returns to mentoring programs, I make the
2. Mentorship programs should undergo a thorough screening
case that NGOs and private-sector groups should promote
of potential mentees. Those adolescents who participate
these types of programs.
must be interviewed along with their (single) parent, pass
Beyond the general support for community-based mentoring a home assessment, receive parental permission, and have
programs, I propose that these groups implement community- a “minimal level of social skills” (Tierney, Grossman, and
Resch 1995).
The Hamilton Project • Brookings 51
SUPPORTING DISADVANTAGED YOUTH
Proposal 4: Designing Effective Mentoring Programs for Disadvantaged Youth
3. Extensive training of mentors is recommended, although A critical component of this analysis is the distinction between
it is not mandatory. The training should address youth returns to the program that are received by the participant
development, communication skills, and suggestions about (private returns) and those that are received by society more
how to interact with a mentee, among other priorities. broadly (social returns). If the private returns of a program are
greater than its costs, then the program is worth it in the sense
4. As mentioned earlier, matches between the mentor and that investing one dollar in the program is better than simply
the mentee should be made based on preferences and transferring one dollar to the participant. An NGO or private-
expediency. Gender, geographic proximity, and availability sector entity that intends to help disadvantaged youth would
are common match factors, along with the interests of both be better off investing in the program than simply giving away
the mentor and mentee. the money. If the social returns are greater than one dollar,
then the program is worth it to taxpayers because they actually
5. Finally, mentorship programs should include an element
profit from making the transfer; the program yields benefits to
of supervision of the mentor–mentee relationship. Case
them that are greater than the investment. In this case, the
managers should routinely check in with the mentor and the
public sector should be willing to invest in the program.
mentee in order to verify that the match has been successful.
Discussions about the value of supporting a public program
These five program components have not been separately
frequently focus on the social benefits. Programs that
evaluated with a rigorous methodology designed to determine
assist underprivileged populations satisfy this condition
their role in the success of the program. Nevertheless, they do
by increasing tax revenue, reducing expenditures for social
provide a starting point; their combination has been found
programs, and reducing crime. A perfect example is the
to be effective in Big Brothers Big Sisters community-based
Perry Preschool program, which Elizabeth Cascio and Diane
mentoring programs. NGOs and private-sector entities should
Schanzenbach discuss in their proposal in this series.
ideally combine and implement these elements in mentorship
programs for disadvantaged youth. Generating social benefits that are greater than the program’s
cost, though, is very hard to do. Even when we can increase the
Finally, given that these program components have not been
earnings of disadvantaged individuals, it is hard to increase
thoroughly evaluated, NGOs and private-sector entities
them enough to put them into the range of incomes where
interested in mentoring programs should support the most
tax receipts would be substantial. Typically, when we are able
rigorous possible experimental evaluation. For instance,
to provide strong evidence of generating social benefits in
evaluations should attempt to answer questions such as the
excess of program costs, the key component is a reduction in
following: Is the estimated impact reproducible in other
crime and incarceration. This was true in the Perry Preschool
settings? What screens should be used in the selection of
program. Those outcomes are so costly to society that relatively
mentors? How often and for how long should mentors and
modest effects can provide tremendous public savings.
mentees meet? What types of activities provide the greatest
benefit to the mentee? We cannot answer these questions It is difficult to determine whether traditional mentoring
based on the available evidence, but it would be valuable to programs reduce crime and incarceration. The outcomes most
have these answers, among many others, to be able to identify closely approximating criminal activity in the Big Brothers Big
the key components for program success and help design Sisters evaluation are “number of times stole something” and
the best possible intervention. Evaluation of implemented “number of times damaged property” (Tierney, Grossman,
programs would therefore be a critical aspect of continuing and Resch 1995). The treatment group was not statistically
and expanding these types of programs. significantly less likely to engage in either of these behaviors
(although the point estimates were negative). The outcome
COSTS AND BENEFITS
“number of times hit someone” did drop significantly, but its
Just because community-based mentoring programs appear relationship to crime is less clear. We do see that drug and
to be the best approach to implement, it would be premature alcohol use declined for participants in traditional mentoring
to judge these programs to be “worth it.” I argue that these programs, and it is possible that this would translate into
programs are worth expanding from the perspective of an reduced criminal activity subsequently, but that is a rather
NGO or private-sector group looking to improve outcomes substantial leap. In the end, it is possible that Big Brothers Big
for at-risk youth, but whether it is worth it for the government Sisters could pass a societal benefit–cost test, perhaps even
to financially support these programs is a higher hurdle that convincingly, but it is not clear that it could do so based on the
mentoring programs would be less likely to overcome. available evidence.
52 Policies to Address Poverty in America
Phillip B. Levine
This does not mean that it is not beneficial for the government programs with a set of best practices in mind. Should altruism
to invest in mentoring programs, but rather that the become a recognized goal of public policy, governmental
investment would need to be supported by another form support of these programs would be desirable as well.
of return. In particular, society may receive value simply
by helping the poor improve their outcomes from a purely
altruistic perspective. It makes us happier if individuals who Questions and Concerns
are having difficulty getting by have an easier time of it. Of
Have the previously conducted evaluations provided enough
course, providing a value to altruism to incorporate into a
guidance to inform the design and implementation of new
formal benefit–cost comparison is a difficult proposition. That
programs?
determination would have to result from the political process.
The simple answer to this question is that it is rare to have
For the private sector, however, altruism is the goal. The goal
enough evidence to be certain of all the best elements that
of the private sector is to spend its money wisely in a way that
should be incorporated into new programs. In this case,
yields the greatest impact. Again, that sector can always just
we have evaluations that enable us to rule out certain types
transfer money to targeted populations directly, so a program
of programs (like those that are school-based), and one
is only desirable if the private benefit the program generates
evaluation that provides strong support for advancing
in the form of higher incomes for its participants is greater
community-based programs. That evaluation was extensive,
than the dollar cost of providing these programs. To satisfy
but there are always limitations in going forward with new
this criterion means comparing private benefits to the cost
programs based on the results of a single experiment. Clearly,
of implementation. This is the form of benefit–cost analysis I
additional experimentation should be conducted to fill some
conduct here.
of the holes in our knowledge.
The good news for mentoring programs is that they easily
This may be an example where the best is the enemy of the good.
satisfy this test. Levine and Zimmerman (2010) provide details
By the standards of program evaluation, the Big Brothers Big
of the approach that lead to this conclusion, but I summarize
Sisters community-based program is an effective one and it
it here. The general idea is to obtain program effects in
should be emulated. Certainly, future experimentation should
terms of some form of educational outcome and then use a
continue to address these lingering questions and help inform
conversion factor that translates that educational outcome
into higher subsequent wages. In this case, we know from the subsequent program design, but based on what we know now,
Big Brothers Big Sisters community-based evaluation that the Big Brothers Big Sisters community-based program model
program participants experienced a 0.08 point improvement is an approach that is worthy of expansion.
in their GPAs. Levine and Zimmerman (2010) then used Is it possible to expand the scale of community-based
data from the 1979 National Longitudinal Survey of Youth mentoring programs like Big Brothers Big Sisters to address
to generate a conversion factor between GPA and wages. This the size of the adolescent population in need of those
wage effect is presumed to be constant over the remainder of services?
the individual’s life; the analysis then calculates the present
discounted value of this higher-earnings stream throughout The estimates I provided above suggest that millions of
the individual’s life. The results of this analysis indicate that adolescents could benefit from mentoring programs. Existing
Big Brothers Big Sisters generates about a $7,500 expected programs like Big Brothers Big Sisters do serve about 200,000
benefit relative to the program cost of about $1,600 (where individuals now, though, suggesting that it is possible to run
all values are measured in 2013 dollars). Benefits exceed costs programs like this on a large scale. It is prudent to be realistic
by a ratio of almost 5:1.5 From this perspective, mentoring on the ability of programs like this to satisfy existing need,
programs are a great investment. however. The logistical difficulties associated with managing
such a large number of mentoring relationships, let alone the
Mentoring programs thus appear to generate private returns recruiting and training of so many mentors, are substantial.
that are considerably in excess of their costs, but it is less clear Moreover, at approximately $1,500 per mentor, supporting
that they will generate a positive benefit–cost ratio when the just 1 million mentors would cost $1.5 billion, and the need is
focus is on social returns. The focus on altruism in justifying considerably greater than that. Despite my earlier claim that
the intervention is better suited for those NGOs and private- mentoring is an intervention that is better suited for NGOs and
sector entities that are trying to accomplish exactly that goal. private-sector entities, the extent of the need may be beyond
On the whole, I am in full support of these groups moving full- these groups’ means. Yet that does not lessen the importance
speed ahead in implementing community-based mentoring of the policy proposal I am making here. It is better to make a
The Hamilton Project • Brookings 53
SUPPORTING DISADVANTAGED YOUTH
Proposal 4: Designing Effective Mentoring Programs for Disadvantaged Youth
sizeable dent in an important social problem than to ignore it in their lives. Valiant attempts have been made to alleviate
because it cannot be solved completely. the difficulties associated with this caring gap in the lives
of disadvantaged youth. In fact, five thousand mentoring
Are there any circumstances under which the federal programs currently provide services to 3 million young
government should intervene to provide mentoring services? people; Big Brothers Big Sisters alone serves almost 200,000
children. Yet many disadvantaged youth still remain without
The purpose of this proposal is not to rule out federal
a mentor.
intervention to help address the lack of adult, caring
relationships in the lives of many of America’s youth. The Evaluating mentoring requires combing through extensive
argument I am making is that the hurdle is higher for justifying research on the programs and components already in play.
a role for public-sector intervention. Since it is unlikely that This policy memo tackles that task. After a careful review
mentoring programs can effectively demonstrate social of the best available evidence, I maintain that community-
benefits beyond program costs, justification for supporting based mentoring programs in the vein of the traditional Big
them is largely based on altruism. At least some component Brothers Big Sisters model are effective and should receive
of the private sector has that as an explicit goal, making it further support of NGOs and private-sector groups, with a set
a more-natural fit for that sector to tackle this issue. Some of best practices in mind as well as with rigorous evaluation to
government programs provide benefits for largely altruistic determine the important components for effective mentoring.
reasons, though; the Low Income Home Energy Assistance
Program is an example. If mentoring programs could A key consideration is whether government provision of
satisfy the altruistic goals of the public sector, then there is mentorship programs is justified—in other words, whether
no reason (outside budgetary constraints) why it could not the social returns of the program (e.g., in terms of crime
support them. reduction and increased tax revenue) exceed its costs. I find
that public spending on mentorship is not justified on these
grounds, and that mentorship programs should instead be
Conclusion provided by NGOs and private-sector entities looking to
improve outcomes for at-risk youth. Indeed, altruism is a part
Well-designed mentoring programs could go a long way
of the mission for these groups, and the benefit of providing
toward giving better opportunities to the more than 9 million
mentorship to disadvantaged youth outweighs the costs.
children growing up in America who have no caring adults
54 Policies to Address Poverty in America
SUPPORTING DISADVANTAGED YOUTH
Proposal 5: Expanding Summer Employment
Opportunities for Low-Income Youth
Amy Ellen Schwartz
New York University
Jacob Leos-Urbel
Claremont Graduate University
Introduction We propose that the federal government make grants to
state and local governments to work with local community-
Youth employment rates have decreased dramatically over based organizations (CBOs) on the expansion of summer
the past decade as the economy has faltered and the youth job programs. Targeting low-income youth ages sixteen to
population has grown, as shown in figure 5-1 (Bureau of nineteen (enrolled in or graduated from high school), these
Labor Statistics n.d.). Unemployment rates among youth are expanded programs would provide employment and training
especially acute during the summer, as more teens temporarily to young people who currently face many barriers to entering
enter the labor force (Morisi 2010; Sum et al. 2008). In response the workforce.
to this problem, the American Recovery and Reinvestment Act
of 2009 (ARRA) provided summer jobs for low-income youth EVIDENCE OF EFFECTIVENESS
with the goal of improving workforce readiness, although this Emerging evidence indicates that summer jobs can do more
increase in the availability of summer jobs was temporary than put a low-income youth to work. Using a rigorous
(Bellotti et al. 2010). This policy memo offers a proposal to lottery design, Leos-Urbel (forthcoming) finds that getting
strengthen and expand work-related summer activities with a job during the summer of 2007 through the SYEP in New
the goal of fostering the skill development, education, and York City (NYC) led to increases in school attendance in
economic success of low-income youth. the following school year of roughly 1 percent overall and 3
percent for students who may be at greater educational risk.
Summer jobs should be part of a broader strategy for poverty These effects are of a similar magnitude to some rigorously
alleviation, with the potential to benefit disadvantaged youth evaluated interventions that are explicitly designed to increase
in multiple ways. In addition to providing work experience school attendance (Dee 2011; Riccio et al. 2010). A follow-up
and an immediate income transfer to low-income youth, an study by Schwartz et al. (in progress) examining the impact of
emerging body of research also suggests that summer youth the NYC program from 2006–2009 also finds small increases
employment programs (SYEPs) can improve educational in school performance.
outcomes and social and emotional development, and
decrease negative behaviors (including criminal behaviors), at An earlier rigorous random-assignment evaluation of intensive
least in the short term (Heller 2014; Leos-Urbel forthcoming; summer jobs programs that included an academic component
Sum, Trubskyy, and McHugh 2013; Walker and Viella-Velez found that the programs increased reading and math scores
1992). A number of states and localities offer SYEPs on varying in the short term (Walker and Viella-Velez 1992).1 Notably,
scales, although the availability of jobs fluctuates year to year. the evaluation found no long-term academic or employment
The Hamilton Project • Brookings 55
SUPPORTING DISADVANTAGED YOUTH
Proposal 5: Expanding Summer Employment Opportunities for Low-Income Youth
FIGURE 5-1.
Employment-Population Ratio, Youth Ages 16 to 19, 2003–2012
50
Employment–population ratio (in percent)
45
40
35
30
25
20
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Annual July
Source: Bureau of Labor Statistics various years.
Source:
Note: Employment–population Bureau
ratio is the of of
ratio Labor Statistics
employed various
youth to allyears.
youth in the civilian noninstitutional population.
Note: Employment–population ratio is the ratio of employed youth to all youth in the civilian non-institutional population.
differences, which has led some to conclude that summer for high school students during the school year in Chicago—
jobs are not a good strategy for reducing poverty.2 Another found improvements in behavior and social and emotional
experimental evaluation found that a summer jobs program development, although it found no effect on academic
in Philadelphia increased the likelihood of youth getting a outcomes (Hirsch et al. 2011). Importantly, 91 percent of
job, but did not change intermediate academic or employment students in the comparison group in Chicago were involved in
outcomes (McClanahan, Sipe, and Smith 2004). This proposal some other after-school activity (most common) or paid work,
builds on the lessons of these earlier programs but takes some indicating that the availability of other opportunities (i.e.,
different approaches, as we describe below. the counterfactual of what they would have done without the
program) may differ considerably during the school year, and
Recent research also suggests that summer jobs can help suggesting that the summer may be an especially promising
reduce violent behavior and crime. Preliminary results from time for such interventions.
an experimental evaluation of One Summer Plus—which
combined summer jobs with a cognitive behavioral therapy–
based program aimed at reducing youth violence—indicate The Challenge
that the program led to a large decrease in violent-crime
arrests (Heller 2014).3 A study of a summer jobs program in High youth unemployment rates and a shrinking supply of
Boston finds that, compared to a comparison group of eligible traditional summer job opportunities for youth can have
youth from the program waiting list, program participants serious implications for their financial well-being and ultimate
were significantly more likely to reduce risky and violent labor market success (Rees 1986). Employment during high
behaviors, including the use of drugs and alcohol, physical school is linked to higher incomes as they become adults
fighting, damaging property, and threatening someone with (Painter 2010; Ruhm 1995). Furthermore, from an equity
a weapon (Sum, Trubskyy, and McHugh 2013).4 The program perspective, the availability of work opportunities for youth
also created much-needed jobs for program participants, as often varies by race and socioeconomic status (Morisi 2010).
just 27 percent of youth in the comparison group were able to For instance, Entwisle, Alexander, and Olson (2000) find that
find a summer job. Finally, an experimental evaluation of the white youth are more likely to work, though African American
After School Matters—an after-school apprenticeship program youth apply for jobs more often than whites. Also, in contrast
56 Policies to Address Poverty in America
Amy Ellen Schwartz, Jacob Leos-Urbel
to many publicly funded out-of-school programs that struggle proposal process to encourage states and localities to innovate
to recruit and retain high school students, jobs programs for in providing training and services to youth, and to build on
youth often face demand that far exceeds supply. best practices.
Public policies to support summer jobs are not new, though Our central proposal calls for extending the program
the availability of jobs fluctuates. At the federal level, ARRA nationwide. We recognize, however, that such a rapid
provided a temporary influx of funding for summer jobs expansion may face severe budget and administrative
for low-income youth that has since dried up (Bellotti et al. constraints. An alternative to an immediate nationwide scale-
2010). In particular, it provided $1.2 billion for employment up of the program is to implement a multiyear pilot program,
and training for disadvantaged youth ages fourteen to twenty- along the guidelines presented below, to a select diverse group
four, and employed 345,000 youth in the summer of 2009.5 of cities and localities. Program outcomes would be subject to
These jobs were in high demand, as indicated by an evaluation comprehensive review and evaluation, and initial funding for
of the program’s implementation, which found that the the pilot program would be set at one-tenth of the cost of the
number of applications received exceeded the number of job full-scale implementation. If the multiyear program is found
slots available at nineteen of the twenty job sites (ibid.). The to be effective at improving educational and labor market
U.S. Department of Labor, lacking funds to pay for summer outcomes for the targeted population, the pilot program would
jobs, coordinated the Summer Jobs+ program in the summer be expanded with the goal of reaching all disadvantaged youth
of 2012; this program sought pledges from companies and across the country.
nonprofit organizations to provide summer work experiences
for youth nationwide. The current iteration of this program is We model our proposal on NYC’s SYEP—the largest program
called Youth Jobs+. of its type in the United States—and we also borrow from and
integrate best practices from programs in other localities.
Many cities and states also offer summer jobs programs. The Based on the lessons learned from the summer jobs created
largest is NYC’s SYEP, which operated with a budget of $45.6 through ARRA and from the NYC program, we anticipate
million in federal, state, local, and private funds in 2013. high demand and propose allocating slots through a random
That same year, the program received more than 135,000 lottery system. This has the dual benefits of allocating
applications and served almost 36,000 participants, down positions fairly, and of allowing for rigorous evaluation of
from more than 52,000 participants in 2009 when ARRA program effectiveness by randomly creating treatment and
funds were available. In 2013 in Washington, DC, 14,000 control groups of lottery winners and losers, respectively.
youth participated in the summer jobs program, which
was administered by the city’s Department of Employment JOB PLACEMENTS
Services. (See table 5-A1 in appendix 5-A for information on We propose that the federal government, through the U.S.
other SYEP programs in select cities across the country.) Department of Labor, make grants to states to regulate
and coordinate these jobs programs, which will then be
Despite these efforts, both the fluctuating availability of administered by city and county governments.6 This grant-
jobs and funding constraints have limited the number of based program, in turn, will develop a request-for-proposal
disadvantaged youth who are able to participate in summer process to identify qualified CBOs that will administer the
employment programs, presenting an opportunity within program locally. In the case of NYC, the city’s Department of
public policy to meet this important need. Youth and Community Development administers the SYEP, and
contracts with CBOs throughout the city to place and supervise
youth in summer jobs and to provide training. Appropriate
A New Approach agencies could include city or county agencies responsible for
We propose expanding summer jobs programs for low-income youth development, workforce development, and/or education.
youth—ages sixteen to nineteen, in both urban and rural Local agencies then contract with CBOs, which place youth
communities, and who are enrolled in or have graduated from in summer jobs supervise and monitor these placements, and
high school—through a program that will pay participants provide the program’s education and training component. The
the federal minimum wage for working twenty-five hours most successful job training programs include experienced
per week for six weeks. (These eligibility parameters were staff and close connections between the program training and
chosen in an effort to keep down program costs by targeting work (Greenberg, Michalopoulus, and Robins 2003; Stanley,
the youth most likely to see the largest gains from a summer Katz, and Krueger 1998). CBOs should be selected through a
work program.) In addition, the jobs program will contain competitive process to ensure they have the experience and
an education and training component, and a request-for- qualifications to provide disadvantaged youth with effective
The Hamilton Project • Brookings 57
SUPPORTING DISADVANTAGED YOUTH
Proposal 5: Expanding Summer Employment Opportunities for Low-Income Youth
training and mentoring. Additionally, providers should have lack of success in previous programs such as the Job Training
knowledge of the local labor market to ensure that the training Partnership Act (JTPA) is that the participants did not
is relevant and necessary for participants’ success. Funds should enter the program with a baseline level of skill necessary to
be allocated in proportion to the number of students ages benefit from the work and training experience (Foster 1995).
sixteen to nineteen in each state, in school or just graduated, Additionally, JTPA training focused on remedial education
and living in poverty. Ideally, contracts with CBOs will be fixed rather than workforce-related training, a feature that more-
term and will be re-competed on a regular basis with specific successful youth employment programs tend to provide
performance evaluation criteria required for renewal to ensure (Greenberg, Michalopoulus, and Robins 2003).
the most qualified organizations operate the program.
In order to avoid any stigma associated with participation and
TRAINING to minimize the administrative burden, our proposal does not
include an income requirement. Requiring documentation of
The proposed training component provides an important
income can serve as a substantial barrier to program enrollment
opportunity for innovation and collaboration between
and can distract from the implementation and monitoring of
multiple youth-serving agencies and organizations to address
program quality (Curnan and Hahn 2010). That said, localities
issues specific to their target population and to the job skills
should be encouraged to target communities with low-income
important in the local labor market.
populations; it is likely that take-up will be higher among low-
To capitalize on existing expertise, the training component income populations. As an example, although NYC’s SYEP is
could be connected to the local high school curriculum, open to all city youth, approximately 90 percent of applicants
focusing on college and career readiness training aligned to are eligible for free or reduced-price lunch, which implies very
state or Common Core standards. For example, NYC’s Career low household income.
and Technical Education (CTE) Summer Scholars program
To estimate the size of the target population, we begin with
matches students with part-time summer internships and
the fact that there are roughly 17 million youth ages sixteen
engages students in a classroom experience to build workforce
to nineteen in the United States, of whom approximately 75
readiness skills. The program includes two full days per week
percent are ages sixteen to eighteen (Bureau of Labor Statistics
of classroom training focused on career readiness skills and
2013). To calculate the number of low-income youth, we
matches students with paid internships that are purposefully
assume that the number of youth ages sixteen to nineteen
aligned to the content of their CTE track, such as information
living close to or below the poverty level is the same as the
technology or media (Weinstein and Leardo 2013).
ratio of households with five-year-olds to seventeen-year-olds
MONITORING PROGRAM QUALITY living at or below 185 percent of the poverty level, which is
approximately 30 percent. This implies a target population of
Metrics for assessing program quality for selection of CBOs to
5 million low-income youth ages sixteen to nineteen, and 3.75
be providers and for contract renewals may include attendance
million low-income youth ages sixteen to eighteen. Both the
and hours worked, program completion or attrition,
sixteen-to-nineteen and the sixteen-to-eighteen age ranges
participant and supervisor evaluations, and feedback from
are appropriate for SYEP, as one would target high school
placement sites. These metrics are directly related to the core
students and recent graduates, and the other would target only
elements of the program and are relatively easy to measure in
high school students.
a standardized way across program sites, requiring a minimal
administrative burden. In addition to providing guidelines The evidence from NYC’s SYEP offers some insight into how
and incentives for program providers, the program should many youth would be interested. Approximately 80,000 low-
offer rewards to students for successful program completion income youth applied for a position for the summer of 2009,
(e.g., high attendance and positive supervisor feedback).7 which is around 40 percent of the roughly 200,000 low-income
youth ages sixteen to nineteen estimated to be living in NYC
TARGET POPULATION AND PROJECTED TAKE UP
(estimate based on data from U.S. Census Bureau 2000).
As mentioned above, the proposal targets youth ages sixteen Funding constraints meant that only half of applicants were
to nineteen who are enrolled in or have recently graduated offered positions and, importantly, approximately three out of
from high school, which we believe to be a population likely to four of those accepted the offer and participated.
benefit from the program. However, our proposal would not
provide training and support of the intensity and duration Combining these figures yields an estimated take-up rate
required to put out-of-school youth on a path to educational among the overall eligible population of 30 percent as a
and career success. In fact, one explanation for the perceived benchmark, which is likely a high estimate due to the relative
scarcity of private sector jobs in the weak economy. Taken
58 Policies to Address Poverty in America
Amy Ellen Schwartz, Jacob Leos-Urbel
together, this suggests that if universally implemented, 1.50 Summer represents an especially efficient area for intervention,
million youth ages sixteen to nineteen would be interested as it is a time when many youth lack opportunities for other
in participating in SYEP; again, it would be roughly three- formal activities. The loss by students over the summer of
fourths of that if the program was limited to students ages some of the skills learned during the school year is well-
sixteen to eighteen. documented in earlier and later grades (Castleman, Arnold,
and Wartman 2012; Entwisle, Alexander, and Olson 2000).
PROJECTED EXPENDITURES
The benefits from this federal investment go beyond providing
If implementing a multiyear pilot program, we propose
summer employment. Research suggests that SYEPs can also
dedicating $300 million annually for five years, at a total cost
have small positive effects on school attendance and academic
of $1.50 billion. As seen in table 5-1, we estimate that the total
outcomes (Leos-Urbel forthcoming; Walker and Viella-Velez
costs of expanding this nationwide to low-income youth ages
1992). In a preliminary investigation of the short-run impacts
sixteen to eighteen would be about $2.25 billion (assuming
of summer jobs programs, Leos-Urbel (forthcoming) finds
1.50 million participants, as calculated above, at a cost of
that these programs produce small increases in attendance
$2,000 per participant); if the program were offered to youth
in the following school year. Increases are larger for students
ages sixteen to nineteen, the estimated costs increase to about
at greater educational risk, namely those ages sixteen and
$3 billion. Importantly, the budget of our proposed program
older with low baseline school attendance. For this group,
is not a social cost. About half of the estimated program
participation in a summer jobs program also increases the
budget is the wage paid directly to the youth. From a societal
likelihood of attempting and passing statewide high school
perspective, this is a transfer of funds to low-income youth,
math and English exams. In current work exploring the impact
rather than a change in economy-wide resources.
of these programs on student academic outcomes over more
Table 5-1 breaks down the direct cost of the program, which years, preliminary findings suggest small positive effects of
is determined by the wage paid, number of hours and weeks the program on the number of exams students take; although
of the program, number of participants, and educational and impacts on scores are generally insignificant, there is a small
administrative costs. We estimate each of these factors drawing positive effect on passing key high school exams. Furthermore,
from the features and experiences of existing programs: national another study finds that the impacts increase with the number
data on the size and composition of the youth population, data of years a student participates in the jobs program—with
from the largest summer jobs program (NYC’s SYEP), and the impacts being larger for second-time participants and largest
administration of other social programs. for those participating for the third time or beyond (Schwartz
et al. in progress). Positive effects, even small effects, are
We propose that the national SYEP pay an hourly wage encouraging; as we have seen, numerous previous efforts have
of $7.25 for jobs that last twenty-five hours per week for failed to produce returns.10
six weeks during the summer. These program parameters
generally mirror features common to existing programs. Moreover, these small increases may translate into meaningful
While some programs offer higher wages, most SYEPs pay the gains in lifetime earnings. Rose (2005) finds that students
federal minimum wage (currently $7.25 per hour). Similarly, who made test score gains in high school were more likely to
twenty-five hours per week is in the middle of the range of be employed and have larger earnings seven years after high
hours offered, which typically ranges from twenty to thirty school compared to students whose test scores improved very
hours a week (see table 5-A1 in appendix 5-A). There is also little. Specifically, a one-point increase in a student’s test score
variation in program duration across the country from five gain from grade 8 to grade 12 predicted an increase of 0.62
to eight weeks during the summer, but six weeks is the most percent in earnings. Similarly, work by Deming and colleagues
common. While administrative overhead costs will vary with (2013) examines the impact of increased student performance
program features, we use the 15 percent overhead rate that the on high-stakes exams, postsecondary attainment, and
California Department of Education allows for public after- subsequent earnings. The authors find that students in high
school programs.8 We also include $650 per participant for an schools that raised test scores in response to accountability
educational component.9 pressure were more likely to attend and graduate from a four-
year university and had higher earnings at age twenty-five.
POTENTIAL OUTCOMES Impacts were strongest for students with the lowest baseline
Summer jobs programs introduce participating youth to the achievement. Specifically, increased test score performance
workforce, and these early work experiences have the potential led to 1 percent higher labor market earnings at age twenty-
to foster noncognitive skills, which prepare youth to enter the five. Given the approximate average earnings of $30,000 at this
labor force (Heckman 1998; Lillydhal 1990; Mortimer 2003). age, this effect would translate to $300 per participant.
The Hamilton Project • Brookings 59
SUPPORTING DISADVANTAGED YOUTH
Proposal 5: Expanding Summer Employment Opportunities for Low-Income Youth
TABLE 5-1.
Program Budget
Estimated costs Ages 16 to 19 Ages 16 to 18
Target population
Total low-income population in United States (in millions) 5 3.75
Take-up rate 30% 30%
Estimated participants (in millions) 1.50 1.13
Average cost
Participant compensation
Wage $7.25 per hour $7.25 per hour
Hours per week 25 25
Duration (in weeks) 6 6
Total $1,088 $1,088
Average other costs
Educational cost per participant $650 $650
Administrative overhead 15% 15%
Cost per participant $1,998 $1,998
Total cost
Total annual cost of SYEP (in millions) $2,997 $2,248
Sources: New York University Institute for Education and Social Policy 2014; authors’ calculations.
Recent research also suggests that summer jobs programs SYEP, the most common job placements are in summer camps
can reduce crime and violent behavior among individuals and day-care centers.
(Heller 2014; Sum, Trubskyy, and McHugh 2013). Heller
(2014) examines the impact of a program that provided youth COSTS AND BENEFITS
from low-income, high-crime high schools in Chicago with a Expanding summer jobs for low-income youth would yield
part-time summer job and cognitive behavioral therapy. The benefits in many dimensions, including to the individual
study provides credible, experimental evidence of a significant participant and to society. Benefits to the individual
link between crime and summer jobs, thus providing a social participants include income received, workforce readiness,
benefit that substantially exceeds program costs.11 reduction in risky behavior and crime, increase of earnings
over the long run, and improvements in educational outcomes.
Finally, paying low-income youth for work reduces poverty. By For example, as noted above, these programs have been shown
offering low-income youth an opportunity to earn wages, this to increase attendance among students in the school year
program would immediately increase the economic resources following the summer intervention, especially among those
available to participants and their families. This increased students with poor attendance records.
income would bring households on the poverty margin above
the poverty level and would ease the depth of poverty for all Social benefits include the services provided by participants,
others. In addition, expanding youth employment can assist such as service as a camp counselor, and improvements
nonprofit organizations in providing services to low-income in communities, such as reductions in crime. As noted
neighborhoods and communities. For example, in NYC’s above, prior research suggests a link between participation
60 Policies to Address Poverty in America
Amy Ellen Schwartz, Jacob Leos-Urbel
in summer jobs programs and lower crime rates. The high reach populations, such as out-of-school youth or those
social cost of each individual committed—numbering involved in the juvenile justice system (see, for example,
in the thousands for even low-level nonviolent crimes— Bloom et al. 1997; Farkas, Smith, and Stromsdorfter 1983;
suggests that even relatively small reductions in burglary or Orr et al. 1996). In contrast, the specific evidence on summer-
vandalism could provide sufficient benefits to offset the costs only programs that target in-school youth suggests that youth
of the SYEP program. summer employment programs hold promise for improving
youth outcomes, particularly educational outcomes, social
The costs of the program are measured by program outlays. and emotional developmental outcomes, and reduced negative
As noted above, we estimate the cost of a pilot program to be behaviors (Heller 2014; Leos-Urbel forthcoming; Schwartz et
$300 million annually, with the cost rising to between $2.2 al. in progress; Walker and Viella-Velez 1992).
billion and $3.0 billion annually if implemented nationwide.
Since a large portion of the program outlays are devoted What are reasonable expectations for the effects of a
to wages paid to participants, much of these outlays can program for youth of this duration, intensity, and cost?
be classified as transfers of income rather than changes in
economy-wide resources. We expect small effects across a range of critical dimensions,
including small increases (1 percent to 2 percent) in
Ultimately, we find that the summer youth program will attendance, educational attainment, and graduation. We also
have a series of relatively modest, but important, impacts on expect slightly larger effects on crime and risky behaviors,
participants and society. Although the effect on any one of particularly during the summer when students are employed.
these dimensions may be small, taken together they suggest
benefits that outweigh the relatively modest costs.
Conclusion
Questions and Concerns While there is a broad consensus that education can provide
a path out of poverty for low-income youth, out-of-school
How is this different from past youth employment programs time—including both summer and after-school activities—
that were considered by some to be a failure? can also enrich youth development. Recognizing this, middle-
class families routinely invest in travel, camps, internships,
Our proposal differs from prior federal programs in a few key and summer jobs, providing their children with experiential
features. First, we propose to serve youth who are enrolled learning and work experience while minimizing the amount
in or have just graduated from high school, a population of unsupervised idle time and the potential opportunities to
of students whom we believe are most likely able to take engage in risky behavior. We believe summer jobs can provide
advantage of the program. We recognize that this limits the some of the same benefits to low-income youth: increasing
potential of the program to help all disadvantaged youth. This their engagement in school, providing job experience, and
population is in contrast to those served by JTPA, for example, reducing participation in risky activities across a broad range.
which targeted out-of-school youth, a population who likely
have lower skills and require support that is more intensive. To be clear, our proposed SYEP is a very modest intervention.
Second, our program requires a regular workforce training It would be naive to imagine that this sort of low-cost
component closely aligned to the local employment context intervention will dramatically improve outcomes. Instead,
that is provided by qualified CBOs with expertise either we hope summer employment will lay a foundation on which
connecting individuals to the local job market or providing future success can be built.
local youth with support services and mentoring. The JTPA
education component for youth, in contrast, focused on While investment in early childhood education has captured
remedial education.12 the imagination of policymakers and the public alike, such
interventions will not address the inequality in opportunities
Why does this proposal make sense given the lack of and life-chances of today’s youth for whom completing school;
evidence on long-term effects of summer jobs programs on avoiding crime, pregnancy, and drug use; and other negative
education or employment outcomes? behaviors are critical steps on the path to future success.
Summer jobs may be an effective tool in the effort to reduce
Much of the available research on youth employment focuses inequality at the beginning of adulthood and may level the
either on effects of employment year round or on hard-to- playing field for low-income youth.
ACKNOWLEDGMENTS: The authors thank Michele Leardo and Megan Silander for outstanding research assistance on all
aspects of this proposal.
The Hamilton Project • Brookings 61
SUPPORTING DISADVANTAGED YOUTH
Proposal 5: Expanding Summer Employment Opportunities for Low-Income Youth
Appendix 5-A.
TABLE 5-A1.
Summer Youth Employment Programs, Select Cities
Program name Location Description
Midwest
Youth Opportunities Unlimited (Y.O.U.) Cleveland, OH Y.O.U. helps match thousands of teenagers from Cuyahoga County to meaningful
summer work experiences.
One Summer Chicago (OSC) 2013 Chicago, IL OSC connects young people to summer jobs, internships and training programs
that are offered throughout the city. Through OSC, young adults have the
opportunity to learn job skills, develop their résumés and explore career interests.
Workforce Development Board Detroit, MI The WDB-SYEP is introducing a program that will allow local businesses to
(WDB)–Summer Youth Employment contribute to summer work experiences for Detroit youth.
Program (SYEP)
Step Up Minneapolis, MN The program primarily serves youth from lower-income families, or youth with
significant barriers to finding a job.
Northeast
Boston Summer Jobs Boston, MA This program provides youth with training related to job readiness and career
exploration and job opportunities during the summer at a variety of private,
community, faith-based and government organizations.
WorkReady Summer Youth Philadelphia, PA Summer employment models offer educationally-enriched work opportunities to
Employment in-school and out-of-school youth that foster the acquisition of the twenty-first
century skills through work-based learning.
Summer Youth Employment New York City, NY The program provides New York City youth with paid summer employment and
Program provides workshops on job readiness, career exploration, financial literacy and
opportunities to continue education and social growth.
Mayor's Summer Youth Employment Norwalk, CT The program prepares youth for jobs via pre-employment workshops and
Program matches them with employment opportunities where they can explore a
profession, learn a skill, learn to navigate in a business environment, contribute
to the community, and earn money.
Rensselaer County Summer Youth Rensselaer County, The program provides income-eligible youth with a unique opportunity to gain
Employment Program NY meaningful job skills during the summer months through employment.
RochesterWorks! Summer of Rochester, NY RochesterWorks! is an employment and training program for youth who are still
Opportunity Program in high school. The program is designed to provide training and employment
opportunities to youth while making a direct connection to success in school.
Source: New York University Institute for Education and Social Policy 2014.
Note: Information downloaded from various Internet sites.
62 Policies to Address Poverty in America
Amy Ellen Schwartz, Jacob Leos-Urbel
Ages Program features Youth served Education component
Hourly wage Hours per week Duration (in weeks)
14–18 $7.95 25 6 4,600 in 2008 Y
14–24 — — 6 17,000 job opportunities Y
in 2013
14–21 $7.50 30 6 — Y
14–21 $7.25 — 6 1,280 in 2009 Y
16–24 $8.00–$12.00 25–35 7 10,000+ in 2009 Y
14–21 $7.25 20 6 5,144 positions in 2012 Y
14–24 $7.25 20–25 6 35,957 in 2013 Y
14–18 $8.70 25 6 — Y
14–19 $7.25 20 5 — Y
14–20 $7.25 — 6–8 845 in 2013 Y
The Hamilton Project • Brookings 63
SUPPORTING DISADVANTAGED YOUTH
Proposal 5: Expanding Summer Employment Opportunities for Low-Income Youth
TABLE 5-A1. CONTINUED FROM PREVIOUS SPREAD.
Summer Youth Employment Programs, Select Cities
Program name Location Description
South
Youth Employment Program Denver, CO This program provides career advising, mentorship, job readiness, financial
literacy and life skills training and work experience programs to allow students
to explore long-term career interests.
Workforce Partnership Summer Youth Kansas Funded through American Recovery and Reinvestment Act of 2009, SYEP is
Employment Program (SYEP) specifically intended for low-income youth with barriers to employment. The
central objective is to introduce and reinforce the demands and rewards of
holding a job.
Summer Youth Employment Program Virginia Beach, VA The program provides jobs and workplace readiness skills to young people who
often lack the skills required to obtain employment and succeed in the workplace.
Summer Youth Employment Program Washington, DC This program is a locally funded initiative sponsored by the Department of
Employment Services (DOES) that provides District youth with enriching and
constructive summer work experiences through subsidized placements in the
private and government sectors.
Summer Youth Employment Program Wilmington, DE The Summer Youth Employment Program provides students with a summer work
experience with the purpose of fostering job-related and personal skills and habits
important for success in future careers.
West
Hire L.A. Los Angeles, CA Hire L.A. is designed to emphasize real-world expectations, increase awareness
of services offered by local community-based organizations, and provide
opportunities for college, career, and financial literacy training.
Summer Youth Employment Program San Francisco, CA This program provides low-income youth with hands-on work experience,
job readiness training and ongoing support through partnerships with local
community-based organizations.
Seattle Youth Employment Program Seattle, WA During the summer, the program provides exposure to the world of work.
Internships take place in a range of sectors such as health care, education,
recreation, skilled trades, social services, and technology.
Source: New York University Institute for Education and Social Policy 2014.
Note: Information downloaded from various Internet sites.
64 Policies to Address Poverty in America
Amy Ellen Schwartz, Jacob Leos-Urbel
Ages Program features Youth served Education component
Hourly wage Hours per week Duration (in weeks)
14–21 $8.00 an hour for up to 160 hours — Y
16–24 $7.25 (in 2009) 20-30 6–8 515 in 2009 Y
16–21 $7.25 35 7 — Y
14–21 $7.25 25 6 14,000+ in 2012 Y
14–20 $7.25 25 5 150 in 2012 Y
14–21 — — 6 — Y
16–21 — — — — Y
15–17 — — 7 — Y
The Hamilton Project • Brookings 65
SUPPORTING DISADVANTAGED YOUTH
Proposal 5: Expanding Summer Employment Opportunities for Low-Income Youth
Appendix 5-B.
Previous Youth Employment Programs
Youth Incentive Entitlement Pilot Projects (YIEPP) Summer Youth Employment Training Program (SYETP)
• Youth sixteen to nineteen, low-income, who have not yet • Youth ages fourteen to twenty-one who are economically
completed high school. (Open to all teenagers in targeted disadvantaged and of school.
communities.)
• Funded by the JTPA, which repealed CETA.
• Funded by the Comprehensive Employment and Training
Act (CETA). • 1982–97.
• 1978–81. • Program components included employment. Training in
the form of remedial education was added after 1986.
• Program components included employment, but no training
or job search assistance. • Served approximately 500,000 to 700,000 youth annually.
• Served approximately 82,000 youth over the course of the The JTPA of 1982 provided federal funds to establish
program, 1978-1981. programs to prepare economically disadvantaged youth and
unskilled adults for employment, including funds to establish
The YIEPP provided low-income youth age sixteen to the Summer Youth Employment Training Program (SYETP).
nineteen who had had not yet graduated from high school This program, operated by the U.S. Department of Labor
with part-time jobs during the school year and full-time jobs and coordinated and regulated by states and administered
during the summer in exchange for meeting academic and by city and county governments, served youth ages fourteen
job-related performance standards. Specifically, to be eligible to twenty-one, and was initially designed to provide short-
participants were required to be enrolled in high school or in a term financial assistance in exchange for work. Youth worked
GED program. This federal program was established through in a variety of public, nonprofit, and private sector jobs and
the CETA, preceded JTPA, and operated from 1978 to 1981. were paid the minimum wage. In later years (after 1986), the
Students participated for an average of fifty-six weeks in the program also included an educational component for students
program. who were identified as needing education remediation
(Doolittle et al. 1993). Although little evidence is available
A matched-comparison study found that the program regarding the effectiveness of SYETP, some research suggests
increased employment in the short term and decreased the that the program provided jobs that would otherwise not have
unemployment gap between white and African American been available to youth (Stanley, Katz, and Krueger 1998).
youth, and increased school enrollment rates.
In addition to summer employment, approximately one-
The study also found that students were more likely to third of the population served under JTPA was economically
employed six months after the program ended (Farkas, disadvantaged out-of-school youth enrolled in year-round
Smith, and Stromsdorfer 1983). However, the study found no programs. Participants enrolled in programs for fifteen
impacts on school outcomes such as high school graduation months on average, but the length of the program varied by
(Gueron 1984). local site. Evidence from a randomized experimental study
found no effect of the program on the youth’s earnings thirty
months after participants were assigned to the program
Impacts also did not differ based on the type of training or
job search assistance youth applicants received. The program
did, however, have small positive impacts on educational
attainment—obtaining a high school diploma thirty months
after program assignment—for youth dropouts, particularly
female youth (Bloom et al. 1997; Orr et al. 1996). An important
caveat to this study is that the comparison group received non-
JTPA educational services, so that the estimated impacts are
not compared to receiving no program services at all.
66 Policies to Address Poverty in America
SUPPORTING DISADVANTAGED YOUTH
Proposal 6: Addressing the Academic
Barriers to Higher Education
Bridget Terry Long
Harvard Graduate School of Education
Introduction English or mathematics, but this figure can be as high as 60
or 70 percent of students at some institutions (Bettinger,
A postsecondary education confers numerous benefits both Boatman, and Long 2013; Complete College America 2012;
to the individual and to society, including higher earnings, National Center for Education Statistics [NCES] 2003).2
lower rates of unemployment and government dependency, Students placed into remedial or developmental programs are
an increased tax base, and greater civic engagement. Access most often held back from taking college-level courses, and as
to higher education remains a challenge for many families, a result, remediation has effectively become the gateway (or
however. In 2010, approximately 82 percent of students from barricade) to postsecondary-level training.3
high-income families attended college in comparison to only
52 percent of students from low-income families (National While the aim of remedial and developmental courses is to
Bureau of Economic Research n.d.).1 There are also large provide academically underprepared students with the skills
differences in rates of college completion by income: among they need to succeed in college and in the labor market, being
students who met a minimum standard of being academically placed into the courses also has important implications for a
qualified for college, 89 percent of high-income students student’s higher-education prospects. Students are forced to
completed a bachelor’s degree within eight years, whereas only pay college-level prices for high school–level courses; there
59 percent of low-income students did so (Adelman 2006). are also large government subsidies at stake given federal
funding and state appropriations that subsidize college costs
There are many barriers to college access and success. One and operating budgets. Time spent in remediation can also
major barrier is affordability, as college prices and student delay completion of a postsecondary degree. Credits earned
debt levels have risen to alarming heights. For many students, from remedial courses often do not count toward a student’s
however, academic preparation may be an equally formidable degree. Thus, it takes students longer to complete their studies,
barrier to postsecondary education. This is not due to college and this increases the chances that a disruption will derail
selectivity—about 80 percent of four-year colleges and nearly them from progressing. The extended time needed to obtain a
all two-year colleges have little to no admissions requirements. degree could also affect a student’s financial aid, as a student’s
Instead, students are required to pass academic placement eligibility for aid may expire; students who need to complete
tests and demonstrate sufficient readiness for postsecondary significant remediation could run out of financial support
study. Those who do not pass are placed into remedial or before being able to finish.
developmental courses.
Unfortunately, research suggests that remediation programs
Estimates suggest that more than one-third of all first-year do not do a good job of improving students’ outcomes. When
students take some form of remedial coursework in either comparing similar students in and out of remediation, some
The Hamilton Project • Brookings 67
SUPPORTING DISADVANTAGED YOUTH
Proposal 6: Addressing the Academic Barriers to Higher Education
researchers have found small positive effects, but most of
the research suggests no long-term effects—or even negative
effects—from being placed into a remedial or developmental
course (Bettinger and Long 2009; Boatman and Long 2010;
Calcagno and Long 2008; Martorell and McFarlin 2011). While
there are still unanswered questions about how the effects
differ by type of student, most researchers, practitioners, and
policymakers have concluded that the current remedies we
have to address the fact that so many students are academically
underprepared for college are not sufficient, and may in fact
involve serious costs for students, institutions, and taxpayers.
There is ongoing debate about the best way to address students’
academic needs. Many states are confronting questions about
who should deliver remediation and how it should be offered.
Some are considering ways to limit the courses, shift their
locations, or pass on the costs of the courses to students or
school districts. While states lament the need for remediation
and debate how to manage it, however, most of the current
policy efforts do not focus on how to improve programs or
help students avoid remediation altogether.
This policy memo offers three key recommendations for better
addressing the academic preparation problem with the hope
of improving rates of college success. The recommendations
focus on actions that could be taken by states, university
systems, and school districts. The federal government could
also play an important role by creating incentives for states
and institutions to address these issues or by supporting a
central organization with the purpose of providing guidance
on best practices to states and institutions. This proposal’s
recommendations are as follows:
1.Improve placement in college remediation classes.
Improving how students’ academic preparation levels are
assessed is the first step in better tailoring supports for
their needs. Better assessment is also necessary to reduce
the number of students who are incorrectly placed into
remediation due an opaque process or bad testing day.
2. Provide better college remediation services. By using
technology, support services, and innovative pedagogies,
remediation programs could doa much better and faster job
in helping to prepare students for future success with college-
level material. Several states are already experimenting
with promising practices, including combining basic-
skill attainment with college-level coursework, and using
learning technology to better target students’ needs.
3. Adopt measures to prevent the need for remediation.
Several states are encouraging students to take college
readiness assessments in high school so that they can use
this early information to make better course selections
68 Policies to Address Poverty in America
and avoid remediation altogether. Working to better align
curricula and strengthen links between K-12 and higher
education could also improve the likelihood that students
are academically prepared for college.
The Challenge
BACKGROUND: POSTSECONDARY REMEDIATION IN
THE UNITED STATES
Multiple studies point to the fact that high school graduates
are often not academically prepared for college. Some
estimates suggest that only about one-quarter of high school
graduates complete a rigorous academic curriculum (NCES
2010). While academic preparation is a problem for many
students, it is a problem that especially affects low-income and
minority students. According to Greene and Foster (2003),
only 32 percent of students leave high school at least minimally
prepared for college, and the proportion is much smaller for
African-American and Hispanic students (20 and 16 percent,
respectively).° Low levels ofacademic preparation are the result
of poor course selection, lack of academic rigor, and a limited
supply of advanced courses at some schools. In addition,
the lack of alignment between the K-12 and postsecondary
education systems frequently results in confusing messages
about how and what students should do to enter and succeed
in college (Venezia, Kirst, and Antonio 2003).
Although many underprepared high school students will fail
to continue their educations, the large proportion of those
who enter higher education will be placed into remediation.
A substantial number of adult students, including recent
immigrants and workers displaced by structural shifts in
the labor market, also enroll in remedial and developmental
Traditionally, the purpose of remedial or
developmental education has been to address whatever was
missed in high school (Education Commission of the States
2012). Nonselective public institutions provide the bulk of
remediation, with rates being highest at two-year colleges
(Bettinger and Long 2009).
courses.
The need for remediation is established based on an exam or
assessment taken when the student first arrives on campus.
Colleges then assign students to a specific course level based
on their scores on the placement test as well as, possibly,
high school courses and grades. Placement into mathematics
remediation is more common than placement into English
(ie., reading and/or writing) remediation, but participation in
English remediation may be a more serious concern as some
evidence suggests that reading and writing deficiencies have
more-negative effects on a student’s college success (Bailey,
Jeong, and Cho 2010; Bettinger and Long 2009; McCabe 2001).
Bridget Terry Long
The vast majority of institutions require students to complete in remedial courses have lower levels of preparation than
their remedial courses before they are allowed to enroll in those who are not placed into remediation, one would expect
college-level courses (NCES 2003). For students in need of remedial students to be less likely to persist and complete a
multiple remedial courses, this could mean more than a degree even in the absence of a remediation program. The
year of coursework before progressing to actual college-level key to understanding whether remedial programs work is to
material. Although remedial courses are offered for credit and compare students with similar preparation levels.
count toward a student’s overall GPA, remedial courses rarely
count toward graduation requirements (Bettinger and Long The results are mixed when new data sources that compare
2007). similar students are used to study the effects of remediation
on student outcomes. For example, Bettinger and Long (2009)
As such, remediation becomes a costly investment incurred examine the effects of remediation in Ohio and conclude that
by students, institutions, and the government. Although remedial students at Ohio colleges were more likely to persist
estimates vary depending on the source, they all suggest in college and complete a bachelor’s degree than students with
that remediation is expensive in multiple ways and for similar test scores and backgrounds who were not required to
multiple stakeholders. Alliance for Excellent Education take the courses. In contrast, focusing on Florida, Calcagno
(2006) estimated that the cost of the delivery of remediation and Long (2008) suggest that remediation might promote
nationwide totaled $1.4 billion in the form of direct costs to early persistence in college, but it does not necessarily help
students and institutions. Further costs would result from the community college students make long-term progress toward
lost earning potential of those remedial students who drop a degree. In Texas, Martorell and McFarlin (2011) find that
out of college without completing a degree. Another study remediation programs had little effect on persistence, degree
estimated the annual cost of remediation to be between $1.9 completion, or a range of other educational outcomes. They
and $2.3 billion at community colleges and another $500 also find no effect on labor-market earnings. It is important to
million at four-year colleges (Strong American Schools 2008), note that much of this research focuses on students just on the
while yet another study estimates that states and students spent margin of needing remedial courses (i.e., students who either
more than $3 billion on remedial courses in 2011 (Complete need one remedial course or go directly into college-level
College America 2012). The most recent estimate suggests that work). Far less is known about the effectiveness of remediation
the national direct cost of remediation is actually as high as in helping students with greater academic needs, though there
$7 billion annually (Scott-Clayton, Crosta, and Belfield 2012). is some suggestive evidence that more-intensive remediation
This estimate does not account for the opportunity cost of can have positive effects (Boatman and Long 2010).
time for students enrolled.6
Even with an incomplete and mixed understanding of whether
EVIDENCE OF THE PROBLEM AND THE CURRENT remediation works or how to improve it, this is a critical time
POLICY DEBATES in terms of remediation policy. In several states, including
Most current models of remediation are not working well: Indiana, South Carolina, and Tennessee, four-year institutions
students placed into remediation are far less likely to persist are prohibited from offering remedial education and are
and graduate from college. Fewer than 50 percent of students expected to make arrangements with community colleges to
referred to remediation actually complete the entire sequence. handle the remediation of students accepted for admission
This percentage is even lower for men, older students, African- (Long and Boatman 2013). The shifting of remediation to only
American students, part-time students, and students in community colleges could have important repercussions on
vocational programs. The students assigned to the lowest student success because community colleges receive far less
levels of math remediation are the least likely to advance into in funding, and transfer rates to four-year institutions are
college-level courses, with only 10 percent of this group ever low due to numerous structural and financial barriers (Long
completing a college-level math course (Bailey, Jeong, and and Kurlaender 2009). In addition, there has been a general
Cho 2010). increase in admissions standards at many institutions to
screen out less-prepared students. In some cases, academic
While disconcerting, these statistics on completion tell only deficiencies are so severe that colleges choose to expel new
part of the story. Longer-term educational outcomes, such students rather than remediate them.7
as total credit accumulation and degree completion, are also
Other states and institutions are considering how to control
much lower for students placed into remediation (Adelman
the costs of remediation. Some limit the percent of students
2006; Bailey 2009; Bettinger and Long 2005; Complete College
who need remedial courses that can be accepted by an
America 2012). This fact alone is not evidence that remedial
institution, while others limit the amount of time students
programs do not work, however. Since students who are placed
The Hamilton Project • Brookings 69
SUPPORTING DISADVANTAGED YOUTH
Proposal 6: Addressing the Academic Barriers to Higher Education
have to complete remediation or the number of times they can kind of standardized placement exam to assign students to
repeat a remedial course. For example, students who do not remedial or developmental courses (Hughes and Scott-Clayton
meet the minimum standards for college-level work within 2010).8 Typically, administrators make these designations based
the University of Georgia system are placed into Learning on hard cutoffs—students scoring below a given threshold
Support classes. Students may only take one Learning are assigned to a remedial course. In fact, Parsad, Lewis,
Support class in English language arts and have only two and Greene (2003) found that the two-year colleges where
attempts to pass the course. In terms of math, students can remediation is particularly concentrated almost exclusively
take up to two Learning Support classes and must pass these use brief, standardized tests administered to new students just
courses within three attempts, with no appeals (Georgia prior to registration to determine who should be placed into
Board of Regents 2010). In 2012, at least seven states restricted remediation. The strong reliance on a single exam is fraught
or eliminated state funding for remedial courses at some of with problems, however, and high-stakes placement exams are
their four-year colleges, thereby forcing these institutions to poor predictors of college readiness (Complete College America
fund remedial courses strictly through the use of tuition and 2012). Moreover, misplacing students who do not actually need
fees (Smith 2012). remediation into these courses can have a discouraging effect
on college enrollment and persistence (Scott-Clayton and
The policy decisions of where to allow remediation and Rodriguez 2012).
whether to limit it in some way have huge implications for
access to college-level training and for whether attending There is increasing attention to the fact that the diagnostic value
college is truly an avenue out of poverty. If the goal is to of remediation placement exams may be limited. Examining
improve educational attainment and skill levels, as well as multiple contexts, researchers have found repeatedly that
reduce government dependency, then states and institutions placement tests do not yield strong predictions of how students
should carefully consider how to govern and provide will perform in college. For example, Scott-Clayton (2012)
remediation (Long 2012). As described below, better placement examines data on over 42,000 first-time students at a large,
policies, improved services, and initiatives to reduce the need urban, community college system to determine the predictive
for remediation would significantly help address this major validity of one of the most commonly used remediation
barrier to postsecondary education. assessments. Her analysis suggests that one-quarter to one-
third of students assigned to remedial classes based on test
scores alone could have passed college-level classes with a
A New Approach grade of B or better.9 Looking at two large community college
systems, Scott-Clayton, Crosta, and Belfield (2012) find that
Given that remediation often acts as a major barrier—instead approximately one in four and one in three test takers in math
of as a gateway—to postsecondary education for many and English, respectively, are severely misassigned under
students, this memo offers three key recommendations for current test-based policies. They conclude that more students
improving remediation services, and thus rates of college are incorrectly assigned into remediation than are incorrectly
completion. States, university systems, school districts, and passed on to college-level coursework.
even the federal government could take up and encourage any
or all of the following steps for improving the remediation There is, however, an easy way to improve student placement:
system and for ultimately removing its need altogether. in addition to test scores, institutions could use information
about a student’s high school GPA, courses taken, and years
IMPROVE PLACEMENT IN COLLEGE REMEDIATION
since high school graduation. Scott-Clayton (2012) argues that
CLASSES
incorporating these multiple measures could reduce what she
Improving how students’ academic preparation levels are defines as “severe misplacements” by 15 percent. This could
assessed is the first step in better tailoring remediation have the added effect of reducing the remediation rate by 8
supports for their needs. Rather than a single remediation to 12 percentage points while still maintaining or increasing
placement exam, one alternative for determining a student’s success rates in college-level courses.
college readiness is to use multiple measures, including
information about a student’s high school GPA, courses taken, Focusing on a different set of colleges, Scott-Clayton, Crosta,
and/or years since high school graduation. and Belfield (2012) come to a similar conclusion: using
information from a student’s high school transcript, either
Currently, there is wide variation in what colleges use to assess instead of or in addition to placement-test scores, would
students and what thresholds they use to determine who should substantially reduce the number of students placed into courses
be in remediation. Most colleges and universities use some incorrectly. Most importantly they conclude, “If institutions
70 Policies to Address Poverty in America
Bridget Terry Long
took account of students’ high school performance, they Redesigning developmental courses could take a number
could remediate substantially fewer students without lowering of forms. Some states and institutions have focused on
success rates in college-level courses.” interventions that accelerate progress through remedial
courses by mainstreaming students into college-level courses
Findings like this have increasingly led states and university while also providing additional supports, such as tutoring,
systems to reevaluate their placement policies. Given advising, or targeted sections outside of class. Other programs
the importance of high school preparation in predicting combine basic-skill courses with college-level coursework in
college success, it is not entirely surprising that taking into a coordinated fashion. Still other programs have focused on
account information about high school course-taking and using technology and/or targeted teaching modules to reduce
performance would improve placement decisions, and the the content students are required to complete. Such programs
potential benefits are large. The surprising fact is that high allow for more customization and personalization based on
school grades and coursework are not already widely utilized diagnostic assessments. Table 6-1 summarizes some of the
as screening tools for many institutions (Belfield and Crosta major state and system efforts.
2012; Scott-Clayton 2012). This is a completely feasible policy,
however, as demonstrated by the fact that some schools and For example, the Community College of Baltimore County
systems already engage in the practice. The costs, beyond has the Accelerated Learning Program, which places students
some additional staff attention, are predicted to be small, who placed into upper-level English developmental courses
especially in comparison to the potential cost savings of into the first college-level composition course instead. It then
avoiding unnecessary classes. requires the student to co-enroll in a support section taught
by the same instructor. Cho and colleagues (2012) find that
In addition to better placement, there are also calls to do a the program significantly increased the rate of completion
better job diagnosing students’ specific needs to better match in the first and the second college-level composition classes
them with appropriate resources. The major remediation within three years. Such programs do not appear to reduce
placement exams contain multiple parts that could be used to the percent of students who pass their college-level courses.
pinpoint the exact needs of students. Using the full value of Edgecombe and colleagues (2012) find that students at another
these assessments to get a better sense of a student’s specific school who elected to use an accelerated pathway into college-
weaknesses could result in improved matching of students level work had passage rates at or above students who first took
with effective resources and supports, along the lines of those developmental education courses.
described in the second recommendation.
Complete College America (2012) has also concluded that this
PROVIDE BETTER COLLEGE REMEDIATION SERVICES is a promising approach; they suggest that students with few
The second key step is for states and institutions to collaborate academic deficiencies should be placed in college-level courses
on systems that provide better remediation services and with corequisite built-in supports such as just-in-time tutoring
supports. Currently, the primary effect of remediation appears and required self-paced computer labs. In addition to the
to be diversionary: students simply take remedial courses Community College of Baltimore County, other institutions
instead of college-level courses, but the research suggests the that have initiated similar programs include the University
remedial courses are doing little to improve student skills of Maryland at College Park, Austin Peay State University in
on average (Scott-Clayton and Rodriguez 2012). Given the Tennessee, and Texas State University–San Marcos.
growing number of students in need of remediation and
the small, mixed results about whether students achieve A program that combines basic-skills attainment with college-
academic success from these courses, an increasing number level coursework is the state of Washington’s Integrated
Basic Education and Skills Training (I-BEST) program.
of institutions are beginning to rethink the ways that they
In the I-BEST program, remedial instructors and college-
offer and teach their remedial and developmental courses.
level faculty jointly teach courses that combine basic-skills
I propose promoting the use of innovative pedagogies,
attainment with college-level material. Using this approach,
technology, and support services to better equip students
the students gain their basic skills through job training.
academically. Such methods could also help to streamline
Evaluations of the I-BEST program show higher rates of credit
the pathway through remediation to increase the proportion
accumulation among recipients over time, as well as higher
of students who complete remedial courses and progress to
rates of persistence to the second year (Jenkins, Zeidenberg,
higher-level academic work (Edgecombe 2011; Zachry and
and Kienzl 2009).
Schneider 2011).
The Accelerated Study in Associate Programs (ASAP) at the
City University of New York (CUNY) is another example
The Hamilton Project • Brookings 71
SUPPORTING DISADVANTAGED YOUTH
Proposal 6: Addressing the Academic Barriers to Higher Education
TABLE 6-1.
Possible Approaches to Redesigning Remediation
Definition Examples Effects
Mainstreaming
Place students into college-level courses Accelerated Learning Program, Participation in the program increased the
and provide additional supports (e.g., Community College of Baltimore County completion rate of college-level composition
tutoring, special sections, and advising). (Maryland): This program allows students classes within three years.
to take the first college-level composition
course and co-enroll in a support session.
Austin Peay State University (Tennessee): Students exposed to redesigned
This program offers enhanced sections developmental courses had more positive
of two core college-level courses and outcomes than similar students not in
linked them to Structured Learning remediation or in traditional remediation.
Assistance workshops.
Linked remedial and college-level courses
Combine remedial courses with college- Integrated Basic Education and Skills Recipients had higher rates of credit
level coursework in a coordinated fashion. Training Program (I-BEST) (State of accumulation and higher rates of
Washington): Remedial instructors and persistence to the second year.
college-level faculty jointly teach courses
that combine basic-skills attainment with
college-level material.
Accelerated Study in Associate Programs ASAP students were 66 percent more likely
(ASAP), City University of New York (New to complete an Associate degree.
York): ASAP links developmental courses
with other college-level courses and
provides supplemental supports to the
classes; it requires students to attend
full-time.
Learning Communities, Kingsborough Students in the learning community moved
Community College (New York): more quickly through their developmental
This program organizes students requirements, enrolled in and passed more
into cohorts that take paired remedial courses, and earned more credits in their
and college-level courses. first semester.
Technology-enhanced learning and modularization
Use assessments to determine students’ Emporium Models: In this program, Descriptive trends suggest students are more
specific needs and have targeted, short students move at their own pace through likely to complete developmental and college-
modules designed to address those online tutorials with support from teaching level courses.
needs. assistants.
Sources: Boatman 2012; Cho et al. 2012; Jenkins, Zeidenberg, and Kienzl 2009; Scrivener and Weiss 2013; Sommo et al. 2012; Twigg 2011.
of a promising program that links developmental courses college full-time and providing them with a rich array of
with other college-level courses and provides supplemental supports for three years, including tuition waivers, free use
supports to those classes. In their evaluation of the effects of textbooks, block-scheduled classes, enhanced advising,
of ASAP on student outcomes, Scrivener and Weiss (2013) career services, and free subway cards for transportation.
describe the program as requiring students to attend Their evaluation found that after two years, ASAP increased
72 Policies to Address Poverty in America
Bridget Terry Long
the proportion of developmental education students who Overall, these cases demonstrate that redesigning remediation
completed an Associate degree by 5.7 percentage points, an programs can take many different forms. The costs of these
increase of 66 percent. innovations and redesigns are currently being documented,
and they will depend on several factors, including the number
Other redesign efforts focus on changing the traditional of students served as well as the costs of instruction and
structure of a remediation course, which is typically a fifteen- supplemental supports. It will also be important to distinguish
week, semester-long lecture or seminar format in which a between the initial costs entailed to establish a new program,
student takes one remedial course in a given subject before which might include investments in technology, and the long-
moving on to the next course in the sequence. Institutions are run costs of having a new program. However, these costs must
experimenting with incorporating learning technology such be compared to the benefits gained and to the current level of
as self-directed learning labs and online-learning models, expenditure.
and with using high-tech classrooms (Epper and Baker 2009).
These newer models of remediation attempt to better target ADOPT MEASURES TO PREVENT THE NEED FOR
students’ academic needs and help them to move more quickly REMEDIATION
through their remedial courses. The final recommendation is for high schools, higher-
Emporium models are an increasingly popular strategy that education institutions, and states to adopt measures with
aims to help students complete their remediation faster. With the aim of preventing the need for remediation altogether.
this approach, students typically attend class in a computer lab Indeed, the need for remediation in college is closely tied to a
and move at their own pace through online tutorials. Students student’s high school curriculum. A study by the Ohio Board
not requiring much help might move through the material of Regents (2002) finds that students who had completed an
in a few weeks, while other students could take multiple academic core curriculum in high school were half as likely
semesters. Students have access to teaching assistants to help to need remediation in college compared to students without
them as they complete the modules, and professors track their this core, and other research also emphasizes the importance
progress (Boatman 2014). Descriptive trends suggest students of academic preparation in high school for success in college.
are more likely to complete developmental and college-level Numerous studies link the courses students take in high
courses using this approach, and that they do so at a lower cost school to their performance in higher education (Attewell
(Twigg 2011), but more research is needed. and Domina 2008; Long, Conger, and Iatarola 2012). For
example, Adelman (1999) tracked a cohort of students and
Texas is currently engaged in such an effort. The Texas found that their academic backgrounds, as measured by their
Higher Education Coordinating Board is working with the high school curriculum, academic intensity, class rank, and
College Board to develop a diagnostic testing system that GPA, were the most critical factors in determining college
informs students not only of their placement, but also of enrollment and success. In a later update, Adelman (2006)
what specifically they do not understand about the material. finds that students differ significantly in the types of courses
As profiled by Boatman (2014), students who receive the they take by background. He concludes that a high school
diagnostic will be required to take only the modules addressing curriculum is becoming even more compelling in terms of
their specific academic needs. its role in degree completion.
Many of the examples noted above demonstrate what could Completion of a high school core curriculum does not ensure
be done at the institutional level to redesign remediation that a student will avoid remediation in college, however.
programs, yet several reform efforts involve state policies Upon enrolling in college, students are often surprised
and higher-education systems. For example, in 2007–8, to learn they need to take such courses. Many students
the Tennessee Board of Regents implemented a redesign and families believe that meeting high school graduation
of remediation that initially involved six campuses. While requirements will adequately prepare them for college. But
the details of each institution’s redesign effort differed, they to avoid remedial college coursework, students often need to
focused on using learning technology, both in and out of take a more-rigorous and more-demanding secondary school
the classroom, to enable students to work at their own pace curriculum than that required by the district or state. Poor
and focus their attention specifically on the particular skills alignment between the K–12 and postsecondary education
in which they were deficient. Boatman (2012) concludes systems results in confusion about how and what students
that students exposed to these redesigned developmental should do to be able to enter and succeed in college (Venezia,
mathematics courses had more positive outcomes than similar Kirst, and Antonio 2003).
students from both nonredesign institutions and from prior
cohorts at the same institutions.
The Hamilton Project • Brookings 73
SUPPORTING DISADVANTAGED YOUTH
Proposal 6: Addressing the Academic Barriers to Higher Education
The use of college placement exams as early diagnostic tools in authors conclude that EAP increased students’ academic
high school is one promising policy aimed at better connecting preparation in high school but did not discourage poorly
student high school preparation with the requirements of prepared students from applying to college. This research
postsecondary courses. For example, several states administer suggests the promise of early assessment programs in reducing
to younger students the same remediation placement test the need for remediation.
that is ordinarily given to college freshmen. Most often this
testing is done in tenth or eleventh grade. Such tests are Another state involved in a large-scale early testing initiative is
designed to improve college-preparatory information for high Tennessee. In 2013, more than one hundred high schools in the
school students and to encourage those who fall short to take state offered the Seamless Alignment and Integrated Learning
additional coursework in their senior year. With assistance Support program. This program identifies high school juniors
from teachers, counselors, and parents, students can then who are on track to need college remediation, and allows
determine what courses to take while they are still in high them, while they are still in high school, to complete the same
school in order to avoid college remediation. remedial math course they would eventually have needed to
take in college (Boatman 2014).
Several states have experimented with early-testing policies,
including California, Kentucky, North Carolina, Ohio, and The summer before college matriculation is another important
Oklahoma.10 As shown by their examples, state-level early time when students could try to address their academic needs
placement testing policies can take a variety of forms. The and avoid remediation. Summer bridge programs can take
tests used range from standardized tests, (e.g., ACT’s Plan) to many forms, from trying to enhance study skills to giving
exams closely resembling those that colleges give to entering students the opportunity to begin their coursework. The
freshmen (e.g., Computerized Adaptive Placement Assessment California State University system has the Early Start policy,
and Support Systems [COMPASS] and ACCUPLACER). The which requires incoming first-time freshmen who are not
timing also varies among existing programs: some policies college-ready to begin their remediation during the summer
target high school juniors, while others test high school before enrolling (Reed 2010). Similarly, the CUNY Start
sophomores or even eighth graders (Long and Riley 2007). program has students spend the semester before beginning
college taking a developmental course. Logue and Mogulescy
The design and structure of a program, as well as the policies (2013) finds that the program has been successful in helping
developed beyond the test to support the program’s intentions students avoid remedial courses once enrolled in comparison
of giving early diagnostic information, are key dimensions to a similar group of students who did not enroll in the
that could affect whether the policy has its intended impact. program.
For instance, a program that is not mandatory and requires a
high school or teacher to opt into the program to participate Other institutions have targeted students during the summer
may not reach many of the students who would benefit. before registration with tips and resources to help them
Moreover, research suggests that taking a test and receiving prepare for the remediation placement exam. For example,
a score report falls short of providing many students with a Santa Monica College offers an online orientation to its
clear signal. Students must be supported after the test with placement test, which explains the content and format of the
counseling to encourage additional course enrollments. It test and offers tips on how to best prepare. In a similar fashion,
may even be necessary to develop new courses and pathways the Community College of Denver published a workbook
to fill gaps. for students to review the material on the ACCUPLACER
placement exam and offered free tutoring sessions for
The experience of California with its Early Assessment interested students. Still another example is Guilford Technical
Program (EAP) is informative for other states and higher- Community College in North Carolina, which created an
education systems. The California EAP aims to provide online course designed to prepare students to take or retake
high school juniors with information about their academic the COMPASS placement test (Quint et al. 2013).
readiness for coursework at California State University
campuses. After the test in eleventh grade, interventions are Finally, additional ways to improve prevention include
developed for the student to pursue during twelfth grade. The strengthening the links between K–12 and higher education.
EAP also includes professional development for teachers. An This could be done by better aligning curricula and including
evaluation of the program found that participation in the EAP higher-education representatives in conversations about K–12
reduced a student’s probability of needing remediation in assessments. For instance, bringing together high school
college by 6.2 percentage points in English and 4.3 percentage English teachers with college English professors would foster
points in math (Howell, Kurlaender, and Grodsky 2010). The smoother transitions for students. Links between the systems
could also be built into K–12 accountability systems and report
74 Policies to Address Poverty in America
Bridget Terry Long
cards. As many districts have already started to do, college cost savings for students, institutions, and taxpayers in the
enrollment rates of recent graduates could be publicized. long run. When considering the social benefits of college
Taking this a step farther, statistics on the placement of recent education, the rewards to improved remediation seem likely
high school graduates into college remediation would be a to be worth the initial investment.
useful way to judge secondary-school rigor and success in
preparing students for college-level material.
Questions and Concerns
COSTS AND BENEFITS
Is remediation worthwhile at all? If remediation is so
Strengthened remedial education has the potential to
expensive, should we just get rid of it?
improve the effectiveness of education spending. At the high
school level, improved diagnostic tests can allow schools To eliminate remediation would be counterproductive to the
to tailor educational curricula before students even attend goal of increasing degree attainment. As noted by Cloud (2002),
college, significantly reducing the need for college-level doing so would “effectively end the American experiment
remediation. For example, as noted above, California’s EAP with mass postsecondary education.” The low levels of
reduced a student’s probability of needing remediation into academic preparation inherited by higher-education systems
6.2 percentage points in English and 4.3 percentage points are certainly a challenge, but solutions need to be found to
in math. In addition, improved placement into college-level address the problem if the country is going to succeed in
remedial courses can save both student and college spending increasing educational attainment and reducing government
on remediation. Academic evidence suggests that a large dependency, especially among low-income individuals who
share—between one-quarter and one-third—of remedial might otherwise be in poverty and lack the skills necessary for
students are misassigned to remedial courses; assigning these advancement. Moreover, research in recent years highlights
students to more-appropriate courses will lower educational promising practices that would improve student preparation
costs and allow students to complete courses that better and outcomes, as well as reduce unnecessary costs.
improve their abilities and knowledge.
Why not just focus efforts on improving the K–12 education
Better administration of remedial courses can have important system?
impacts on educational and labor-market outcomes.
Interventions aimed at improving supports for students in Improving the K–12 system would have benefits, but the
remedial courses—such as the state of Washington’s I-BEST problems facing high schools are numerous: insufficient
program or CUNY’s ASAP program—can lead to improved academic rigor, a lack of alignment with postsecondary
college persistence and higher graduation rates. These institutions, and a limited supply of advanced courses at
outcomes are particularly promising for low-income and some schools. Even if these problems were solved, the country
minority students who exhibit low rates of college completion. would still have to contend with addressing the needs of
Higher rates of college completion can then translate into older, nontraditional students, who make up approximately
improved labor-market outcomes, namely higher rates of 40 percent of college students today. Moreover, students
employment and elevated earnings. sometimes make poor choices about their courses, and while
improving early information about college preparedness
Depending on the nature of the intervention, better levels would help (as recommended above), some students will
remedial education may temporarily raise spending in the not decide that they need a college education until after high
implementation phase. However, even though redesigned school. Therefore, colleges and universities need to improve
courses and improved remedial supports will incur initial their efforts to address the needs of these students. With
outlays, the short-term costs of starting a new program should remediation rates being as high as 70 percent at some colleges,
be measured against long-term cost savings. For example, focusing on K–12 alone will not solve the problem.
programs that reduce the need for remedial education
can lead to lower overall spending over time. In addition,
programs that require an initial capital investment—such as Conclusion
technology-based programs that require new computers and
programming—will incur costs early in the development Remediation plays an increasingly important role in the lives
process, but these costs are expected to decline over time. of students and the colleges and universities they attend.
Traditional remedial courses are costly in terms of time
In sum, improved remediation may lead to slightly higher and resources, however, and fail to improve the chances
educational outlays in the short run, but will likely lead to that students will be successful in college and graduate with
The Hamilton Project • Brookings 75
SUPPORTING DISADVANTAGED YOUTH
Proposal 6: Addressing the Academic Barriers to Higher Education
a credential. As a result, remediation is a major barrier to for remediation by better aligning curricula and having high
postsecondary-level training for many students, and currently school students take college readiness assessments earlier so
the system is not designed to help students get over that that they can make better decisions about the courses they
hurdle. While some states debate how to manage or limit take before entering college.
remediation, most of the current policy efforts do not focus on
how to improve programs or help students avoid remediation Reforming remediation and better supporting the students
altogether. Improving the placement process, redesigning the who need it will be essential if the country is to improve
courses and supports, and adopting policies to help students educational attainment levels. Currently, 40 percent of first-
avoid remediation, however, are three meaningful ways to year students are placed into remediation, and most do not
improve student outcomes and increase their educational complete the courses or persist until they earn a credential. As
attainment. Improving placement policies by incorporating the national nonprofit Complete College America highlights
high school course-taking and performance information would in its 2012 report, “Remediation: Higher Education’s Bridge
reduce the chance that students are assigned to remediation to Nowhere,” the “broken remedial bridge is travelled by some
incorrectly and would help schools to better target services. In 1.7 million beginning students each year, most of whom will
addition, redesigning remediation programs with innovative not reach their destination—graduation.” Now is the time
pedagogies and support services in order to streamline the for the federal government, states, colleges, and high schools
pathway through remediation and enhance student progress to consider the growing number of promising practices and
would reduce the time needed to complete the courses and additional supports that could improve students’ chances for
improve rates of success. Finally, we could reduce the need educational success.
76 Policies to Address Poverty in America
Section 3.
Building Skills
The Hamilton Project • Brookings 77
78 Policies to Address Poverty in America
BUILDING SKILLS
Proposal 7: Expanding Apprenticeship
Opportunities in the United States
Robert I. Lerman
American University and Urban Institute
Introduction resources, and dealing with supervisors and a diverse set of
coworkers. The course work is generally equivalent to at least
Reducing inequality and expanding opportunity are central one year of community college. Completing apprenticeship
challenges increasingly acknowledged by leaders across the training yields a recognized and valued credential attesting
political spectrum. Policymakers generally agree that one key to mastery of skill required in the relevant occupation. Unlike
solution is to prepare young people and adults with the skills the normal part-time jobs held by high school and college
to earn a good income. Unlike other advanced countries, students, apprenticeship integrates what young people learn
however, reform proposals in the United States have typically on the job and in the classroom. Box 7-1 describes a successful
included little or nothing about apprenticeship—a highly youth apprenticeship program in Georgia.
cost-effective mechanism for developing workplace skills
and for reducing youth unemployment. However, interest in In some ways, apprenticeship offers an alternative to
apprenticeship models is building in the United States, partly the “academic-only” college focus of U.S. policymakers.
because of the recent successes of Britain and South Carolina Increasingly, placing all of our career-preparation eggs in
in stimulating major expansions of apprenticeship training. A one basket is leaving young adults, especially minority young
robust apprenticeship system is especially attractive because men, well behind. Among young adults ages twenty-five to
of its potential to reduce youth unemployment, improve the thirty-four in 2013, 49 percent of all women and 37 percent
transition from school to career, upgrade skills, raise wages of of African American women had earned at least an Associate
young adults, strengthen a young worker’s identity, increase degree; for men, the comparable figures were 40 percent and
U.S. productivity, achieve positive returns for employers and 28 percent, respectively.1 Furthermore, in 2011–12, nearly two
workers, and use limited federal resources more effectively. African American women earned a bachelor’s degree for every
African American male who earned one (National Center for
Apprenticeship prepares workers to master occupational Education Statistics 2013). Despite the well-documented high
skills and achieve career success. Under apprenticeship average returns to college, variations in interests, capacities,
programs, individuals undertake productive work for and learning styles suggest many young people would benefit
their employer, earn a salary, receive training primarily far more from alternative pathways to rewarding careers than
through supervised work‐based learning, and take academic they do from academic-only pathways.
instruction that is related to the apprenticeship occupation.
The programs generally last from two to four years. Apprenticeship can narrow the postsecondary achievement
Apprenticeship helps workers to master not only relevant gaps in both gender and race. Having learning take place
occupational skills, but also other work‐related skills, mostly on the job, making the tasks and classroom work highly
including communication, problem solving, allocation of relevant to their careers, and providing participants with wages
The Hamilton Project • Brookings 79
BUILDING SKILLS
Proposal 7: Expanding Apprenticeship Opportunities in the United States
BOX 7-1.
The Georgia Youth Apprenticeship Program
In 1992, the Georgia General Assembly passed a law directing the Departments of Education, Labor, and Technical Adult
Education to develop and implement youth apprenticeship programs by 1996. Today, the program operates successfully with
more than 7,000 participants.
During their freshman and sophomore years of high school, students learn about the possibility of joining the apprenticeship
program as juniors and seniors. Students can then apply to participate in a structured program of at least 2,000 hours of
work-based training and 144 hours of related coursework. Apprentices complete not only their high school diploma, but also
a postsecondary certificate or degree, and certification of industry-recognized competencies applicable to employment in a
high-skill occupation. The fields vary widely from energy to information technology, manufacturing, and transportation and
logistics. Mentorship is a key part of the program, as are employer evaluations of the student’s job performance and the building
of professional portfolios. As of 2009, more than 7,000 students in Georgia were participating in a youth apprenticeship.
High schools are responsible for recruiting and counseling students, supporting career-focused learning, and assisting in
identifying industry partners. Postsecondary schools participate in developing curriculum and dual credit arrangements.
Businesses offer apprenticeship positions, provide each apprentice with a worksite supervisor, and ensure that apprentices
gain experience and expertise in all the designated skill areas. The worksite supervisors must participate in mentor orientation
and training so that they can guide students through all the skill areas and serve as coaches and role models. Parents must
agree to and sign an educational training agreement and provide transportation to the student. Finally, apprentices must
maintain high levels of attendance and satisfactory progress in classes (both academic and career-oriented) and in the
development of occupational skills at the worksite.
Employers report high levels of satisfaction with the apprentices and the apprenticeship program. Over 95 percent say the
program has been highly beneficial to the company and that they would recommend the program to other companies.
Participating companies also report good quality student performance in problem-solving and communication skills. There
has been no rigorous evaluation of the impact of apprenticeship participation on students in Georgia, but participation has
been growing among both companies and students.
while they learn are especially beneficial to men, particularly to apprenticeship are strikingly high. U.S. studies indicate
minority men. Apprenticeship can give minorities increased that apprentices do not have to sacrifice earnings during their
confidence that their personal efforts and investment in skill education and training and that their long-term earnings
development will pay off, giving graduates a genuine sense of benefits exceed the gains they would have accumulated after
occupational identity and occupational pride. graduating from community college (Hollenbeck 2008). The
latest reports from the state of Washington show that the gains
Additionally, apprenticeship is a useful tool for enhancing in earnings from various education and training programs
youth development. Young people work with natural adult far surpass the gains from all other alternatives (Workforce
mentors who offer guidance but allow youth to make their own Training and Education Coordinating Board 2014). A broad
mistakes (Halpern 2009). Youth see themselves judged by the study of apprenticeship in ten states also documents large and
established standards of a discipline, including deadlines and statistically significant earnings gains from participating in
the genuine constraints and unexpected difficulties that arise apprenticeship programs (Reed et al. 2012).
in the profession. Supervisors provide the close monitoring
and frequent feedback that helps apprentices keep their focus On the demand side, employers can feel comfortable upgrading
on performing well at the work site and in the classroom. their jobs knowing that their apprenticeship programs will
ensure an adequate supply of well-trained workers. High
Furthermore, apprenticeship is distinctive in enhancing levels of apprenticeship activity in Australia, Canada, and
both the worker supply side and the employer demand side Britain demonstrate that even companies in labor markets
of the labor market. On the supply side, the financial gains with few restrictions on hiring, firing, and wages are willing
80 Policies to Address Poverty in America
Robert I. Lerman
to invest in apprenticeship training. While no rigorous apprenticeship could be counterproductive without a major
evidence is available about apprenticeship’s costs and benefits increase in apprenticeship slots.
to U.S. employers, research in other countries indicates
that employers gain financially from their apprenticeship Developing a more robust support system for apprenticeship
investments (Lerman 2014). programs requires action at various levels of government.
This proposal consists of a series of targeted initiatives that
In general, firms reap several advantages from their rely on both state and federal support. At the state level,
apprenticeship investments. They save significant sums in governments could develop marketing campaigns to persuade
recruitment and training costs, in reduced errors in placing employers to create apprenticeship programs, and to build on
employees, in excessive costs when the demand for skilled existing youth apprenticeship programs. At the federal level,
workers cannot be quickly filled, and in all employees being the government could provide federal subsidies to encourage
well versed with company procedures. One benefit to firms take-up of existing vouchers for apprenticeship programs;
that is rarely captured in studies is the positive impact of designate occupational standards for apprenticeship through
apprenticeship on innovation. Well-trained workers are more a joint Office of Apprenticeship (OA)–Department of
likely to understand the complexities of a firm’s production Commerce (Commerce) team; and develop an infrastructure
processes and therefore to identify and implement technological of information, peer support, and research within the
improvements, especially incremental innovations to Departments of Commerce and Labor.
improve existing products and processes. A study of German
establishments documents this connection and finds a clear
relationship between the extent of in-company training and The Challenge
subsequent innovation (Bauernschuster, Falck, and Heblich
Today apprentices make up only 0.2 percent of the U.S.
2009). In the United States, evidence from surveys of more than
labor force, far less than in Canada (2.2 percent), Britain
900 employers indicates that the overwhelming majority of
(2.7 percent), and Australia and Germany (3.7 percent). In
them believe their programs are valuable and involve net gains
addition, government spending on apprenticeship programs is
(Lerman, Eyster, and Chambers 2009). Nearly all sponsors
tiny compared with spending by other countries and spending
reported that apprenticeship programs help them meet their
on less-effective career and community college systems that
skill demands—87 percent reported that they would strongly
recommend registered apprenticeship programs, and another provide education and training for specific occupations. While
11 percent recommended apprenticeship programs with some total annual government funding for apprenticeship in the
reservations. Other benefits of apprenticeship include reliably United States is only about $100 to $400 per apprentice, federal,
documenting appropriate skills, raising worker productivity, state, and local annual government spending per participant
increasing worker morale, and reducing safety problems. for two-year public colleges is approximately $11,400 (Cellini
2012). Not only are government outlays sharply higher, but
While apprenticeship offers a productivity-enhancing approach the cost differentials are even greater after accounting for
to reducing inequality and expanding opportunity, activity in the higher earnings (and associated taxes) of apprentices
the United States has declined in recent years to levels about compared to college students. Given these data, at least some of
one-tenth of those in Australia, Canada, and Britain. Some the low apprenticeship penetration can be attributed to a lack
believe the problems include inadequate information and of public effort in promoting and supporting apprenticeship
familiarity with apprenticeship, an inadequate infrastructure, and to heavy subsidies for alternatives to apprenticeship.
and expectations that sufficient skills will emerge from
community college programs. Others see the main problem as However, the historical reasons for apprenticeship’s low
an unwillingness of U.S. companies to invest, no matter how penetration in the United States are less important than the
favorable government subsidies and marketing policies are. potential for future expansion.2 Recent experiences in Britain
In considering these explanations, we should remember that and in selected areas of the United States suggest grounds for
even in countries with robust apprenticeship systems, only a optimism, but the barriers to expansion are significant.
minority of firms actually hires apprentices. Since the number One significant barrier is limited information about
of apprenticeship applicants already far exceeds the number apprenticeship. Because few employers offer apprenticeship
of apprenticeship slots, the main problem today is to increase programs, most employers are unlikely to hear about
the number of apprenticeship openings that employers offer. apprenticeship from other employers or from workers in other
Counseling young people about potential apprenticeship firms. Compounding the problem is both the difficulty of
opportunities is a sensible complementary strategy to finding information about the content of existing programs
working with the companies, but encouraging interest in and the fact that developing apprenticeship programs is
The Hamilton Project • Brookings 81
BUILDING SKILLS
Proposal 7: Expanding Apprenticeship Opportunities in the United States
complicated for most employers, often requiring technical two to four related occupational courses. The program draws
assistance that is minimal in most of the country. The on industry skill standards and awards completers with a
experiences in Britain and South Carolina (discussed below) Certificate of Occupational Proficiency in the relevant field.
demonstrate that effective marketing is critically important Some students also receive technical college academic credit.
for expanding the number of firms offering apprenticeship In Georgia, the industry sectors offering apprenticeship range
programs. from business, marketing, and information management to
health and human services and technology and engineering.
A second barrier is employer misperceptions that The Wisconsin youth appenticeship programs are in food and
apprenticeship will bring in unions. There is no evidence natural resources, architecture and construction, finance,
that adopting an apprenticeship program will increase the health sciences, tourism, information technology, distribution
likelihood of unionization, but reports about such close and logistics, and manufacturing.
links persist. A third barrier is the asymmetric treatment
of government postsecondary funding, with college
courses receiving financial support and courses related to A New Approach
apprenticeship programs receiving little financial support.
Policies to reduce the government spending differentials Recent proposals by the administration and some members
between college subsidies and apprenticeship subsidies can of Congress suggest apprenticeship expansion would require
help overcome this barrier. substantial government funding. To support apprenticeship,
President Obama included $500 million per year for four
Another significant complication to developing more years in his fiscal year 2015 budget. Senators Tim Scott (R-
apprenticeship opportunities is that U.S. apprenticeship SC) and Corey Booker (D-NJ) have proposed providing tax
programs are categorized in three different ways: registered credits to employers hiring apprentices. Though these steps
apprenticeship with the Department of Labor’s OA, are necessary, they may not be sufficient.
unregistered apprenticeship, and youth apprenticeship.
Official data generally fail to track unregistered apprenticeship; Building a robust apprenticeship system in the United States,
evidence suggests their numbers exceed those of registered even with new resources, will require branding at the state
apprenticeship.3 Small youth apprenticeship programs operate and/or federal levels and marketing at both the general
in a few states. Furthermore, tiny budgets and an excessive level and the firm level. I suggest five strategies: two could
focus on apprenticeship in the field of construction have be accomplished at the state level, and three would be the
hampered expansion of the registered apprenticeship system. responsibility of the federal government.
The federal government spends less than $30 million annually
THE STATE ROLE
to supervise, market, regulate, and publicize the system. Many
states have only one person working under the OA. In sharp Develop high-level and firm-based marketing initiatives
contrast, Britain spends about £1 billion (or about $1.7 billion) Britain’s success in expanding apprenticeship positions from
annually on apprenticeship, which would amount to nearly about 150,000 in 2007 to over 850,000 in 2013 offers one
$8.5 billion in the United States after adjusting for population. example for how to create successful national and decentralized
Unlike programs in Austria, Germany, and Switzerland, the marketing initiatives. Alongside various national efforts,
apprenticeship system in the United States is almost entirely including the National Apprenticeship Service and Sector
divorced from high schools and serves very few workers under Skills Councils, the British government provided incentives
the age of twenty-five. Only a few states, notably Georgia and to local training organizations to persuade employers to
Wisconsin, now operate youth apprenticeship programs that create apprenticeship programs. A similar model could be
provide opportunities to youth ages sixteen to nineteen. State developed in the United States. State governments could build
funding pays for coordinators in local school systems and a state marketing campaign together with incentives and
sometimes for required courses not offered in high schools. In technical support to community colleges and other training
Georgia, 143 out of 195 school systems currently participate in organizations to market apprenticeship at the individual firm
the apprenticeship program, serving a total of 6,776 students. level. However, simply marketing to firms through existing
These apprentices engage in at least 2,000 hours of work-based federal and state agencies may not work if the staff lacks the
learning, as well as 144 hours of related classroom instruction. marketing dynamism, sales talent, and passion for expanding
The Wisconsin program includes one-year to two-year apprenticeship. Pay for performance is recommended:
options for nearly 2,000 high school juniors and seniors, technical education and training organizations would earn
requiring from 450 to 900 hours in work-based learning and revenue only for additional apprenticeship programs that each
college or organization developed with employers.
82 Policies to Address Poverty in America
Robert I. Lerman
Each apprenticeship slot stimulated by the college/training and associated tax revenues, as well as reduced spending on
organization would increase the work-based component of educational and other expenditures.
the individual’s education and training and would reduce
the classroom-based component. Assume the work-based A good place to start is with Career Academies, schools
component amounts to 75 percent of the apprentice’s learning within high schools that have an industry or occupational
program and the school-based courses are only 25 percent focus. Over 7,000 Career Academies operate in the United
of the normal student course load: by allowing training States; these programs already include classroom-related
providers to keep more than 25 percent of the standard full- instruction and sometimes work with employers to develop
time-equivalent (FTE) cost provided by federal, state, and internships in fields ranging from health and finance, to travel
local governments in return for providing the classroom and construction. Because a serious apprenticeship involves
component of apprenticeship, the community colleges and learning skills at the workplace at the employer’s expense,
other training organizations would have a strong incentive the Academies would be able to reduce the costs of teachers
to develop units to stimulate apprenticeship. State and local relative to a full-time student. If, for example, a student spends
governments could provide matching grants to fund units two days per week in a paid apprenticeship, the school should
within technical training organizations to serve as marketing be able to save at least 15 percent of its costs for that student.
arms for apprenticeship. The marketing effort should Applying these funds to marketing, counseling, and oversight
encourage government employers as well as private employers for youth apprenticeship should allow the Academy or other
to offer more apprenticeship opportunities. school to stimulate employers to provide apprenticeship
slots. Success in reaching employers will require a talented,
South Carolina’s successful example involved collaboration business-friendly staff that is well trained in business issues
between the technical college system—a special unit and apprenticeship initiatives.
devoted to marketing apprenticeship programs—and a
federal representative from the OA. With a state budget for To implement this component, state governments should fund
Apprenticeship Carolina of $1 million per year, as well as marketing and technical support to Career Academies to
tax credits to employers of $1,000 per year per apprentice, set up cooperative apprenticeship programs with employers,
the program managed to stimulate a six-fold increase in using either state or federal dollars. The first step should be
registered apprenticeship programs and a five-fold increase planning grants for interested and capable Career Academies
in apprentices. Especially striking is that these successes— to determine who can best market to and provide technical
including 4,000 added apprenticeship opportunities—took assistance to the Academies. Next, state governments should
place as the economy entered a deep recession and lost millions sponsor performance-based funding to units in the Academies
of jobs. The costs per apprentice totaled only about $1,250 each so they receive funds for each additional apprenticeship.
calendar year, including the costs of the tax credit. Private foundations should offer resources for demonstration
and experimentation in creating apprenticeship opportunities
Build on youth apprenticeship programs within high school programs, especially Career Academies.
State government spending on youth apprenticeship programs THE FEDERAL ROLE
amounts to about $3 million in Georgia and $2 million in
Wisconsin. Although these programs reach only a modest Extend use of current postsecondary and training subsidies to
share of young people, the United States could make a good apprenticeship
start on building apprenticeship programs if the numbers Several postsecondary programs could be set up to subsidize
in Georgia could be replicated throughout the country. The at least the classroom portion of apprenticeship. Already,
focus would be on students who perform better in work-based localities can use training vouchers from the Workforce
settings than in purely school-based ones and who are less Investment Act for apprenticeship programs. To encourage
likely than the average student to attend a four-year college or greater use of vouchers for apprenticeship, the federal
complete a bachelor’s degree. To create about 250,000 quality government could provide one to two more vouchers to
jobs and learning opportunities, the gross costs of such an Workforce Investment Boards for each training voucher used
initiative would be only about $105 million—about $450 per in an apprenticeship program. Another step is to encourage the
calendar year—or about 4 percent of current school outlays use of Trade Adjustment Act training subsidies to companies
per student-year. Moreover, some of these costs would be offset sponsoring apprenticeship, just as training providers receive
by reductions in teaching expenses, as more students would subsidies for Act-eligible workers enrolled in full-time
spend more time in work-based learning and less time in training. In addition, policies could allow partial payment of
high school courses. In all likelihood, the modest investment the Act’s extended unemployment insurance to continue for
would pay off handsomely in the form of increased earnings employed individuals in registered apprenticeship programs.
The Hamilton Project • Brookings 83
BUILDING SKILLS
Proposal 7: Expanding Apprenticeship Opportunities in the United States
Allowing the use of Pell Grants to pay at least for the classroom other countries, along with the list of occupation skills that
portion of a registered apprenticeship program makes perfect the apprentices master. The clearinghouse should include the
sense as well. Currently, a large chunk of Pell Grants pays for curricula for classroom instruction, the skills that apprentices
occupationally oriented programs at community colleges and should learn and ultimately master in the workplace, and
for-profit career colleges. The returns on such investments are up-to-date information on available apprenticeship slots and
far lower than the returns on apprenticeship. The Department on applicants looking for apprenticeship opportunities. The
of Education can authorize experiments under the federal development of the information hub should involve agencies
student aid programs (Olinsky and Ayres 2013), allowing within Commerce as well as in the OA.
Pell Grants for some students learning high-demand jobs
as part of a certificate program. Extending the initiative to The research program should cover topics especially relevant
support related instruction (normally formal courses) in an to employers, such as the return to apprenticeship from
apprenticeship could increase apprenticeship slots and reduce the employer’s perspective and the net cost of sponsoring
the amount that the federal government would have to spend an apprentice after taking into account the apprentice’s
to support these individuals in full-time schooling. contribution to production. Other research should examine
best practices for marketing apprenticeship programs,
The GI Bill already provides housing benefits and subsidizes incorporating classroom and work-based learning by sector,
wages for veterans in apprenticeship programs. However, and counseling potential apprentices.
funding for colleges and university expenses is far higher than
for apprenticeship. Offering half of the GI Bill college benefits COSTS AND BENEFITS
to employers hiring veterans into an apprenticeship program The proposals in this paper would involve only a modest
could be accomplished by amending the law. Unless the amount of new funding, though some shift in the allocation
liberalized uses of Pell Grants and GI Bill benefits are linked of funds for the education and training marketplace would be
with an extensive marketing campaign, however, the take-up necessary. To date, apprenticeship programs have not proven
by employers is likely to be limited. to be very expensive for the government; the majority of
costs stem from the federal and state costs of administering
Designate best practice occupational standards for apprenticeship programs, tuition paid by participants,
apprenticeships instruction costs related to the academic portions of the
To simplify the development of apprenticeship for potential programs, and those borne directly by taxpayers through
employers, a joint OA–Commerce team should designate one higher spending or forgone tax revenue (Reed et al. 2012;
or two examples of good practice with regard to specific areas Workforce Training and Education Coordinating Board 2014).
of expertise learned at work sites and with regard to subjects
learned through classroom components. The OA–Commerce A recent study of apprentices in the state of Washington gives
team should select occupational standards in consultation with an indication of the potential costs and benefits associated
selected employers who hire workers in the occupation. Once with an apprenticeship program (Workforce Training and
selected, the standards should be published and made readily Education Coordinating Board 2014). The average cost per
accessible. Employers who comply with these established participant borne by the individual and government was
standards should have a quick and easy path to the registration about $5,500. In contrast, the per-participant cost associated
of the program. In addition, workforce professionals trying to with participation in a community college professional or
market apprenticeship will have a model that they can sell and technical program are about $16,000 per year.
that employers can adopt, either as-is or after making modest
The potential benefits, as indicated by this study, are stunning:
adjustments. Occupational standards used in other countries
apprentices raised their earnings relative to a comparison
can serve as starting points for the OA–Commerce team
group by an average of nearly $78,000 over two and a half
and for industry groups involved in setting standards and in
years after leaving the program. In comparison, participants
illustrating curricula.
in community college professional or technical programs
Develop a solid infrastructure of information, peer support, netted only about $15,000 in increased earnings. Projecting
and research earnings effects through age sixty-five, these relative earnings
for apprentices amount to roughly $440,000 at a cost of
The federal government should sponsor the development $5,500; the comparison figures for participants in community
of an information clearinghouse, a peer support network, college professional or technical programs are $175,000 at a
and a research program on apprenticeship. The information cost of about $20,000 (Workforce Training and Education
clearinghouse should document the occupations that Coordinating Board 2014). A separate study conducted
currently use apprenticeship in the United States and in
84 Policies to Address Poverty in America
Robert I. Lerman
by Deborah Reed and colleagues (2012) of ten other states but we are highly unlikely to achieve growth without trying
found earnings gains associated with apprenticeship training something along the lines of the proposals in this paper.
amounting to $6,000–$6,500 per year per participant. In
addition to these quantitative benefits, apprenticeship— Will enough workers apply for the additional apprenticeship
in particular registered apprenticeship—also results in slots?
numerous social benefits, including added productivity of Compared to expanding the demand for apprentices, increasing
workers, reduced use of government safety-net programs by supply by attracting sufficient applicants for apprenticeship
participants, and a stronger local economy. is likely to be relatively easy. Although representative data
Two studies of the earnings gains of apprentices and on the number of applicants per apprenticeship slot do not
government costs in the United States find that the social exist, many examples indicate that the number of applicants
benefits outweigh the social and government costs by ratios is far higher than the number of apprenticeship openings.
of 20:1 to 30:1 (Reed et al. 2012; Workforce Training and Take the case of the Apprenticeship School, a program linked
Education Coordinating Board 2014), although the extent to to the shipbuilding tasks of a company in Newport News,
which these benefits are due to government investment as Virginia.4 In 2013, the school had over 6,000 applicants for
compared to employer investment is indeterminate. Investing about 240 positions. Most craft apprenticeship programs in
in extensive marketing aimed at increasing apprenticeship the building trades have far more applicants than apprentice
with appropriate incentives for performance will add only slots. The case of Britain offers additional evidence: the
modestly to government costs while yielding substantial gains massive increase in intermediate or advanced apprenticeship
for workers and the public. positions between 2007 and 2013 was matched by a sufficient
increase in applicants. Nonetheless, providing counseling and
Given the high share of apprenticeship programs undertaken information to prospective apprentices will still be a sensible
through joint union-employer agreements, some share of the investment, especially after an expansion in apprenticeship
earnings gains associated with apprenticeship may actually slots, because a good matching process is critical for the
result from the role of unions in bargaining for higher wages. effectiveness of the program for workers and firms.
Still, workers must have raised their productivity enough
through their apprenticeship in order for employers to afford What role does public perception play in the expansion of
to pay union wages. On the cost side, construction unions apprenticeship opportunities?
and both union and non-union employers certainly invest Public perception and awareness of apprenticeship could
large sums in training apprentices. Manufacturing companies play a major role in its expansion in the United States. In
that train apprentices do so as well. This stimulus to private
the United Kingdom, for example, a large shift in public
investments is one of the reasons apprenticeship increases
perception occurred over the past few decades, leading to a
earnings at a modest cost to the government.
series of pro-apprenticeship campaigns that coincided with
a rapid increase in apprenticeship. In the year following the
implementation of a marketing campaign in London, the
Questions and Concerns number of apprentices in the city more than doubled from
Will enough employers offer apprenticeship positions? 20,350 to 41,400 (Evans and Bosch 2012). Furthermore, a £25
million public apprenticeship fund introduced in 2010, which
Stimulating a sufficient increase in apprenticeship slots is the included a marketing component, coincided with a near
most important challenge. Although it is easy to cite examples doubling of apprenticeship starts in England—from 279,700
of employer reluctance to train, the evidence from South to 520,600—between the 2009–10 and 2011–12 academic years
Carolina and Britain suggests that a sustained, business- (Skills Funding Agency 2014).
oriented marketing effort can persuade a large number of
employers to participate in apprenticeship training. Both Importantly, too, is the culture surrounding both the teaching
programs were able to more than quadruple apprenticeship and learning aspect of apprenticeship. In the United States,
offers over about five to six years. Today, U.S. employers registered apprenticeship in the building trades industry
are far less likely to offer apprenticeship programs than are have been present for more than 100 years and are an integral
their counterparts in many other advanced economies. part of the training for construction-related occupations.
One reason is that federal and state governments have not Many workers in these industries are accustomed to their
provided adequate resources to encourage and help employers role as mentor and teacher. As apprenticeship becomes more
adopt apprenticeship programs. New policies may or may common in other industries, the apprenticeship model—
not succeed in generating significant growth in apprentices, which relies heavily on the participation of existing workers—
The Hamilton Project • Brookings 85
BUILDING SKILLS
Proposal 7: Expanding Apprenticeship Opportunities in the United States
may become a familiar and welcome of aspect of employment Community college graduation rates, especially for low-
in other industries. income students, are dismally low. Even after graduating,
they often have trouble finding a relevant job. For students
Will apprenticeship programs accept disadvantaged in postsecondary education, forgone earnings are one of
workers? the highest costs. In contrast, participants in apprenticeship
programs rarely lose earnings and often earn more than if they
Apprenticeship can play a role in helping the disadvantaged,
had not entered an apprenticeship. Furthermore, apprentices
but not all will benefit. As noted above, apprenticeship
are already connected with an employer and can demonstrate
promotes youth development and provides a pathway to
the relevant credentials and work experience demanded
rewarding careers that is less reliant on classroom instruction.
by other employers. Finally, there are net gains flowing to
This approach is particularly relevant to the learning
employers from apprenticeship programs.
processes of men, especially minority men. In many cases,
employer requirements will limit the opportunity of the The key question is not whether the shift in emphasis from
most educationally disadvantaged from entering various community and/or career colleges toward apprenticeship
professions. Of course, exclusions of this type occur even is desirable, but whether it is feasible. Although some argue
without an apprenticeship. Still, apprenticeship is attractive that the free U.S. labor market and the weak apprenticeship
even to disadvantaged workers because they provide clear tradition pose insurmountable barriers to scaling-up
incentives for low-performing students to work hard to apprenticeship, the dramatic increases in apprenticeship
attain adequate skills to qualify for apprenticeship that leads in Britain offer strong evidence that building a robust
to career jobs that pay well. Finally, there is a distribution of apprenticeship program in the United States is feasible.
apprenticeship occupations; some occupations might not
require advanced education yet still involve apprenticeship The first step is persuading policymakers and employers
that leads to attractive careers. about the desirability and feasibility of apprenticeship.
Once that intellectual hurdle is overcome, the next step is
establishing leadership at the policy and program levels and
Conclusion effective implementation of the new approach. Institutional
change of this magnitude is difficult and will take time, but
Expanding apprenticeship is a potential game-changer
will be worthwhile in terms of increased earnings, enhanced
for improving the lives of millions of Americans and for
occupation identity, increased job satisfaction, and expansion
increasing the efficiency of government dollars spent on
of the middle class.
developing the workforce. Instead of spending over $11,000
per year on students in community college career programs,
why not shift resources toward apprenticeship programs,
which are far more cost-effective? Apprenticeship programs
yield far higher and more-immediate impacts on earnings
than community or career college programs, yet cost the
student and the government far less than college programs.
86 Policies to Address Poverty in America
BUILDING SKILLS
Proposal 8: Improving Employment Outcomes
for Disadvantaged Students
Harry J. Holzer
Georgetown University and
American Institutes for Research
Introduction clearly benefit from having more postsecondary education
or training options that they can successfully complete, and
Improving the skills and earnings potential of poor youth that are more closely linked to the needs of employers in high-
and adults should remain a top priority for state and federal demand fields that pay well.
policymakers. Poorer people lag behind their more affluent
peers in both postsecondary educational attainment and Colleges can expand course offerings in high-demand fields of
earnings, and raising both would contribute strongly to study, but there are other approaches as well to better align
reducing poverty among current and future generations. educational skills with the current labor market. One such
Tapping the full potential of public colleges to provide a leg up approach is sectoral training, in which education providers
to those who need the educational push could go a long way work with employers to educate and train directly for the job
toward alleviating poverty. requirements of high-demand sectors. This approach appears
to have large impacts on earnings in rigorous evaluations.
Students from all family backgrounds already face strong Career pathways are also being developed for these sectors that
financial incentives to pursue postsecondary education. In combine classroom education and work experience in a series
response to the higher earnings of college graduates relative of steps that ultimately lead to these jobs. And other models
to those without college, U.S. enrollment rates have risen of work-based learning, such as apprenticeships or incumbent
dramatically in the past decade, especially during the Great worker training, can accomplish many of the same goals.
Recession, and degree attainment has increased somewhat
at both two- and four-year colleges (Greenstone and Looney Many states and localities are trying to build education and
2011; National Student Research Clearinghouse 2011). We training programs in both four-year and community colleges,
have also greatly increased the nation’s investment in Pell especially in high-demand fields, and bring them to scale. A
Grants and other forms of assistance to improve college access report by the National Governors Association (2013) finds
for the poor (Holzer and Dunlop 2013). that at least twenty-five states are now building partnerships
between key employers or industry associations and
But the dropout rate among low-income youth and adults community colleges for sectoral training and career pathway
in college remains extremely high; even among those who development, and are trying to integrate these programs with
complete certificates or degrees, many choose fields of study their broader economic development goals.1
that are not well compensated in the labor market (Bound,
Lovenheim, and Turner 2009; Robst 2007). These outcomes Anecdotes abound about partnerships and programs developed
hurt the poor, and weaken the impacts of large national in specific industries at the state level. But we have few data so
investments in higher education. Low-income students would far indicating the scale and outcomes achieved, much less data
The Hamilton Project • Brookings 87
BUILDING SKILLS
Proposal 8: Improving Employment Outcomes for Disadvantaged Students
on the impacts on the education or employment outcomes of are likewise very low: less than 25 percent of young students
the disadvantaged students engaged in these efforts. When at community colleges, and even fewer among older ones,
considering future investments, maintaining both the quality complete an AA degree. These rates are low among poor and
achieved in the smaller evaluated programs and a focus on the nonpoor students alike, but many more poor students attend
poor remain important, so as not to simply provide windfalls these colleges (Holzer and Dunlop 2013).
to employers at taxpayer expense.
While there are various factors behind these high dropout
To improve earnings prospects for recent graduates and to rates, one important factor may be the perceived imbalance
encourage two- and four-year colleges to be responsive to labor between the costs of attending college—including the
market demand, I propose that state legislatures implement opportunity cost of forgone employment—relative to the
financial incentives for colleges to steer students toward high- perceived benefits. Low-income students might not enroll in
wage occupations and to industries with especially high labor or complete degree programs in these highly compensated
needs. In addition, while this proposal primarily calls for fields if they lack the information about which fields are well
state-level reforms, I also note opportunities for the federal compensated or about which fields are in high demand among
government to support states in this initiative. those that they could actually complete successfully (Jacobson
and Mokher 2009; Scott-Clayton 2011).
My proposal calls on states to partially base college funding
on graduates’ reported wages five years following graduation By most accounts, community colleges vary enormously
and, where appropriate, on the colleges’ provision of courses in their quality and commitment to responding to labor
that are especially important to the local economy. These demand. Some are torn between their more traditional
incentives may also be accompanied by technical assistance for academic missions of being feeder programs to four-year
states and colleges, plus supports for students. I also propose colleges and their newer vocational missions. In addition,
that states experiment with generating financial incentives for as public institutions that are mostly paid (through state
employers to engage more with colleges in sectoral efforts, and subsidies or private tuition payments) for student “seat
propose that employers expand their own efforts to train and time” rather than education or employment outcomes, they
hire more workers. have little incentive to respond to labor market need. In at
least some fields of study (e.g., nursing, health technology,
and advanced manufacturing), the costs of equipment and
The Challenge instructors are relatively high, deterring community colleges
from building adequate instructional capacity in these areas.
It is widely known that the earnings of less-educated workers— And the instructors they hire may have little incentive to
i.e., those with high school or less education—have greatly keep up effectively with newer developments in dynamic
lagged behind those of more-educated workers in the past few fields such as information technology. Accordingly, students
decades. We also know that poorer people tend, on average, to report difficulty enrolling in classes they need for their majors
have low levels of education and achievement (Greenstone et in such fields. And the high-skill requirements in some of
al. 2013). And their children’s education and achievement lags these areas—such as the math requirements for machinists in
behind as well, with the gaps apparently growing wider over manufacturing—also preclude efforts to expand participation,
time (Bailey and Dynarski 2011; Reardon 2011). especially among disadvantaged students.
While postsecondary enrollments have risen across the entire This challenge has not been resolved by the recent expansion
income spectrum, it is also evident from data sources that in for-profit educational institutions.2 While for-profits are
dropout rates are very high among low-income students, sometimes described as institutions that serve disadvantaged
especially at community colleges and non-elite four-year students and rapidly respond to changing labor market
colleges. For instance, calculations from the National demands (Deming, Goldin, and Katz 2013), these institutions
Educational Longitudinal Survey (NELS) data show that have thus far been largely unsuccessful at appreciably
students from disadvantaged socioeconomic backgrounds improving postgraduation earnings for low-income students
struggle to complete higher education: only about 30 and filling gaps in training and education left by public-
percent of the students from the lowest quartile of families sector colleges. Evidence suggests that students who attend
by socioeconomic status who enroll in four-year colleges for-profit colleges are likely to experience lower earnings,
complete their bachelor of arts degrees within about six years, higher unemployment, and higher student debt burdens than
less than half of the completion rate of the overall student they would if they attended public institutions (Deming,
population (Holzer and Dunlop 2013). This cohort’s associate Goldin, and Katz 2013). In addition, for-profits operate with
of arts (AA) degree completion rates at community colleges
88 Policies to Address Poverty in America
Harry J. Holzer
a motivation for generating profits, not for maximizing the at least partially dependent on student performance in the job
social benefits of education and meeting local labor demand. market. In addition, states can incent colleges to offer courses
While they appear to generate some benefits, for-profit and majors that would better align unmet labor demand
institutions are, at best, a highly imperfect way to accomplish with labor supply in local markets. This strategy will reward
the policy goals of these proposals. colleges for focusing more on the labor market outcomes
of their students, leading to higher earnings for graduates
Instead, a more prudent approach is to create financial and stronger local economies. While these proposals are
incentives similar to those that motivate for-profit colleges to not specifically targeted to low-income individuals, a large
supply high-demand classes in public institutions. This would segment of the program beneficiaries will be in the lower part
likely avoid at least some of the negative outcomes associated of the income distribution, making these policies an effective
with for-profit institutions as suppliers of education to the antipoverty initiative.
disadvantaged but would still target class offerings toward
those in high-return fields. As it stands, public institutions State governments can incent public colleges to improve
already rely on public funds and make course offering graduates’ outcomes through a primary mechanism that
decisions based in part on the set of incentives that results partially ties funding to postgraduation reported earnings.
from the structure of those subsidies. The prudent path would Specifically, the earnings of students over the subsequent
be to structure these subsidies in a way that will likely result five-year period beyond graduation would form the basis to
in colleges offering the classes that students demand and that reward states. Extra subsidies could be granted, for example,
will generate the highest level of social and economic benefits. to colleges whose students subsequently earn above the
median level for those with such a credential in that state;
Indeed, several states have already instituted incentives those whose students have relatively low earnings would get
for colleges that are tied to performance. According to the lower subsidies. Rewarding the earnings of minority or low-
National Conference of State Legislatures (2014), at least income students would be critically important as well, since
twenty-five states have embraced some form of performance- these are the students whose employment in high-demand
based subsidies for their public colleges, and another five are fields most lags behind. Allowances would be made for
planning to implement such policies. Table 8-1 summarizes graduates who transferred to four-year universities or who
some recent information from the National Conference of sought further education.
State Legislatures on what some states are doing in this area. It
shows that most of these states reward colleges for successful In local labor markets that demonstrate a major imbalance
course completion, credits earned, and ultimately credentials between labor demand and labor supply, colleges might also
earned (or successful transfers to four-year colleges). And be rewarded for putting students in high-demand occupations
at least some of the states additionally reward schools for and industries within that locality or state. For example, such
successful outcomes achieved among their low-income occupations or industries might include nursing, health
or minority populations, which would be an important technology, or advanced manufacturing. In general, this
consideration for antipoverty policy. mechanism is inferior to tying colleges’ funding to wages,
which represent the market valuation for various types of
But, to date, such policies focus only on student academic work. However, in local markets that seem to experience labor
outcomes at the colleges and universities, rather than those shortages—such as being unable to fill nursing slots or to find
that occur afterwards in the labor market. Basing subsidies at qualified special-education teachers—there is an argument for
least partly on job market outcomes will make public colleges using public funds to incent specialized training.3
more responsive to labor demand, especially by building higher
instructional capacity and hiring high-quality instructors States could also implement concurrent reforms that help
in high-demand fields. And, as noted below, it might also be colleges better target their curriculum. For example, states
helpful to encourage more employers to participate in such could provide technical assistance to their public colleges
partnerships with colleges, or to directly train, or at least hire, as they implement reforms that better align courses to the
more workers who are somewhere on their career pathways. labor market. Networks of states that are developing sectoral
programs and career pathways on a larger scale are working
with supporting organizations to provide guidance on how to
A New Approach best implement these changes (Choitz 2013).
My proposal calls on states to encourage public colleges to be In addition, a set of supports for students—such as career
more responsive to students’ prospects and the job market. and labor market counseling—would likely raise completion
One way to accomplish this is by making their public subsidies rates and the earnings of program graduates. Simplifying
The Hamilton Project • Brookings 89
BUILDING SKILLS
Proposal 8: Improving Employment Outcomes for Disadvantaged Students
TABLE 8-1.
Some States Using Performance-Based Higher Education Subsidies
State Funding amount Metrics measuring performance Type of Administrative body
institution
Arizona For FY 2013 and FY 2014, $5 million Metrics are based on degrees awarded, In place at four- Arizona Board of
per year was allocated based on completed student credit hours, and external year institutions Regents
performance. research and public service dollars brought
into the university system.
Colorado Beginning in 2016-17 and for each Metrics are based on attainment, student In transition Each institution’s
year that state funding is at or above success, diversity in enrollment, reducing governing board
$706 million, 25% of the amount over attainment gaps among students from negotiates a contract
$650 million will be appropriated underserved communities, and financial with the department of
based on each institution’s stewardship. Institutions then design higher education
performance. separate sets of common and institution-
specific metrics.
Georgia Beginning in FY 2017, all new money Metrics are based on student progression, In transition Higher Education
appropriated will be allocated based degrees conferred, success of low-income Funding Commission
on institutional performance. and adult learners, and institution-specific
success on strategic initiatives.
Illinois Funding amount is less than 1% of For four-year universities, metrics are In place at two- Illinois Board of Higher
base funding. based on bachelor’s, master’s, doctoral, and four-year Education’s Higher
and professional degrees, undergraduates institutions Education Performance
per 100 FTE, research and public service Funding Steering
expenditures, graduation rate, and cost of Committee
attendance. For two-year universities, metrics
are based on degree completion, completion
rates for "at risk" students, transfers to
four-year institutions, remedial and adult
education, momentum points, and diversity.
Kansas New state higher education funds Metrics are specific to each institution, but In place at two- Kansas State
are allocated based on performance must be selected from a proscribed list of and four-year Board of Regents,
incentives. performance indicators. institutions contingent upon each
institution meeting
its individualized
Performance Agreement
Maine Performance funding will start as Metrics are based on degrees awarded, In place at four- University of Maine
5% of base funding in FY 2014, prevalence of STEM and priority fields, year institutions System Board of
and increase by 5% increments number and dollar value of research grants Trustees
each subsequent year until it and contracts received, and degrees
reaches 30%. awarded per $100,000 of net tuition and fee
revenues.
Michigan For FY 2014, $21.9 million in new Metrics are based on completions in critical In place at two- Performance funding is
appropriations for universities skill areas, research and development and four-year included in the annual
and $5.8 million for community expenditures, graduation rates, institutional institutions higher education
colleges was allocated based on support as a percent of core expenditures, appropriations
performance metrics. with mandatory requirements of limiting
resident tuition increases to 3.75% per
year, participation in at least three reverse
transfer programs with community colleges,
maintaining a dual enrollment credit policy,
and participation in the Michigan Transfer
Network. Separate allocation criteria exist for
community colleges.
90 Policies to Address Poverty in America
Harry J. Holzer
State Funding amount Metrics measuring performance Type of Administrative body
institution
Mississippi After a base amount is set aside The Board of Trustees sets priorities based In place at four- Mississippi Public
for operational support, 90% is on a range of specified metrics relevant year institutions Universities Board of
allocated based on completion to attainment, intermediate educational Trustees
targets and 10% is allocated outcomes, research, and productivity.
based on progress toward
established priorities.
Montana 5% of base funding will be at stake Metrics are currently being developed, but In transition Montana University
during the FY 2015 trial phase. The are expected to vary based on the mission System Performance
amount of performance funding of each institution and include measures of Funding Steering
for long-term priorities has not yet completion and retention. Committee
been determined.
New Mexico Performance-based funding is Metrics are based on number of certificates In place at two- Performance
set to increase, but is currently and degrees awarded in both general and and four-year funding is included
5% of instruction and general priority areas, degrees earned by at-risk institutions in the annual
formula funding to colleges and students, grant/contract funding, and higher education
universities. momentum points. appropriations
North Dakota Nearly all base funding is The funding formula is based on the number In place at two- The state board of
calculated by the number of credit of credit-hours completed by students. A and four-year education, based
hours completed. completed credit-hour is one for which a institutions on per-credit dollar
student met all institutional requirements amounts specified in
and obtained a passing grade. legislation
Oklahoma Performance funding only applies Metrics are based on first-year retention, In place at two- Oklahoma State
to new appropriations. first-year retention for Pell recipients, student and four-year Regents for Higher
completion of twenty-four credits in their institutions Education
first academic year, cohort graduation rates
anywhere in the system, degrees granted, and
program accreditation.
South Dakota Nearly all base funding is Metrics are based on funds appropriated In transition Council on Higher
calculated by the number of credit according to degrees awarded, STEM Education Policy
hours completed. degrees awarded, and growth in research Goals, Performance,
expenditures. and Accountability
Texas Funding amount is 10% of 10% of the formula funding is allocated based In place at two- Higher Education
formula-based state higher on developmental education completion rates, year institutions Coordinating Board
education funding. number of students who complete first college
level course in mathematics, reading intensive
and writing intensive courses, interim student
attainment, and number of degrees and
certificates awarded, with additional points
awarded for degrees in STEM or allied
health fields.
Virginia 50% of funding is expected to be Metrics are based on number of degrees In transition State Council of
allocated based on performance awarded and number of additional degrees Higher Education
and incentive funding. awarded each year with emphasis on STEM
attainment, degrees earned within 100% of
time-to-degree, and degrees awarded to
students from under-represented populations.
Source: National Conference of State Legislatures 2014.
Note: FY = fiscal year; FTE = full-time equivalent; STEM = science, technology, engineering, and mathematics.
The Hamilton Project • Brookings 91
BUILDING SKILLS
Proposal 8: Improving Employment Outcomes for Disadvantaged Students
financial aid, and conditioning it on maintaining some educational and earnings outcomes. As noted below, states
adequate level of academic performance, tends to have are increasingly generating the data needed to observe these
positive effects on student outcomes. Accelerating outcomes; to measure the full scale of programs effects,
remediation efforts, and combining them with labor market however, data on student participation in occupational
information and other kinds of supports, would likely help programs must also be included (Choitz 2013).
as well (Bettinger, Boatman, and Long 2013).
More broadly, the use of state labor market data to monitor
While simply creating rewards to institutions and employers employment growth across sectors (as well as job vacancy data
based on outcomes might be sufficient, the federal or state to indicate where firms are having difficulty filling available
governments might also help by paying for some of these jobs) might enable colleges to better target sectors in which
supports directly, or by helping to make them more easily demand remains somewhat unmet, and where investments
available. For instance, high-quality career counseling might in training would be most useful. Keeping track also of the
be more available to students at community colleges if the Job full range of credentials achieved by workers, including
Centers (formerly known as One-Stop offices) funded by the those provided and recognized by employers and others,
U.S. Department of Labor were increasingly colocated with is important so that the supply of skills can be measured as
college campuses or if Job Centers increased the number of well as the demand. Finally, states should also evaluate these
staff available to counsel students.4 programs regularly to see whether their impacts justify
ongoing expenditures.
IMPLEMENTATION OF THE PROPOSAL
Although most of these subsidies to public colleges and
This proposal would be implemented by state legislatures in
participating employers will be financed by states, the
their ongoing budget processes. States would explicitly tie a
federal government could help as well. For instance, the
specified share of two- and four-year college funding to the
Obama administration plans to implement the last round
reported earnings of graduates in the five years following their
of competitive grants in 2014 in its Trade Adjustment
graduation. In states with specialized labor force needs, state
Assistance Community College to Career program, worth
legislatures could also introduce additional funding criteria
$500 million. It has already given out $1.5 billion in three
based on labor force outcomes in designated industries or
previous rounds, with the funding going almost exclusively
occupations. Since many states are already tying subsidies to
to individual community colleges or consortia of colleges
academic outcomes of students, this proposal would call for
in each state. There are plans to partner with specific high-
approximately half of all incentive payments to be based on
demand and high-wage industries. In the last round, the
the subsequent labor market outcomes of students, while the
administration will hopefully reward states directly that
other half might continue to be based on academic outcomes.
institute some of the performance measures described
The share of funding explicitly tied to employment (as well above, or offset some of the state financial supports for
as academic) outcomes will vary by state. Existing state participating employers. In addition, the administration
structures have varied considerably in this respect: Tennessee, could use its Workforce Innovation Funds in the Department
for instance, is already transitioning to making student of Labor to encourage such state activities, or some of
performance the entire basis of its higher education subsidies, the new grants proposed in its FY 2015 budget (Office
while Texas bases just 10 percent of its funding on various of Management and Budget 2014). It might also consider
educational measures. As recommended by the National using some of these funds to offset additional expenses
Conference of State Legislatures, states may benefit from incurred by the states (or their colleges) in developing the
gradual implementation of their performance-based subsidies, new data systems and analysis that would be needed to
with small but steady increases over time in the percentages implement these proposals, to prevent them from viewing
of funding based on performance, as Maine is doing. States these changes as something like an “unfunded mandate.”
might also decide to implement these approaches to all new
EXPERIMENTING WITH INCENTIVES FOR
or additional funding above some base level, which Georgia,
EMPLOYERS
Mississippi, and Oklahoma appear to be doing.
Another challenge that might limit the effectiveness of
Quality assurances should also be considered. For instance, education or training aimed at high-demand sectors is the
it is crucial that the states, their local workforce boards, and reluctance of employers to participate in partnerships with
their colleges carefully monitor the progress associated with colleges and to hire their trainees, or to directly train more
these additional investments, by measuring the numbers of workers themselves.
students that participate in these programs, as well as their
92 Policies to Address Poverty in America
Harry J. Holzer
Employer decisions on whether to train workers involve a high completion rates and large impacts on the earnings of
set of considerations. As Gary Becker (1996) pointed out, adults and youth who complete them, such as the sectoral
employers have little incentive to invest in general training programs mentioned above, where education providers work
for workers who might leave at any time. If they question the closely with employers to train workers for existing jobs.
quality of the workers’ basic skills, and their ability to handle
technical material, they have even more reason to avoid such Maguire et al. (2010) provide evidence from a randomized
investments. Some employers provide such training mostly control trial study of three such programs: Jewish Vocational
to their professional and managerial employees (Lerman, Services in Boston, which trains disadvantaged workers for
McKernan, and Riegg 2004). Many training models, such as careers in health care; Per Scholas in New York, which focuses
apprenticeships and internships and other models of work- heavily on IT services; and the Wisconsin Regional Training
based learning, require less-educated workers to largely pay Partnership, which prepares trainees for jobs in construction,
for general training out of their own wages.5 manufacturing, and other industries. In addition, Roder
and Elliott (2012) used randomized control trial evidence to
To incent more private sector employers to engage in sectoral evaluate Year Up, a program that trains youth for jobs in IT
partnerships and provide employment and/or training to and related industries.
workers, states could offer tax credits or subsidies per employee
hired or trained in this fashion. Though the evidence to date All of these programs, which take roughly six months to
on tax credits for employers who hire or train workers is complete, generated large impacts on earnings (of roughly
somewhat weak, experimentation by states could add to the $4,000 per year, or about 30 percent higher than earnings of
available pool of knowledge about what works (or does not the control groups) within two years of random assignment.
work) in this area.6 Though some important questions remain about whether the
impacts fade out over time (especially when workers change
How might such tax credits or subsidies be structured? jobs and move across sectors as well as firms), and exactly
Activities that cost employers more, such as direct provision who is served by these programs (some require at least a
of training to new hires (or incumbent workers), might high school diploma or a GED), these impacts compare very
require relatively higher subsidies, while simply hiring those favorably with other education or training programs (Holzer
trained by a local community college or other providers might 2013). And, though the training providers in these programs
require lower subsidies. Those who implement apprenticeship were generally not community colleges, other well-known
programs, or other models of work-based learning, might sectoral efforts (like Quest in San Antonio) rely more heavily
need some encouragement if some of the costs cannot be on colleges to provide training.
passed on to the worker or if administrative hurdles are posed.
By limiting the subsidies to students with only a high school Other evidence also shows large impacts on earnings from
diploma or GED at the outset, states could more effectively other approaches, including work-based learning (from
target their lower-income populations with these policies apprenticeship or incumbent worker training). Some of this
without stigmatizing them as efforts for the poor only. evidence is based on careful matching studies, rather than
on randomized control trials, so they should be viewed as
It is still unclear how large tax credits or subsidies should suggestive rather than conclusive, but they are encouraging
be to successfully encourage employer participation. But nonetheless (Hollenbeck 2008, 2012; Reed et al. 2012). Even
Hollenbeck (2008) reports that spending under $1,000 per remediation programs in community colleges appear to be
worker in participating firms was sufficient to generate more more successful when they integrate labor market information
incumbent worker training in Massachusetts. Holzer and or skills training directly into the remedial classes, as has been
Lerman (forthcoming) also report that South Carolina now done in the LaGuardia Community College’s GED Bridge
offers employers $1,000 per apprentice, though we need more program in New York and the Integrated Basic Education and
evidence on its impacts. Total costs can be further reduced, Skills Training program in Washington state (Martin and
for example, by limiting such tax credits to employees with Broadus 2013; Zeidenberg, Cho, and Jenkins 2010).
less than bachelor degrees in entry-level nonmanagerial jobs.
COSTS AND BENEFITS
EVIDENCE OF EFFECTIVENESS
Perhaps the most obvious private benefits for students who
There is no doubt that improving the extent to which low- experience a better targeted public college curriculum are
income students gain high-education credentials will raise higher earnings and improved employment prospects.
their earnings. Rigorous evidence exists on the kinds of Jacobson, LaLonde, and Sullivan (2005) find that one year
education and workforce development programs that have of technically oriented community college education raises
The Hamilton Project • Brookings 93
BUILDING SKILLS
Proposal 8: Improving Employment Outcomes for Disadvantaged Students
earnings by 14 percent for men and 29 percent for women— training of 1.3 million workers nationally, though it is not
at least for a sample of displaced workers. Similar impacts clear how much of this training represents net impacts of the
have been found for sectoral workplace training programs. As expenditures. An estimated expenditure of $2 billion a year
noted above, experimental studies of the impact of training nationally by federal and state governments could therefore
programs showed wage increases of approximately 30 percent. be associated with the occupational training for as many as 2
million students or new workers in the short-run.
In addition, a host of social benefits can be attributed to
improved college education and employee training. Higher
earnings can move families out of poverty and reduce reliance Questions and Concerns
on social safety net programs. Greater economic success among
a local economy’s residents also stimulates economic activity Would colleges and universities have the administrative
and generates tax revenue. And, if the incentives are successful capacity and data to measure the subsequent labor market
at inducing relatively greater labor market rewards among performance of their students?
disadvantaged or minority populations, the states might value
This proposal would create very serious data needs for
this outcome on equity grounds even though the rewards accrue
colleges in each state. But many states are now developing
mostly to private individuals. Finally, if public colleges tailor
administrative data systems that link education and earnings
their curriculum to meet critical local labor market shortages—
records (Zinn and Van Kluenen 2014), so the data by which
such as those for nurses—consumers of the targeted industry
states could measure these earnings outcomes for graduates (as
will benefit as well.
well as nongraduates) of each college are potentially available.7
Better-tailoring public college curriculum potentially carries Technical assistance from federal and state governments
very little, or even zero, costs to states and higher education would help colleges follow their student earnings would be
institutions. At least in theory it is possible that the incentives crucial here. The federal government might also incent local
in state subsidies could be implemented with no net increase in states in a region to merge their data systems, so that students
costs to the colleges or the state by simply restructuring existing who move out of state can be tracked as well.8 In many cases,
subsidies. But if teaching in the high-demand fields is also the state and local workforce boards will have experience
costlier to the colleges, due to higher teacher or equipment costs, using the local earnings data, and can also help local colleges
then the average cost of instruction per credit hour to colleges develop an infrastructure for routinely measuring the earnings
will rise, which they might view as an unfunded mandate. of their graduates as well as their academic outcomes.
If so, how might states and their colleges respond to such higher Won’t colleges have strong incentives to game the system in
costs? First, they could keep total costs constant by cutting various ways, to improve their measured performance along
expenditures on other services (in noninstructional costs), the requisite dimensions?
though this might be costly to college outcomes in other ways
Poorly designed performance measures for public programs
(Webber and Ehrenberg 2010). Second, they could reduce their
instructional offerings and capacity in low-reward (in terms of can potentially generate unintended consequences.9 States
the labor market) academic fields. Average student completion do not want to encourage colleges to improve their outcomes
rates and labor market rewards to students may rise as a result through “cream-skimming” from their applicant pool, by
of these changes even if the colleges offer fewer total credit raising entrance requirements, nor do they want to improve
hours of instruction per term to students enrolled there completion rates by lowering the bar for graduation. Specific
and fewer such enrollments over time. Third, to avoid these rules prohibiting such practices plus careful monitoring
options, states may opt to modestly raise tuition costs, perhaps to enforce them would be necessary to ensure that such
partially offsetting the burden that higher costs may impose manipulation is not used to improve the outcomes that
on students by higher needs-based scholarship assistance. As generate rewards.
noted earlier, the federal government might also provide some If states train too specifically for occupations or industries
financial assistance to states making this transition to help in high demand, what happens to students if they ultimately
them offset the higher costs they would likely incur. leave those fields, or when labor demand shifts over time to
Subsidies for workplace training could also be limited to other sectors?
modest sums. For instance, Hollenbeck (2008) reported that There is always some tension between providing workers with
the sum of expenditures by all states providing incumbent the specific skills they need for getting jobs in the targeted
worker training was under $1 billion per year before the sector, and the more general skills they will need in the job
Great Recession, and this sum financed incumbent worker
94 Policies to Address Poverty in America
Harry J. Holzer
market, especially if/when they leave the jobs they obtain with
these skills and even cross into other sectors. To maintain
Conclusion
longer-term earnings improvements, especially in dynamic State and—and in some instances federal—policymakers
labor markets where high-demand sectors today can become should focus on improving the skills and earnings potential
low-demand sectors tomorrow, workers must have at least of poor youth and adults as an important multigenerational
some skills that are clearly general and portable. Certain antipoverty initiative. The earnings of less-educated workers
approaches, like stackable credentials in the career pathways have greatly lagged behind those of more-educated workers
framework, explicitly aim to achieve this result.10 The colleges in the past few decades. Although postsecondary education
must also be encouraged to be nimble, and to adjust their enrollment is up among all income levels, dropout rates are very
offerings over time to labor market changes. high among low-income students, especially at community
colleges and nonelite four-year colleges. Educational
Won’t the workers who are trained in high-demand fields institutions should be incentivized to better guide students
just displace other workers, who would otherwise have into the workforce and to concentrate the school curriculum
obtained the same jobs? on the skills valued or unmet in the local labor market.
Economists have worried for some time that their estimates of Specifically, I propose that states partially base public college
training impacts for individuals might overstate the aggregate funding on graduates’ reported earnings five years following
impact due to such displacement (Heckman, LaLonde, and graduation and, where appropriate, on the colleges’ provision
Smith 1999). But evidence from simulations performed of courses that are especially important to the local economy.
by Davidson and Woodbury (1990), in a paper estimating Rigorous research and evaluation of training programs
the size of displacements created by incentive bonuses for have demonstrated that sectoral programs, with associated
unemployment insurance recipients to accept employment career pathways, can have the largest positive impacts on the
earlier, find displacement effects that are relatively small. In subsequent earnings of disadvantaged workers. I propose
the short run, with wages fixed, displacement could mean to create incentives for more colleges to participate in these
that jobs are rationed away from other workers toward those programs, along with technical assistance to help them do
receiving a particular treatment. But in the longer run there so. States might also experiment with incentives to encourage
is less cause for concern, as employers will presumably create employers to participate in partnerships with community
more jobs in response to an outward shift in the supply of colleges or to directly hire and train more workers on their own.
workers with the requisite skills (though perhaps along with
some reduction in wages). Significant private and social benefits would accrue with
carrying out the provisions of this proposal. Most notably, the
nation would realize increased productivity, higher earnings,
and better opportunity to find gainful employment. The
higher earnings can move families out of poverty, reduce
reliance on social safety programs, and raise local economies.
On the revenue side, better-tailoring public college curriculum
potentially carries very little, or even zero, costs.
Finally, the best preparation for low-income students in the
long run will give them not only the specific skills they need
for jobs in the targeted sectors, but also some general skills
that are valued across firms and sectors. Developing curricula
and pathways that maintain this balance should be a high
priority as well.
The Hamilton Project • Brookings 95
96 Policies to Address Poverty in America
BUILDING SKILLS
Proposal 9: Providing Disadvantaged Workers
with Skills to Succeed in the Labor Market
Sheena McConnell
Mathematica Policy Research
Irma Perez-Johnson
Mathematica Policy Research
Jillian Berk
Mathematica Policy Research
Introduction for the WIA Adult program and other sources of vocational
training has been declining over the past several decades.
Millions of Americans cannot obtain jobs that pay enough WIA was scheduled for congressional reauthorization in 2003,
to lift them out of poverty. For many, the principal barrier but more than ten years have passed without new legislation.
to obtaining these good jobs is their lack of specialized In May 2014, policymakers announced that they reached a
occupational skills increasingly sought by employers. bipartisan deal to reauthorize WIA through new legislation,
Research has shown that vocational training can be effective the Workforce Innovation and Opportunity Act.
in boosting the earnings of disadvantaged adult workers. This
proposal argues that, by helping workers acquire the skills This paper outlines why Congress should increase funding
that employers demand, vocational training could be wielded for vocational training for disadvantaged adult workers.1
as an effective antipoverty tool. Specifically, we argue that Congress should increase funding
for the WIA Adult program. Decades of research on the
The 1998 Workforce Investment Act (WIA) Adult program effectiveness of vocational training of the type provided
is one of the most important sources of government-funded by the WIA Adult program, as well as an evaluation of the
vocational training for disadvantaged workers—workers WIA Adult program itself, suggest that the program can
with both low levels of education and low levels of skills. be effective in increasing the employment and earnings of
Accessed through the American Job Center network, this disadvantaged workers.
program provides vocational training funds for adults aged
eighteen or older who are determined to need, and be suitable We also argue, however, that Congress, and the state and
for, vocational training, with priority of service given to low- local workforce investment boards that administer the WIA
income workers. Eligible workers are provided a voucher, Adult program, should explore ways to improve the vocational
known as an individual training account, that they can training that is available to adult disadvantaged workers. In
use to purchase training at any program as long as it is on particular, policymakers should focus on addressing two
a state-approved list of programs that includes courses at concerns about training programs: (1) too many people who
both community colleges and private training providers. The start training programs do not complete them, and (2) too many
WIA Adult program, currently funded at about $800 million, people do not find a job in the occupation for which they are
serves more than one million workers annually. Funding trained. We recommend experimentation with four evidence-
The Hamilton Project • Brookings 97
BUILDING SKILLS
Proposal 9: Providing Disadvantaged Workers with Skills to Succeed in the Labor Market
based approaches to address these concerns: (1) providing (figure 9-1). Since 1985 the amount budgeted for key U.S.
more guidance to workers so they make appropriate decisions Department of Labor training programs has declined by
about training, (2) investing in more services to support the about 20 percent in real terms.2
workers while they are enrolled in a training program, (3)
developing training programs that provide the skills demanded Even among supporters of vocational training, there is
by employers, and (4) developing training programs that are legitimate concern that many people who start programs
more suited to the needs of disadvantaged adult trainees. In do not complete them. Within three years of enrollment in
the absence of federal action on reauthorization to fund this a community college, fewer than half of all enrollees have
experimentation, we encourage state and local workforce attained an associate’s degree or vocational certificate,
boards that oversee the American Job Centers to take advantage transferred to a four-year institution, or remain in college
of grant opportunities to test the proposed strategies aimed at (Horn and Weko 2009). Only about 55 percent of the people
improving outcomes for trainees. who begin two-year colleges obtain either an associate’s
degree or a certificate (Holzer and Dunlap 2013). Analysis of
data on training vouchers provided by the WIA Adult and
The Challenge Dislocated Worker programs found that only 64 percent of
workers who enrolled in training programs at community
Low-skilled workers are much more likely to be unemployed colleges completed a training program within three years
and living in poverty than are more-skilled workers. In (Perez-Johnson, Moore, and Santillano 2011). Although the
2013 the unemployment rate was 11.4 percent among rate of completion for those enrolled in training at a private
people twenty-five and older without a high school diploma, training provider was higher, about 15 percent of trainees still
compared with 5.4 percent among those with an associate’s did not complete a training program within three years.
degree (U.S. Bureau of Labor Statistics 2014). Similarly, in
2013 the median weekly earnings of people twenty-five and A second concern is that too many workers who complete
older with an associate’s degree was more than 60 percent training cannot subsequently find a job to use the acquired
higher than those without a high school diploma ($777 skills. A study of training vouchers provided through the WIA
compared with $472). Poverty rates are highest among Adult and Dislocated Worker programs reported that only
people who are unemployed, do not work full time, or have about 40 percent of the participants found employment in the
low wages (Meyer and Wallace 2009). occupation for which they received training (Perez-Johnson,
Moore, and Santillano 2011). Similarly, a study of the Trade
The supply of skilled workers is not keeping up with the Adjustment Assistance program found that only 37 percent of
demand for them (Goldin and Katz 2012). Employers report people who participated in training funded by that program
shortages of workers with occupation-specific skills (Holzer held a job in the occupation for which they were trained in
et al. 2011). A recent survey of 2,000 U.S. companies found the fourth year after they were initially laid off (Schochet et al.
that 30 percent had been unable to fill skilled job positions for 2012). These statistics suggest that there is often a missing link
more than six months (Manyika et al. 2012). between employers and training programs.
Many low-income workers would not be able to access
vocational training without assistance from government
programs. Although the vast majority of vocational training
A New Approach
in the United States is provided by employers (Mikelson We propose five evidence-based recommendations to
and Nightingale 2004), employers are less likely to provide improve publicly funded vocational training. The first
training for their lower-skilled positions, which tend to have recommendation requires congressional support for
higher rates of turnover (Lane 2000). Hypothetically, workers additional funding for the WIA Adult program. While
could pay for their own training, but many unemployed and the other four recommendations could be congressionally
low-skill workers do not have the financial resources or the mandated when WIA is reauthorized, they could also be
ability to borrow to pay for training. implemented by the state or local workforce investment
boards that administer the WIA Adult program even without
The United States does not currently invest heavily in reauthorization. Funding for these recommendations can
vocational training compared with other countries, and be obtained from federal grants. For example, in 2012 the
funding for vocational training has declined over the past U.S. Department of Labor issued $147 million in grants from
decades. Whereas the United States spends less than 0.05 the Workforce Innovation Fund to states or local workforce
percent of its gross domestic product on vocational training, investment boards to demonstrate and evaluate innovative,
other industrialized nations invest up to ten times as much
98 Policies to Address Poverty in America
Sheena McConnell, Irma Perez-Johnson, Jillian Berk
FIGURE 9-1.
Labor Market Training Expenditures as a Percent of GDP in OECD Countries, 2011
0.6
0.5
Percent of GDP spent on training
0.4
0.3
0.2
0.1
0
Fin
De lan
nm d
Au rkst
Fr ia ar
Po nce
rt
Ge uga a
rm l
Sp yan
N ai
Sw or n
itz wa
er y
Be and
lg liu m
Ne th Ita
er ly
Ne Ca ds
w na la
Ze da
al n
Es nd
to
Isr
Sl ae
ov le ani a
Sw nia
Lu K en
xe o
Un mb rea
ite ou
d rg ed
St
Au ate
st s
Hu ile
ng ra
Ch lia
Cz
Sl ec J ry
h ap
Re an
pu
Po lic
laba
ov M nd
ak ex
Re ico
pu bl ic
Source: OECD 2013.
Note: Data were not available for Greece, Ireland, and the United Kingdom. Training expenditures for Mexico and the Slovak Republic are less than 0.005 percent of GDP. The OECD defines
labor market training as “measures undertaken for reasons of labor market policy, including both course costs and subsistence allowances to trainees, when such are paid. Subsidies to
employers for enterprise training are also included, but not employer’s own expenses” (OECD 2008).
evidence-based approaches to improve the workforce in the training, and the gains are sustained over time (U.S.
system. Another $60 million for these grants is proposed Government Accountability Office [GAO] 1996). One review
in the president’s fiscal year 2015 budget. The Long-Term of the evidence suggests that low-skilled workers can increase
Unemployed Ready to Work Partnerships to be awarded this their earnings by between about $300 and $900 per quarter
summer, or the Trade Adjustment Assistance Community (Heinrich 2013). These gains are large and sustained enough
College and Career Training grants—both funded by the that they are likely to cover the cost of the programs.
U.S. Department of Labor—could also be used.
Even with the evidence of the effectiveness of training for
RECOMMENDATION #1: CONGRESS SHOULD disadvantaged workers, the budget for the WIA Adult program
INCREASE FUNDING FOR THE WIA ADULT PROGRAM has declined markedly over the past decades. Between fiscal
years 1998 and 2013, the budget for the WIA Adult program
Multiple rigorous evaluations conducted over the past
declined by 41 percent in real terms (figure 9-2). Anecdotally,
decades in Europe and the United States suggest that access to
many local workforce investment administrators report not
vocational training increases the employment and earnings of
providing training for eligible workers because their training
low-skilled adults (Bloom et al. 1993; Card, Kluve, and Weber
funds run out.
2010; Heinrich et al. 2013; Hollenbeck 2009). Low-skilled
adults who receive training through these programs typically We recommend that Congress reverse this decline in
enroll in relatively short-term, inexpensive training programs. funding for vocational training and, more specifically,
A typical program funded by the WIA Adult program lasts that it funnel the increased funding to the WIA Adult
less than a year and costs between $3,000 and $6,000. While program. We recommend expansion of the WIA Adult
in training, participants earn less than they would if they program rather than other sources of training funding for
were not in training; after they complete training, however, three main reasons. First, that program has been shown, at
they earn more than they would if they had not participated least by a nonexperimental study, to increase the earnings
The Hamilton Project • Brookings 99
BUILDING SKILLS
Proposal 9: Providing Disadvantaged Workers with Skills to Succeed in the Labor Market
FIGURE 9-2.
Total Funding for the WIA Adult Program, Fiscal Years 1998–2013
1,500
1,200
Millions of 2013 dollars
900
600
300
0
FY 1998 FY 2001 FY 2004 FY 2007 FY 2010 FY 2013
WIA Adult Program WIA Adult Program, with ARRA funds included
Source: U.S. Department of Labor 2014.
of its participants (Heinrich et al. 2013). 3 In contrast, RECOMMENDATION #2: THE WORKFORCE
studies of training for other populations have been less BOARDS SHOULD EXPERIMENT WITH PROVIDING
encouraging. For instance, a study of the WIA Dislocated STRUCTURED, DIRECTIVE GUIDANCE TO WORKERS
Worker program, a program that is structured identically WHO REQUEST TRAINING
to the WIA Adult program but serves dislocated workers When contemplating training, workers need to make complex
who have been laid off and are typically more skilled and decisions. They need to decide whether to undertake training,
experienced than the WIA Adult program participants, was and, if so, what courses to take, and through which training
found to be ineffective at increasing earnings (Heinrich et provider. They may need to find child care or support for
al. 2013). Other studies of training programs for dislocated themselves and their families while they are in training.
workers have found either no evidence of positive impacts Workers may not have the information or analytical ability
on earnings (Schochet et al. 2012) or impacts that are to make good decisions, which could result in incomplete
smaller than that for disadvantaged workers (Hollenbeck training or in the acquisition of skills that are not in demand
2009). Second, because the WIA Adult program is offered by employers.
through American Job Centers, workers can access other
employment services and supports such as labor market A study of different approaches to providing training
information, job listings, and other services at the same time vouchers to trainees in the WIA Adult and Dislocated Worker
that they are being trained. Third, funding an established programs found that WIA Adult program participants benefit
program rather than setting up a new program will avoid from counseling (Perez-Johnson, Moore, and Santillano 2011).
concerns voiced by the GAO and others about fragmenting Those participants in the WIA Adult program who expressed
employment and training services (GAO 2011). interest in training and were required to discuss their training
decision with an employment counselor earned on average
$474 (about 8 percent) more per quarter six to eight years later
as compared to program participants who were not required
to discuss their training decision, but who were offered the
same amount of training funds. The study also suggests that
100 Policies to Address Poverty in America
Sheena McConnell, Irma Perez-Johnson, Jillian Berk
BOX 9-1.
Example of a Tool to Assist in Occupation Selection: My Next Move
Accessed online at http://www.mynextmove.org/, this assessment tool enables job seekers to explore the requirements of and
their suitability for different occupations. The interest assessment, accessed by clicking on “Tell us what you would like to
do,” requires the job seeker to rate sixty work activities based on her interest in performing the task. The tool then categorizes
the interests into six career types: realistic, investigative, artistic, social, enterprising, and conventional. The job seeker is
then asked to indicate her job zone, or the level of experience and education she either has or is willing to pursue. The tool
then compiles a list of potential occupations for the job seeker to explore given her interests and the results of the Job Zones
activity. For each occupation, the tool provides its education and training requirements, and the typical personality traits,
skills, and abilities of people in the occupation. The tool also notes if the occupations are high-demand and high-growth,
green, or part of a registered apprenticeship program. (This box is based on Laird and Holcomb 2011.)
such counseling should be mandated. When meeting with used in a wide range of programs) for a U.S. Department of
a counselor was not required to receive the voucher, only 4 Labor–sponsored study; that study is publicly available (Perez-
percent of workers chose to do so. Johnson, Moore, and Santillano 2011). Box 9-1 provides
an example of an assessment tool that workers could use to
We recommend that WIA Adult program participants be explore occupations. Structured tools could help guide workers
provided structured and directive counseling. By structured, through the processes of program research, comparing
we mean that counselors consistently cover the same set of program and provider options, estimating a training budget,
topics with program participants. By directive, we mean that and projecting income and expenses while participating in
counselors guide program participants to a training option training. One tool could be similar to the training report card
and have the authority to refuse funding for training decisions proposed in a prior Hamilton Project brief (Jacobson and
that they view as unwise. Currently, while most WIA training LaLonde 2013). In addition to the factors in this report card,
programs require workers to discuss their training choices counselors should also help workers consider the amount they
with an American Job Center employment counselor before expect to earn once they complete training, what they could
their funding is approved, typically this counseling is neither earn if they took a job instead of attending training, and the
structured nor directive (D’Amico et al. 2004). number of additional years they expect to work. This would
We propose that employment counselors consistently discuss help workers examine their expected returns to training.
with program participants the factors that influence the To implement this recommendation, even without federal
benefits and costs of training and the likelihood that the action, local workforce investment boards will need to invest
worker will complete the training. Counselors should also in more counseling staff and in additional training and
be empowered to not fund training that they deem unlikely oversight of the staff, as well as in collecting and refining the
to lead to success in the labor market. During their meetings tools. The study of individual training accounts found that,
with workers who request training, counselors will need to on average, counselors spent about seventy-five minutes with
consistently conduct assessments to collect information about each program participant on her training decision when
workers’ interests, basic skills, aptitudes, and transferrable counseling was required but unstructured (Perez-Johnson,
skills. They should discuss barriers to employment, the type Moore, and Santillano 2011). We expect that more-structured
of training they seek, the providers they are considering, the counseling would require an additional thirty minutes per
number of additional years they expect to work, their training trainee. To minimize staff burden, some of the proposed
costs and budget, and their need for and potential sources of activities could be delivered within group workshops.
income support while participating in training. Information
on possible earnings trajectories after participating in training To guide workers, counselors need accurate and timely
should be discussed as well as the likelihood of obtaining a job information to understand the skills demanded by employers
with the training. and the potential returns to different training paths. Two
new data sources—Real Time Labor Market Information and
To facilitate this counseling, the programs should provide linked administrative data—offer promising opportunities
tools to help counselors and workers examine the anticipated to enhance counselors’ understanding of local labor markets
benefits and costs of training. A complete suite of worksheets and increase their confidence in offering workers directive
and counseling tools was developed (drawing from exemplars
The Hamilton Project • Brookings 101
BUILDING SKILLS
Proposal 9: Providing Disadvantaged Workers with Skills to Succeed in the Labor Market
counseling. Real Time Labor Market Information uses This approach is promising, and we recommend that it be
information in online job postings to make inferences about rigorously evaluated.
labor market conditions. Providers of Real Time Labor
Market Information use a daily Web crawler to scrape job RECOMMENDATION #4: WORKFORCE BOARDS
postings from the Internet and aggregate this information to SHOULD EXPLORE DEVELOPING TRAINING
capture trends in employer demand, emerging occupations, PROGRAMS IN PARTNERSHIPS WITH EMPLOYERS
and skill requirements (Vollman 2011). Real-time data One of the most promising new vocational training programs
provide a snapshot of the market and can reveal the extent of for low-skill adults strengthens this link between training and
demand for a particular credential or the emergence of a new employers’ needs (Maguire et al. 2010; Richburg-Hayes 2008;
occupation. Counselors can use real-time data to understand Woolsey and Groves 2010). Sector-based programs focus on
their local labor markets and guide workers accordingly. a particular industry (such as health care, manufacturing, or
Linked administrative data increasingly allows states and local information technology) and engage with employers in that
areas to track the outcomes of workers who enroll in different sector. Using both labor market statistics and information
training programs (Jacobson and LaLonde 2013). Federal grant collected directly from employers, the programs identify the
funding awarded under the State Longitudinal Data System skills that employers need. Training providers and employers
grants and the Workforce Data Quality Initiative grants have work collaboratively to develop training curricula tailored to
allowed states to make infrastructure investments to improve specific job opportunities; training providers carefully screen
linkages between the workforce system, community colleges, applicants to ensure that matches with the targeted occupation
and administrative earnings records. States need to take the are appropriate. When trainees complete the program, they
next step to analyze these data and provide counselors and receive a credential that employers recognize. In addition,
workers with information on the distribution of educational the programs develop strong relationships with employers to
and employment outcomes for workers who enrolled in help quickly match workers who complete their training with
similar training programs. available job vacancies.
RECOMMENDATION #3: WORKFORCE BOARDS Evaluations of sector-based programs have yielded promising
SHOULD EXPERIMENT WITH PROVIDING MORE results. A study of three relatively mature, sector-based
SUPPORTIVE SERVICES programs estimated that participants earned about $4,500
An important barrier to low-income workers completing (18 percent) more over the two years after they had enrolled
training is lack of financial assistance to cover child care, in the study than similar workers who did not participate in
transportation, and basic needs (Goldrick-Rab and Sorenson the program (Maguire et al. 2010). Importantly, significant
2010). Although the WIA Adult program and other programs earnings gains were estimated for program participants
at the American Job Centers can provide funds for supportive with diverse characteristics—including men, women,
services, many trainees do not receive this help. We estimate African Americans, Latinos, immigrants, people who were
that, of those who obtained training in the WIA Adult and formerly incarcerated, welfare recipients, and young adults.
Dislocated Worker programs, fewer than 40 percent received This suggests that sectoral programs could be an appealing
any support to pay for child care, transportation, tools, or training option for a wide range of low-skilled workers and
uniforms (Perez-Johnson, Moore, and Santillano 2011). could be accessed by WIA Adult program participants using
the individual training account vouchers in the same way that
The WIA Adult program could increase the value of its they access other training programs. Box 9-2 describes one of
training vouchers, or individual training accounts, and allow the sector-based programs found to be successful.
the program participant to use the voucher to also cover
supportive services to ease their participation in training. For Sector-based training programs require significant up-front
example, in The Thumb Area Michigan Works! program, staff investment to develop and refine. Individual training providers
members determined all the programs for which a worker and employers may be unlikely to make the investment,
was eligible and consolidated the individual’s funding into especially with uncertainty about whether public funding
one Tool Chest voucher (U.S. Department of Labor 2002). would be available to pay for the training. The state and local
The individual could use this voucher for education, training, workforce boards should be willing to invest in developing
or any other services that were consistent with the funding the necessary partnerships between employers and training
sources and approved by a staff member at the American Job providers and to assist in designing the programs. They could
Center. The consolidation relieved the worker from applying involve intermediaries to develop the partnerships. One
to multiple programs and allowed use of the voucher for sector-based program found to be effective was developed
a wider range of purposes and at a wider set of vendors. by the Wisconsin Regional Training Partnership, a nonprofit
102 Policies to Address Poverty in America
Sheena McConnell, Irma Perez-Johnson, Jillian Berk
BOX 9-2.
Example of a Successful Sector-Based Program: Medical Office Occupations at
Jewish Vocational Services–Boston
Jewish Vocational Services (JVS)-Boston is a community-based nonprofit organization that provides vocational training to
disadvantaged youth and adults. Having previously received grants to create incumbent training for health-care providers
and administered an American Job Center in Boston, it has a long history of working with employers. It employed a full-time
employer-relations staff member to identify employers’ needs and assist with placement of trainees. Employers served on
committees to advise on the content of the programs and the eligibility requirements. JVS’s medical office training program
was included in the Maguire and colleagues (2010) study. To be eligible for the program, workers needed to have a high school
diploma or GED, possess the ability to read at the sixth-grade level or higher, and show during an interview that they have the
interest and ability to succeed in the training. The training program lasted twenty to twenty-two weeks and took twenty to
twenty-five hours per week. In addition to vocational skills, the program provided job readiness training, case management,
postemployment services, and a four- to six-week internship. The program was found to increase trainees’ earnings by 21
percent over the two years after enrollment. (This box is based on Maguire et al. 2010.)
organization (Maguire et al. 2010). The U.S. Department before they begin a vocational training program. An alternative
of Labor has announced the availability of $150 million in approach that has been found to be promising is to integrate
grants under its Long-term Unemployed H-1B Ready to the teaching of basic and vocational skills into the same
Work Partnerships grant program to fund the development course. This provides a context for learning the basic skills
of partnerships between employers, nonprofit organizations, and reduces the length of time taken to acquire the vocational
and workforce investment boards to develop innovative skills. Washington state has implemented an Integrated Basic
sectoral training programs for the long-term unemployed. Education and Skills Training program for some occupations
Grants that could be released under the U.S. Department of throughout its community and technical colleges. A study of
Labor’s Workforce Innovation Fund could also be used for that program found that it increased the probability the trainee
this purpose. earned a certificate or degree and improved other educational
outcomes, but did not increase earnings (Zeidenberg, Cho,
RECOMMENDATION #5: WORKFORCE BOARDS and Jenkins 2010). The findings were positive enough for this
SHOULD EXPLORE PARTNERING WITH TRAINING program model to have been replicated in other community
PROVIDERS TO DEVELOP TRAINING PROGRAMS colleges, and it merits further study.
MORE SUITED TO THE NEEDS OF ADULT TRAINEES
Some of the factors that make participating in training difficult As much as possible, training should be divided into multiple
for adult disadvantaged workers may be ameliorated by three discrete courses that build on each other. For example, a
types of changes in the structure of training programs: (1) two-year course that serves as a means to an occupational
providing a flexible schedule for course offerings, (2) providing credential is better provided as a series of four separate
basic skills training at the same time as vocational skills, and sequential courses, each one providing an interim credential
(3) providing training in more discreet, stackable modules. and building on the skills taught in the prior course. This
approach avoids trainees participating in programs that teach
Providing courses more frequently and in the evenings as skills that they already possess or do not need. It also provides
well as during the day would make it easier for workers to more flexibility in when the courses are taken and provides
work or care for dependents while in training. Waiting for some interim credentials to workers who may not be able to
the beginning of a semester at a community college can complete the full sequence of courses. Many of the career
significantly increase the length of time before training pathways programs identify sequences of courses to generate
can begin and hence the cost of participating in training in credentials that will lead to sufficient skills for an occupation.
terms of forgone earnings. Online training courses can also For example, a program could provide a series of instruction
accommodate the need for more flexibility. modules that prepare students for certification in progressively
higher-paying health-care occupations—certified nursing
Lack of math and reading skills is often a barrier to accessing assistant, patient care technician, and licensed practical nurse
and completing training programs. Typically, the WIA Adult (Fein 2012).
program requires workers to take basic education courses
The Hamilton Project • Brookings 103
BUILDING SKILLS
Proposal 9: Providing Disadvantaged Workers with Skills to Succeed in the Labor Market
While these approaches seem promising, we do not yet have experiences, skills, and other characteristics of the workers.
rigorous evidence of their effectiveness. The Departments of Workers with many barriers to employment are at high risk of
Labor and Education have allocated $2 billion in grant funds to not completing the training and of not being able to find and
community colleges through the Trade Adjustment Assistance retain a job after training. Resources for those workers may be
Community College and Career Training grant program better spent on addressing their employment barriers directly
to facilitate open or rolling enrollment and to structure and providing job readiness training and assistance with job
training programs to facilitate training while working. Many search, retention, and advancement. For other workers with
community colleges are conducting evaluations of their more skills, the increased earnings from participating in
reform efforts; hopefully, these evaluations will provide strong training may not offset the cost of the earnings forgone while
evidence on approaches that could be adopted more broadly. participating in training. In this case, training would not be
cost-effective even for the trainees. Our recommendation is
COSTS AND BENEFITS that employment counselors in the WIA Adult program assess
The main benefit of our proposal to increase public the suitability of training and provide training only to workers
investments in vocational training is an expansion in the for whom the expected benefits exceed the costs.
number of low-income individuals who participate in training
The other recommendations in this memo—providing
and experience earnings gains once they finish training. The
guidance on the type of training, providing more supportive
size of the benefits from increased earnings depends on the
services, partnering with employers, and developing training
persistence of the earnings increase. While research evidence
more-suited to the needs of adult workers—still need to
is not conclusive on how long the increased earnings from
be evaluated. These evaluations should examine not only
training persist, a study of multiple programs in the United
whether the interventions are effective in increasing retention
States and Europe found that the impacts of vocational
in programs and the earnings of trainees, but also whether the
training on earnings over two to three years are on average
total benefits—in terms of increased earnings, reduced use of
larger than the impacts over one year (Card, Kluve, and
public assistance, and increased taxes—exceed the total costs
Weber 2010), suggesting the benefits from training last for
of these programs.
several years at least. In addition to the benefit of increased
earnings for the trainees, the government also benefits
from the increased tax payments and reduced use of public
assistance (such as Temporary Assistance for Needy Families Questions and Concerns
[TANF], Supplemental Nutrition Assistance Program [SNAP, Are you suggesting that funding for disadvantaged workers
formerly Food Stamps], and unemployment insurance [UI]) be increased at the expense of dislocated workers?
that accompany trainees’ increased earnings. The main cost
of vocational training is the amount the government pays for While studies of the effectiveness of training for low-skilled,
the training program. The opportunity cost of the time spent inexperienced workers consistently show that it is effective,
in training—that is, if the trainees were not in the program, studies of the effectiveness of training programs for dislocated
they may be working and earning money—should also be workers are less encouraging. Some dislocated workers
considered as a cost, however. can obtain earnings gains from participating in training
that are large enough to offset the cost of that training, but
On average, training is likely to be a sound investment for low- the evidence suggests that, on average, this is not the case.
income disadvantaged workers. As discussed above, evidence A recent evaluation of the Trade Adjustment Assistance
suggests that earnings may increase by between $300 and $900 program also finds that even when dislocated workers are
per quarter from participation in the WIA Adult program. offered longer-term training programs, on average, workers
Assuming earnings increased by $600 per quarter (the middle would have been better off finding a job rather than investing
of the range suggested by research), that the forgone earnings in training (Schochet et al. 2012). Synthesizing the evidence
are small (which is likely for low-skilled adults), and that the from several multistate, matched comparison-group studies,
impacts on earnings persist for about three years, the benefits Hollenbeck (2009) concludes that the return to WIA-funded
from training programs that cost less than $5,000 (which training is lower for dislocated workers than it is for other
many do) would likely exceed their cost. training recipients. A study of older dislocated workers in the
Washington state found that attending community college
While training is cost-effective for the average disadvantaged
increased earnings, but that the return was lower for these
worker, it may not be for all disadvantaged workers. This is
workers than was the return for younger workers (Jacobson
because the expected benefits of training compared to its
LaLonde, and Sullivan 2005). Given limited training funds, it
costs—the return on investment—can vary depending on the
104 Policies to Address Poverty in America
Sheena McConnell, Irma Perez-Johnson, Jillian Berk
is better that they be targeted to those workers for whom the Notably, the residential component of youth programs like
return is greatest. Job Corps is likely to be important for two reasons. First, it
removes youth from the environment in which they were not
Policymakers, however, should not ignore dislocated workers succeeding. When asked about the benefits of moving away
who, even though they typically have more resources from home to a center, Job Corps participants talked about the
than disadvantaged workers, are still at risk of sliding into negative influences of their peers in their home neighborhoods
poverty because of their longer unemployment spells and and their relief from family obligations (Johnson et al. 1999).
inability to secure jobs that pay as much as they had earned Second, a residential program provides more time to address
before. Training may still be appropriate for some dislocated youth’s challenges—Job Corps provides structure and
workers, especially for younger dislocated workers who have supervision for most of the youth’s day. Moreover, maintaining
a longer time to reap the benefits from training. Programs, regular attendance is easier in residential programs—there
however, should be more selective about which dislocated is no commute and there are fewer distractions. While the
workers are encouraged to pursue training. Findings from residential component of a program may be important, not all
previous evaluations also suggest that there are other program youth can or want to move away from home. Findings from an
refinements—including providing career assessments and ongoing study of YouthBuild, a nonresidential program with
minimizing time to enter training—that merit testing (Berk many of the elements of Job Corps, will provide more evidence
2012). For those dislocated workers who are unlikely to benefit on this issue.
from training, alternatives to training should be developed
and rigorously evaluated.
What do you recommend for vocational training programs Conclusion
for youth? This paper has suggested policy changes to increase and
improve publicly funded vocational training. Yet many
The evidence on training for disadvantaged youth suggests
questions remain about effective vocational training strategies.
that to be effective, the programs need to be intensive. The
How can we increase the likelihood that a trainee completes
most disadvantaged youth face myriad challenges other than
the training program? How can we ensure that trainees find
lack of occupational skills, such as low literacy, the need
jobs in the occupations to which they have been trained?
to learn English as a second language, involvement in the
How can we identify who will benefit from training and who
criminal justice system, or substance abuse; some also face
will not? To address these questions, we need to embark on
the challenges of pregnancy or parenting. Successful training
a policy agenda that involves an ongoing cycle of developing
programs for youth need to address these challenges. We have
new programs that are informed by the lessons already
robust evidence that Job Corps, the largest federally funded
learned, evaluating these new programs, changing them in
program for youth, is effective (Schochet, Burghardt, and
response to the findings, and then testing again. Only then
McConnell 2008). In contrast to participants in Job Corps,
will we be able to identify a full suite of training programs that
youth who participated in the Job Training Partnership Act of
can significantly reduce the number of vulnerable American
1982 programs, which were found to be ineffective, typically
workers who, because they lack the necessary skills, fall into
attended the program part time and for only three to four
long-term poverty.
months (Kemple, Doolittle, and Wallace 1993). Effective
programs are likely to be expensive—it costs an average of
$16,500 for a youth to attend Job Corps.
The Hamilton Project • Brookings 105
106 Policies to Address Poverty in America
Section 4.
Improving Safety Net and Work Support
The Hamilton Project • Brookings 107
108 Policies to Address Poverty in America
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 10: Supporting Low-Income Workers
through Refundable Child-Care Credits
James P. Ziliak
University of Kentucky
Introduction increasing their participation in the labor force (Baum 2002;
Berger and Black 1992; Kimmel 1995; Tekin 2007).
Economic self-sufficiency through labor market work for low-
income families, especially those headed by a single mother, This policy memo introduces a way to restructure an existing
formed a fundamental tenet of both the 1993 expansion in federal child-care tax credit to better incentivize work and
the Earned Income Tax Credit (EITC) and the 1996 Personal improve the financial and child well-being for low-income
Responsibility and Work Opportunity Reconciliation Act families. Specifically, I propose converting the Child and
(aka welfare reform). While both reforms have been credited Dependent Care Credit (CDCC) from a nonrefundable
with expanding employment of single mothers in the years credit—a credit that cannot exceed the income taxes owed by
immediately following implementation (Grogger 2003; Meyer a family—to a refundable credit—one that can result in a net
and Rosenbaum 2001), employment rates of mothers with gain after taxes—that is targeted to low- and middle-income
dependent children have been on a steady decline over the families. Because current law does not limit eligibility for the
past decade, leaving many families unable to make ends meet CDCC based on income, the majority of tax expenditures are
(Blank and Kovak 2008; Bollinger, Gonzalez, and Ziliak 2009; spent on those families with annual incomes between $100,000
Fox et al. 2013). and $200,000 (Maag 2013). I propose capping eligibility at
$70,000 and making the credit a progressive function of
A key financial challenge facing these families is finding income, the age of the child (ages zero to four versus five to
affordable child care. In 2012 the average annual cost for twelve), and utilization of certified, licensed care facilities.
full-day, center-based care of an infant ranged from $4,850 These reforms, to be implemented at the federal level, will make
in Mississippi to $16,450 in Massachusetts; for care of a four- labor market work more attractive to low-income families by
year-old, the cross-state range was $4,300 to $12,350 (Child providing much-needed financial relief from the high cost of
Care Aware of America 2013). As a fraction of average annual child care. In addition, by reducing the out-of-pocket cost of
earnings among single mothers with children under the age of care for low-income workers, the reformed credit will enable
five, child-care costs amount to over one-fourth of earnings in more families to place children in formal instead of informal
Mississippi and over one-third of earnings in Massachusetts.1 care settings.
Evidence suggests that children do better in model, center-
based care than in informal, home-based care on a host of
cognitive and noncognitive measures (Bernal and Keane The Challenge
2011; Blau and Currie 2006; Morris et al. 2009), and that
The fact that mothers are deterred from working in the labor
women respond to reductions in effective child-care prices by
market because of costly child-care options runs counter to
The Hamilton Project • Brookings 109
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 10: Supporting Low-Income Workers through Refundable Child-Care Credits
the national goal of economic self-sufficiency. To fix ideas on However, in a remarkable turn, the level of employment of
the evolution of employment in recent decades, figure 10-1 poor and near-poor married women was lower in 2012 than
depicts employment rates of single and married mothers (by it was in 1980.
the education level of single mothers) with dependent children
under the age of thirteen. This age range of children is selected Figure 10-2 presents a more disaggregated look at the
because the presence of children under the age of thirteen is employment status of single mothers with children under the
a requirement for two of the three major federal child-care age of thirteen by examining part-time and full-time work.
assistance programs.2 The data are drawn from the 1981–2013 The figure shows that, starting in the late 1980s, a plurality of
Annual Social and Economic Supplement of the Current single mothers were employed year-round and full-time, rising
Population Survey, and employment refers to any reported paid to 52 percent in 2000, but then falling steadily to 42 percent by
work in the prior year. The huge surge in employment rates in 2012. In the past decade, the decline in full-time work (both
the 1990s—which occurred coincident with the expansion of full-year and part-year) has been mostly filled by an increase
the EITC, the strong economy, and welfare reform—is most in the share not in the labor force (NILF), and to a lesser extent
evident among single mothers with a high school education or by an increase in full-year, part-time work. Since 2000 there
less, and among those never married (of any education level). has also been an increase in the fraction of married mothers
What is also striking in figure 10-1 is the secular decline in with children under the age of thirteen not in the labor force;
employment after 1999 and the relative absence of a cyclical this mostly coincides with a decline in the fraction of married
effect even in the face of the Great Recession of 2007–2009. mothers working part-year, including those working both
Employment rates in 2012 are about 10 percentage points full-time and part-time.
lower than in 1999 for each group of single mothers, and about
The past decade has witnessed a significant shift away from
7 percentage points lower for married mothers. In results not
employment among mothers, whether single or married, that
depicted, employment trends of mothers living in families
was particularly pronounced among the less skilled and those
with incomes below twice the poverty level are quite similar.
with family incomes below twice the poverty level. While a
FIGURE 10-1.
Employment Rate of Women with Children under Age 13, by Marital Status and
Education
100
90
80
Percent of single mothers
70
60
50
40
30
20
10
0
1980 1984 1988 1992 1996 2000 2004 2008 2012
Single, all education levels Single, less than high school education
Single, high school education Single, more than high school education
Single, never married Married
Sources: U.S. Census Bureau various years; author’s calculations.
Note: Data are derived from the 1981–2013 Current Population Survey Annual Social and Economic Supplement.
110 Policies to Address Poverty in America
James P. Ziliak
FIGURE 10-2.
Employment Status of Single Mothers with Children under Age 13
60
50
Percent of single mothers
40
30
20
10
0
1980 1984 1988 1992 1996 2000 2004 2008 2012
Full-time, full-year work Full-time, part-year work Not in labor force
Part-time, full-year work Part-time, part-year work (NILF)
Sources: U.S. Census Bureau various years; author’s calculations.
Note: Data are derived from the 1981–2013 Current Population Survey Annual Social and Economic Supplement.
full analysis of the reasons behind the decline in employment 20 percent. Note that these estimates are for any out-of-
is beyond the scope of this paper, the high cost of child care, pocket child-care expenses and that if we were to limit the
combined with stagnant real wages and other factors, might sample to only center-based child care, these ratios would be
be a contributing factor. Table 10-1 presents median out-of- significantly higher.
pocket child-care costs, the interquartile range of costs (75th
percentile less 25th percentile), median family earnings, and
median family income for working mothers pooled across A New Approach
the 2012 and 2013 waves of the Current Population Survey.3
A restructured Child and Dependent Care Credit (CDCC)
The table shows that even though the median out-of-pocket
could encourage greater economic self-sufficiency and
costs for child care among married mothers is about 80
improve the economic well-being of low-income families.
percent higher than for single mothers, family earnings
The federal government currently provides direct assistance
(mother plus spouse) of working married women are four
for child-care expenses through a nonrefundable tax credit
times higher; as a fraction of earnings, the burden on single
(CDCC), block grants to states (Child Care and Development
mothers is substantially higher at roughly 16 percent and 11
Block Grant [CCDBG], and Temporary Assistance to Needy
percent for those with children under age five and under age
Families [TANF]), and flexible spending accounts. Indirect
thirteen, respectively.
support for child-related expenses is provided through the
Figure 10-3 highlights the cross-state variation in the ratio nonrefundable Child Tax Credit (CTC), and the partially
of median out-of-pocket child-care costs to median earnings refundable Additional Child Tax Credit (ACTC).4 After briefly
of single mothers with children under age five. The figure summarizing current programs, I offer a new approach for
makes clear that the burden of child care is quite high in some funding child care that could boost employment and subsidize
states. At the median, child-care costs range from 6 percent families to secure quality center-based care.
of earnings in Alaska to 28 percent in Delaware, with twelve
states clocking ratios of child care to earnings in excess of
The Hamilton Project • Brookings 111
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 10: Supporting Low-Income Workers through Refundable Child-Care Credits
TABLE 10-1.
Out-of-Pocket Child-Care Costs, Earnings, and Income of Families with Working Mothers
Single, child under Single, child under Married, child under Married, child under
age 5 age 13 age 5 age 13
Median out-of-pocket child-care costs $3,000 $2,600 $5,400 $4,680
Median family earnings $19,200 $23,088 $82,500 $83,880
Median family income $22,000 $26,445 $85,276 $87,000
Median out-of-pocket child-care costs 15.6% 11.3% 6.5% 5.6%
as percent of median family earnings
Interquartile range of out-of-pocket child- $4,400 $3,800 $7,300 $6,500
care costs
Sources: U.S. Census Bureau various years; author’s calculations.
Note: The interquartile range is the difference between the 75th percentile of out-of-pocket child-care costs and the 25th percentile of out-of-pocket child-care costs.
Data are derived from the Current Population Survey and are pooled across the 2012 and 2013 waves.
CURRENT PROGRAMS mandatory child-care funding in Section 418 of the Social
Child and Dependent Care Credit (CDCC) Security Act. The CCDF, formed by welfare reform, allocates
funds to states to help low- and moderate-income families
The CDCC, established in 1976, is the oldest of the U.S. tax pay for child care, and also establishes state law and licensing
code credits related to child care. This nonrefundable credit for child care. In fiscal year (FY) 2012, about $5.2 billion was
covers qualifying child-care expenses of working parents allocated to CCDF: $2.3 billion in discretionary CCDBG
with children under the age of thirteen. The parent(s) must funds and $2.9 billion in mandatory Section 418 funds
have earned income and/or net positive self-employment (Congressional Research Service 2012).5 In addition, states
income. For married couples filing jointly, one spouse may be may transfer up to 30 percent of their TANF grant to CCDF,
considered having earned income if he or she is a full-time and may also directly spend TANF funds on child care. In
student or disabled; the family may not claim child-care FY2012, states transferred about $1.36 billion to CCDF from
expenses in excess of the lower of the two spouses’ earnings. TANF, and spent about $1.23 billion directly out of TANF on
The credit is worth 35 percent of qualifying expenses (capped child care. Moreover, states spent about $2.4 billion of their
at $3,000 for one child and $6,000 for two or more children) own funds on child care, financed out of Maintenance of
for families with adjusted gross income (AGI) under $15,000. Effort requirements for TANF, and/or Separate State Program
As such, the maximum credit is $1,050 for one child and funds, bringing total federal and state spending in FY2012 to
$2,100 for two or more children. The credit rate is lowered by 1 about $10.2 billion (U.S. Department of Health and Human
percentage point for each $2,000 of AGI above $15,000 until it Services [DHHS] 2012).
plateaus at a 20 percent rate for income above $43,000. There is
no income cap for eligibility, and because it is nonrefundable, To be eligible for CCDF assistance, children must be under
the credit affects only filers with a positive pre-credit tax age thirteen and living with parents who must be working, in
liability. Therefore, many EITC recipients do not qualify for school, or in protective services. Federal law limits eligibility
the current CDCC. The Urban–Brookings Tax Policy Center to those families with incomes less than 85 percent of state
estimated that in 2013 the largest average benefits of the CDCC median income. However, states have the option to impose
accrued to families with annual incomes between $100,000 lower limits, and in fact, in 2012 the median eligibility
and $200,000 (Maag 2013). rate was substantially lower at 54 percent of state median
income. Child-care assistance via CCDF and TANF is not
Child Care and Development Fund (CCDF) and Temporary an entitlement, and in FY2012 twenty-two states either had
Assistance to Needy Families (TANF) active waiting lists or had frozen intake (Schulman and Blank
The 1996 welfare reform law expanded and consolidated the 2013). Estimates show that in FY2009 only one in six children
discretionary child-care funding in the CCDBG of 1990 with
112 Policies to Address Poverty in America
James P. Ziliak
FIGURE 10-3.
Ratio of Median Out-of-Pocket Child-Care Expenses to Median Earnings of Single
Mothers, by State
New Hampshire
24% Maine
Washington
21%
18% Vermont
Montana North Dakota 25%
5% 19% Minnnesota
13%
Oregon
Wisconsin New York Massachusetts
12%
South Dakota 13% 13% 22%
Idaho Michigan
12%
17% Wyoming 7% Rhode Island
Pennsylvania 14%
15%
Iowa 12%
Nebraska 16% Ohio Connecticut
16% Indiana 9% 18%
Nevada Illinois 27% West
22% 10% New Jersey
Utah Virginia
17% Colorado Virginia 18%
20%
17% Kansas Missouri Kentucky 21%
California Delaware
16% 13% 13% 12%
North Carolina 28%
Tennessee 13%
Maryland
20% 14%
Oklahoma South Carolina
17% Arkansas 23%
Arizona New Mexico 16% D.C.
17 % 8% 15%
Mississippi Georgia
10% Alabama 20%
19%
Texas
16% Louisiana
17% Florida
18%
Alaska
6%
Hawaii
12%
9% and below 10% to 14% 15% to 19% 20% and above
Sources: U.S. Census Bureau various years; author’s calculations.
Note: Data are derived from the Current Population Survey and are pooled across the 2012 and 2013 waves.
eligible for CCDF or TANF child care received assistance Child Tax Credit (CTC)
(DHHS 2013).
The CTC was established to partially offset the costs of
Flexible Spending Accounts (FSAs) raising a child as part of the Taxpayer Relief Act of 1997; as
of FY2012, the CTC provided a credit worth up to $1,000 per
FSAs allow workers to set aside a share of pretax income qualifying child under the age of seventeen. In general, the
for designated purposes—including medical costs, CTC is not refundable, but if earnings exceed $3,000 or the
transportation, and dependent care. Dependent care FSAs family has three or more qualifying children, it is possible
allow workers to set aside up to $5,000 annually to pay for the filer to qualify for the ACTC, which is refundable.
qualified dependent care costs. Contributions to these If the value of the CTC exceeds federal tax liability, then
accounts are not subject to income or payroll taxes. Married a refund not to exceed 15 percent of earnings above the
taxpayers must both be working to take advantage of the $3,000 threshold can be received as the ACTC. The Urban–
deduction. Eligible expenses for child care are subject Brookings Tax Policy Center estimated that in 2013, 38
to several limitations, such as the following: Child-care million families claimed credits totaling nearly $60 billion,
expenses are limited to those for dependent children younger but only 13 percent of benefits went to the bottom income
than thirteen. Any given expense cannot be paid through quintile; about 77 percent of benefits accrued to the middle
FSA funds and be claimed for the CDCC. Unspent funds are quintiles, and 10 percent went to the top quintile (Maag and
forfeited at the end of the plan year. Carasso 2013). The CTC is phased out starting at earnings of
$110,000 for married couples filing jointly ($75,000 for head
The Hamilton Project • Brookings 113
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 10: Supporting Low-Income Workers through Refundable Child-Care Credits
of household), and is eliminated at earnings above $150,000. • Place income limit on credit
The lower threshold of $3,000 expires after the 2017 tax year,
• Vary by type of child-care provider
when it will return to its pre-2001 tax reform level (over
$13,000 in 2013); thus, it will no longer offer assistance to For children under the age of five with family AGI of less
families with very low incomes. than or equal to $25,000, I propose a 100 percent refundable
credit up to $4,000 in qualifying expenses for the first child
PROPOSAL: A REFUNDABLE CHILD AND DEPENDENT
in a licensed facility, with a maximum allowable expense of
CARE CREDIT
$6,000 for two or more children. The credit rate declines by
Although the federal government is actively involved in the 10 percentage points for each additional $5,000 in AGI, and is
provision of child-care assistance, two of the programs are poorly phased out for AGI above $70,000. For children between the
targeted—the CDCC and CTC—and the one that is targeted to ages of five and twelve, the rate is 70 percent for families with
low-income families (CCDF) leaves an overwhelming majority AGI below $25,000, declines by 7 percentage points for each
of those eligible for care uncovered. In a marked difference, the additional $5,000 in AGI above $25,000, and is zero for AGI
EITC is very target efficient and is proven to be a highly successful above $70,000. The base of qualifying expenses is the same.
prowork, antipoverty policy tool. A virtue of the EITC is that, as Like the current CDCC, the dollar amount of the credit applies
a cash refund to taxpayers, the taxpayer can spend the money to that portion of AGI received from earnings as defined in
flexibly to meet a host of needs. A case could also be made to Form 2441.
supplement the EITC with a targeted assistance program like a
child-care credit. Workers with dependent children use child In an effort to steer children to licensed, center-based child-
care in tandem with labor-market work, and thus a child-care care facilities, the credit rate is double that available to those
credit can improve the efficiency of the tax system by lowering families choosing unlicensed or informal care settings.
some of the disincentives to work from high marginal tax rates Making the credit twice as valuable for licensed care is justified
(Currie and Gahvari 2008). A survey of the literature on the because of the high expense of this type of care, as well as
employment effects of subsidized care suggests that a 10 percent the evidence pointing to the child-development benefits of
reduction in the price of child care will increase employment of center-based care (note, however, that not all licensed care is
single mothers by 3 to 4 percent and of married mothers by 5 to in a center). At the same time, allowing the refundable credit
6 percent (Ziliak, Hokayem, and Hardy 2008). for those utilizing unlicensed care facilities acknowledges
the fact that many low-income mothers work nonstandard
Another upside of a targeted child-care credit is that a directed shifts—nights and weekends—when formal care facilities
credit ensures that the money is spent on child care. Many are less readily available. The current Form 2441 used for the
low-income working families have insufficient resources to CDCC requires the filer to report the name, address, employer
invest in quality child care, and thus resort to lower-quality, identification (or Social Security number), and amount
but less-expensive, informal care, often relying on friends, paid for care. The refundable CDCC would also require this
family, and others. Research has shown that children in high- information; because licensing of centers is already a function
quality centers experience both short- and long-term benefits carried out by states, a registry of licensed facilities could be
compared to children in informal care settings, ranging from linked to IRS records with this form to verify claims for the
better test scores in the short run to reduced grade retention licensed- versus unlicensed-care credit amount.6
rates, higher graduation rates, higher earnings, and reduced
criminal activity in the long run (Bernal and Keane 2011; Blau Because child care is generally paid weekly or monthly, and
and Currie 2006; Heckman, Stixrud, and Urzua 2006; Morris since many low-income families are liquidity constrained,
et al. 2009). receiving the credit in advance—the Advance CDCC
(ACDCC)—should be made optional for claimants. Until 2010,
A reform that will spur employment among low-income taxpayers had the option of receiving the EITC throughout
parents, and also expand opportunities for families to place the year in their paychecks (Advance EITC). However, the
their children in quality, center-based care, is to convert the experience with the Advance EITC is generally considered a
CDCC from a nonrefundable credit with no income limit to failure because fewer than 3 percent of recipients opted for
a refundable credit that is targeted to low-income working the advance payment, and those that did frequently made
families. Building off current tax law, I propose the following mistakes (Government Accountability Office 2007).
changes to the CDCC (summarized in table 10-2):
Research suggests that EITC recipients prefer to receive the
• Convert to refundable credit credit as a lump sum, and want to avoid situations where they
receive too large a credit during the year and then are forced
• Convert to child age-dependent credit rate
114 Policies to Address Poverty in America
James P. Ziliak
TABLE 10-2.
Schedule for Refundable Child and Dependent Care Credit
AGI ≤ $25,000 $25,000 < AGI ≤ $70,000 AGI > $70,000
Licensed facility rates
Children under age 5
Credit rate 100% Reduced 10 pp for every $5,000 AGI 0
Credit base $4,000 first child; $6,000 max. $4,000 first child; $6,000 max. 0
Refundable Yes Yes 0
Children ages 5 to 12
Credit rate 70% Reduced 7 pp for every $5,000 AGI 0
Credit base $4,000 first child; $6,000 max. $4,000 first child; $6,000 max. 0
Refundable Yes Yes 0
Unlicensed facility rates
Children under age 5
Credit rate 50% Reduced 5 pp for every $5,000 AGI 0
Credit base $4,000 first child; $6,000 max. $4,000 first child; $6,000 max. 0
Refundable Yes Yes 0
Children ages 5 to 12
Credit rate 35% Reduced 3.5 pp for every $5,000 AGI 0
Credit base $4,000 first child; $6,000 max. $4,000 first child; $6,000 max. 0
Refundable Yes Yes 0
Note: pp = percentage points.
to repay the IRS on April 15 (Romich and Weisner 2000). This the payback is not lump-sum unless the taxpayer does not
makes sense when the mental accounting of the EITC is to report the overpayment until his or her submission of the
apply it toward paying off debt or to make a down payment end-of-year tax return. The United Kingdom offers something
(Gao, Kaushal, and Waldfogel 2009; Smeeding, Ross Phillips, similar. A possible structure for the ACDCC, should the
and O’Connor 2000). However, with regular child-care taxpayer elect to receive it, is to cap the advance portion at
expenses, the ACDCC seems more likely to be used, and more 50 percent of the total prior-year credit and to deposit it in
akin to Supplemental Nutrition Assistance Program (SNAP) equal monthly installments. At the time of tax filing the credit
benefits—formerly known as the Food Stamp Program—that amount (under or overclaim) can be reconciled. Capping it at
are received monthly. 50 percent should reduce the incidence of overclaiming, while
also providing needed assistance throughout the year.
The issue then is how to design the ACDCC with greater success
than the Advance EITC. New Zealand, for example, direct COSTS AND BENEFITS
deposits advance tax credits in the recipient’s bank account There are three primary benefits of a refundable CDCC. First,
each week (or every two weeks, or annually, depending on the by offsetting the costs of child care, the reformed CDCC
recipient’s pay period), and any overpayment is balanced by would encourage greater labor force participation by working
a subsequent payback schedule for the recipient. Generally, parents. This higher labor supply would benefit affected
The Hamilton Project • Brookings 115
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 10: Supporting Low-Income Workers through Refundable Child-Care Credits
families and increase our nation’s productive capacity. Second,
the expanded credit would increase the disposable income of
Questions and Concerns
working families, leading to more resources and improved Why create a new refundable credit in lieu of expanding the
well-being for households with children. Third, subsidized CTC and/or EITC?
child care would allow more working parents to move their
children from informal care arrangements into higher-quality A credit that can be used flexibly like the CTC and the EITC
center-based care. is generally favored by economists, and the refundable CDCC
is more administratively burdensome because of the need to
On the cost side, it is important to acknowledge that not all track expenses, and to track whether the provider is licensed.
groups are held harmless by this proposal. Namely, families However, as discussed previously, the evidence suggests that
making greater than $70,000 would lose eligibility for the the EITC is not spent directly on the child, and there is no
CDCC, which will require those families to bear a slightly evidence on how the CTC is spent. There is some limited
higher tax burden. Moreover, shifting the nonrefundable credit evidence that the expanded generosity of the EITC could lead
that currently benefits high-income families to a refundable to improvements in children’s math and reading achievement,
credit that benefits low- and middle-income families could but the mechanisms are as yet unknown (Dahl and Lochner
reduce labor effort among upper-middle-income families. 2012). If a key goal is to focus policy on boosting employment
Any such effect is likely to be minimal because the current and early childhood development, then a targeted child-care
maximum nonrefundable credit—$600 for one child and credit makes sense, and in fact, would be more target efficient in
$1,200 for two or more children—represents a small share of achieving those dual goals than expanding the CTC or EITC.
income for high-income workers and, as such, is unlikely to be Moreover, while the proposed credit is dedicated to child care
a decisive factor in their labor supply decisions. only, it maintains a high degree of consumer sovereignty akin
to the EITC in that the credit can be received across a host of
An additional potential cost comes from the possibility that providers—public, private, licensed, and unlicensed.
the phase-out range of the refundable CCDC will create
additional disincentives to work. In particular, the phase-out Why not expand the CCDF and run all child-care assistance
tax rates of 10.0 percent and 7.5 percent depending on the age through block grants?
of the child (5.0 percent and 3.5 percent for unlicensed care)
will overlap with the phase-out rates of the EITC (16.0 percent The CCDF provides assistance to TANF and other low-income
for one child and 21.1 percent for two or more children). families, and should be used as a first line of child care for
Research by Eissa and Hoynes (2004) suggests that any these families. However, the reach of this program is very low.
reduced labor supply response will most likely come from the As noted, in 2009 only one of six eligible children was reached
work decisions of married women—whether to work and how by CCDF and TANF child-care programs. On the contrary,
many hours—but the effects are modest. A recent proposal by recent estimates place take-up rates in each of the EITC and
Kearney and Turner (2013) to provide a secondary-earner tax SNAP programs at 79 percent (IRS 2014; U.S. Department of
deduction for earnings up to $60,000, if enacted, is likely to Agriculture 2014). Because the refundable CDCC is a blend of
mitigate any disincentive from the refundable CDCC among the latter two programs, it is expected that take-up rates will
low- and middle-income married couples. be much higher than CCDF/TANF child care.
In terms of tax revenue cost to the government, because Does creating a wedge in the credit’s generosity between
the proposed policy would couple the refundability of the licensed and unlicensed care facilities raise the prospects of
credit with an income limit on eligibility, the lost tax revenue fraudulent claims?
associated with this proposal is likely to be modest. Still, even
The concern is that taxpayers may falsely claim that the
considering that the expanded credit could lead to some tax
provider is licensed, or may not know whether the provider
revenue loss, the benefits of the proposed reform outweigh
is licensed, and claim the higher credit amount when they are
the costs. A sizable child-care subsidy for low- and middle-
only eligible for the lower amount. Estimates in 2011 showed
income working parents will increase the work efforts and
that just over 60 percent of children under age five had a
the returns to work for low- and moderate-income families.
regular child-care arrangement, and of those, 25 percent were
It will make the U.S. tax code more progressive in a way that
in an organized care facility and over 40 percent received care
will likely have no discernible work disincentives for higher-
from a relative, most often a grandparent (Laughlin 2013). This
earning individuals.
suggests that there will be opportunities to game the system. A
way to mitigate such false claims is to not distinguish licensed
from unlicensed facilities, and to offer only a single credit
116 Policies to Address Poverty in America
James P. Ziliak
schedule. However, this does not seem desirable because the rate age-dependent and more valuable for placements in
benefits of quality, center-based care are well established and center-based care recognizes the fact that the cost of center-
the proposed credit is designed to incentivize the use of center- based care is much higher for young children, and potential
based care. Moreover, as noted, states already have a process long-term benefits of making center-based care affordable
of licensing care facilities, and the IRS can utilize this system for low-income parents is backed by evidence (Bernal and
to verify claims. One option would be to require child-care Keane 2011; Blau and Currie 2006; Morris et al. 2009). There
providers to file a Form 1098 documenting the dollar amount is also increasing evidence that making the tax code more
of child-care payments received from the taxpayer. This is akin age-dependent brings us closer to an optimal tax structure
to what a bank does for mortgage interest payments received, (Bastani, Blomquist, and Micheletto 2013; Weinzierl 2011). This
or an educational institution for tuition payments received, is based on the notion of tagging proposed long ago by Akerlof
with the presumption that dual filing by both the payer and (1978), who showed that tax efficiency and redistribution can
recipient will reduce the incidence of false claims. be improved if different tax schedules are applied to readily
verifiable characteristics, which could include the age of the
child as proposed here (Mankiw, Weinzierl, and Yagan 2009).
Conclusion While the size of the benefit is on par with, or larger than,
the current EITC, there is precedent for such tax incentives
The proposed refundable CDCC is highly progressive,
in both the British and New Zealand tax codes, and in both
redirecting current tax expenditures of the CDCC from the
of those countries the size of the child-care benefits are larger.
top two income quintiles to the bottom two quintiles. As
Take-up of the credit, however, is likely to be lower than that
such, this proposal directly addresses the issue of widening
of the EITC, especially among married families, as many will
inequality, creating opportunity for upward mobility in
continue to keep one parent at home to raise children. The
the bottom half of the distribution by making work more
latter, coupled with the fact that families with incomes above
attractive. Importantly, unlike the current CDCC available
$70,000 will no longer be eligible for the CDCC, could easily
only to those with positive tax liability, this new credit is more
leave this proposal revenue neutral or better.
of a complement to the existing EITC; the two can be received
in tandem as refundable credits. Moreover, making the credit
ACKNOWLEDGMENTS: I thank Robert Paul Hartley for excellent research assistance. I am grateful for helpful suggestions
on an earlier draft from Janet Currie, Bill Gale, Ben Harris, Melissa Kearney, Tim Smeeding, Ken Troske, Jane Waldfogel, and
seminar participants at the authors’ workshop at The Hamilton Project. All errors are my own.
The Hamilton Project • Brookings 117
118 Policies to Address Poverty in America
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 11: Building on the Success
of the Earned Income Tax Credit
Hilary Hoynes
University of California, Berkeley
Introduction Yet a program of this size and impact could be more equitable
in its reach. Under the current design of the EITC, childless
The Earned Income Tax Credit (EITC) provides a refundable earners and families with only one child, for instance, receive
tax credit to lower-income working families. In 2011, the EITC disproportionately lower refunds.
reached 27.9 million tax filers at a total cost of $62.9 billion.
Almost 20 percent of tax filers receive the EITC, and the In 2014, families with two children (three or more children) are
average credit amount is $2,254 (IRS 2013). After expansions eligible for a maximum credit of $5,460 ($6,143) compared to
to the EITC in the late 1980s through the late 1990s—under $3,305 for families with one child. Married couples, despite their
Democrat and Republican administrations—the EITC now larger family sizes, receive only modestly more-generous EITC
occupies a central place in the U.S. safety net. Based on the benefits compared to single filers.1 Childless earners benefit little
Census Bureau’s 2012 Supplemental Poverty Measure (SPM), from the EITC, and have a maximum credit of only $496—less
the EITC keeps 6.5 million people, including 3.3 million than 10 percent of the two-child credit.
children, out of poverty (Center on Budget and Policy Prominent proposals seek to mitigate these inequalities.
Priorities [CBPP] 2014a). No other tax or transfer program President Obama’s fiscal year 2015 budget includes an expansion
prevents more children from living a life of poverty, and only of the childless EITC, a concept outlined by John Karl Scholz in
Social Security keeps more people above poverty. 2007 in a proposal for The Hamilton Project. Notably, MDRC
Since the EITC is only eligible to tax filers who work, the is currently evaluating Paycheck Plus, a pilot program for an
credit’s impact on poverty takes place through encouraging expanded EITC for workers without dependent children, for
employment by ensuring greater pay after taxes. The empirical the New York City Center for Economic Opportunity (MDRC
research shows that the tax credit translates into sizable 2014). The recent Hamilton Project proposal for a secondary-
and robust increases in employment (Eissa and Liebman earner tax credit addresses the so-called EITC penalty for
1996; Meyer and Rosenbaum 2000, 2001). Thus, the credit married couples (Kearney and Turner 2013). And the more-
reduces poverty through two channels: the actual credit, and generous EITC credit for three or more children was recently
increases in family earnings. This dual feature gives the EITC enacted as part of the American Recovery and Reinvestment
a unique place in the U.S. safety net; in contrast, many other Act of 2009, and is currently scheduled to sunset in 2017.
programs redistribute income while, at least to some degree, Considering this broad set of EITC reforms, and recognizing
discouraging work. Importantly, transferring income while the demonstrated effectiveness of the program as an
encouraging work makes the EITC an efficient and cost- antipoverty program with numerous benefits, this policy
effective policy for increasing the after-tax income of low- memo proposes an expansion for the largest group of EITC
earning Americans.
The Hamilton Project • Brookings 119
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 11: Building on the Success of the Earned Income Tax Credit
BOX 11-1.
The Earned Income Tax Credit Schedule
Figure 11-1 presents the schedule for the EITC for tax year 2014. The EITC schedule has three regions. In the phase-in region,
the credit is phased in at a constant rate, which is 7.65 percent for taxpayers without children, 34 percent for those with
one child, 40 percent for those with two children, and 45 percent for those with three or more children. In the flat region,
taxpayers receive the maximum amount of the EITC benefit. In the phase-out region, the credit is phased out at a constant
rate: one-child families lose 15.98 percent of each dollar earned due to the lost credit, families with two or more children
experience a 21.06 percent phase-out, and childless filers a 7.65 percent phase-out. The dotted lines in figure 11-1 indicate
the somewhat more-generous schedule for married taxpayers—the Economic Growth and Tax Relief Reconciliation Act of
2001 and later legislation expanded the flat and phase-out regions for married couples; in 2014, the phase-out threshold for
married couples is $5,430 larger than for single filers. This expansion of the schedule for married couples was introduced to
reduce the marriage penalties that the EITC creates for lower-income taxpayers.
To illustrate the mechanics of the credit, consider a single mother with one child earning $15,000 per year. Her earnings
would place her in the flat region of the credit—that is, in the range of income in which a tax unit receives the maximum
credit and in which benefits are neither phased in nor phased out; she would receive an EITC of $3,305. If her earnings were
instead $20,000, she would be in the phase-out region and her credit would fall by $347 to $2,958. In other words, her credit
would equal the maximum credit minus 15.98 percent of all earnings that lie in the phase-out region.
recipients: families with one child. In particular, I propose to in poverty. The proposal would also generate social benefits
expand the one-child schedule to be on par with the two-child through the spillover effects that the increase in income plays
schedule, in equivalence scale-adjusted terms. An equivalence in improving health and children’s cognitive skills (Dahl and
scale captures the cost of living for a household of a given size Lochner 2012; Evans and Garthwaite 2014; Hoynes, Miller,
(and demographic composition) relative to the cost of living and Simon forthcoming).
for a reference household of a single adult, and is a standard
component in defining poverty thresholds. The proposal
expands the maximum credit for one-child families to $4,641, The Challenge
from $3,305 under current law, an increase of about 40 percent.
The expansion will lead to a roughly $1,000 increase in after- The EITC is a refundable tax credit that gives a taxpayer with
tax income for taxpayers in the bottom 40 percent of the no federal income tax liability a tax refund for the full amount
income distribution receiving the higher credit. As this paper of the credit. The amount of the credit depends on filing status,
outlines, the expansion is justified on equity and efficiency number of qualifying children, and earned income (and, for
grounds. This expansion is anchored in the equity principle some taxpayers, adjusted gross income). (The EITC schedule is
in that the generosity of the credit should be proportional to explained in box 11-1.) Because the EITC is one of our nation’s
the needs of families of differing sizes; I use the equivalence most effective antipoverty programs, the challenge considered
scale implicit in the poverty thresholds of the Census SPM as in this policy memo is how to leverage this tool to have even
a guide for household needs. This proposal is also supported greater impact.
by efficiency principles given the EITC’s demonstrated success Enacted in 1975, the EITC’s original intent was to offset payroll
at raising labor supply among single mothers. taxes for low-income families. The EITC has been expanded
The target population for the proposal is low-income working by tax legislation five times in the subsequent years: in 1986,
families with children. Implementing this proposal requires 1990, 1993, 2001, and 2009. Figure 11-2 illustrates these policy
legislative action by the federal government; it is important expansions by plotting the maximum EITC credit by number
to note that altering the EITC schedule requires a simple of children for each year between 1985 and 2014 (in real 2014
amendment to the tax code, and not a massive overhaul of our dollars). The 1993 expansion is the most significant, having
nation’s tax system. The revenue cost of the proposal derives introduced the more-generous schedule for those with two or
from additional federal costs of the EITC, less the additional more children. Additionally, the 1993 expansion introduced
payroll and ordinary federal income taxes. The private the relatively small credit for childless taxpayers. The 2009
benefits include increases in after-tax income and reductions expansion, enacted as part of the federal stimulus, introduced
a separate schedule for those with three or more children.
120 Policies to Address Poverty in America
Hilary Hoynes
FIGURE 11-1.
Earned Income Tax Credit Amount by Earnings Level and Number of Children, 2014
7,000
6,000
Earned income credit (in dollars)
5,000
4,000
3,000
2,000
1,000
0
0 10,000 20,000 30,000 40,000 50,000 60,000
Earned income (in dollars)
No children 1 child 2 children 3 or more children
Source: Urban-Brookings Tax Policy Center 2014.
FIGURE 11-2.
Earned Income Tax Credit Maximum Credit by Number of Children, 1985–2014
7,000
Maximum credit (in 2014 dollars)
6,000
5,000
4,000
3,000
2,000
1,000
0
1985 1990 1995 2000 2005 2010
No children 1 child 2 children 3 or more children
Source: Urban-Brookings Tax Policy Center 2014.
The Hamilton Project • Brookings 121
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 11: Building on the Success of the Earned Income Tax Credit
TABLE 11-1.
Earned Income Tax Credit (EITC) Recipients by Number of Children, 2011
Average credit amount Share of EITC returns Share of EITC benefits
(in dollars) (percent) (percent)
No children 264 25 3
1 child 2,199 36 35
2 children 3,469 27 41
3 or more children 3,750 12 20
All recipients 2,254 100 100
Sources: IRS 2013; author's calculations.
Table 11-1 gives a snapshot of EITC recipients for 2011 (IRS the appropriate adjustments to the cost of living between
2013). A total of 27 million taxpayers received the credit, different family sizes. Using the SPM equivalence scale, the
representing almost 20 percent of all tax filers. The total cost maximum credit for families with two children should be
of the credit in 2011 was $62.9 billion. As a comparison, in about 18.7 percent higher than the maximum credit for one-
2011, payments for the Supplemental Nutrition Assistance child families. Under current law it is 65 percent higher.2 I
Program (SNAP), formerly known as the Food Stamp return to this in the proposal below.
Program, totaled $72.8 billion, and Unemployment Insurance
payments totaled $107.0 billion (Bitler and Hoynes 2013). The EITC is explicitly tied to work. As shown in figure 11-1,
EITC benefits averaged $2,199 for one-child families, $3,469 if a family has no earned income, then it is not eligible for the
for two-child families, $3,750 for families with three or more credit. Overall, the credit subsidizes entering and staying in
children, and less than $250 for taxpayers with no children. the workforce, and redistribution occurs while encouraging
About a quarter of the EITC returns went to taxpayers without work. This stands in contrast to virtually all other elements of
children, 36 percent to those with one child, 27 percent to the U.S. safety net—such as SNAP and Temporary Assistance
those with two, and 12 percent to those with three or more for Needy Families, commonly referred to as welfare—where
children. Overall, the majority (97 percent) of EITC dollars go the largest benefits are transferred to those with no earnings.
to families with children; the small share of dollars claimed This work-promoting earnings subsidy is at the core of EITC’s
among those without children (3 percent) reflects their much cost-effectiveness.
lower potential and actual credit amounts. The empirical research provides robust evidence that the EITC
Figures 11-1 and 11-2 illustrate that the EITC is substantially leads to sizable increases in the employment of single mothers
more generous for families with two or more children than it (Eissa and Liebman 1996; Meyer and Rosenbaum 2000,
is for those with one child. For families with two children, the 2001). For example, Meyer and Rosenbaum (2001) find that
maximum credit is $5,460 and the phase-out range extends to a 10 percent increase in EITC income leads to a 6.9 percent
earned income of $43,756, while for families with one child, increase in employment rates (an elasticity of 0.69). Because
the maximum credit is $3,305 and extends to earned income of the two potential earners in the household, the labor supply
of $38,511. Standard equity arguments would imply that larger predictions are more complex for married couples, generally
families should receive a higher credit than smaller families. suggesting a reduction in employment for secondary earners.
But what is the right adjustment? The needs of a family grow The existing evidence shows that the EITC leads to modest
with each additional child but, due to economies of scale in reductions (an elasticity of 0.267) in the employment of
consumption, not in a proportional way. I use the family-size married women (Eissa and Hoynes 2004). In contrast, we have
adjustment that forms the basis of the poverty thresholds in little empirical evidence on the possible employment effects of
the Census SPM to capture the varying needs across family the credit for taxpayers without children; the MDRC pilot of a
sizes. Known as equivalence scales, they are used to establish childless EITC currently in the field in New York City should
fill this important gap in our knowledge.
122 Policies to Address Poverty in America
Hilary Hoynes
TABLE 11-2.
Details of Policy Proposal by Number of Children, 2014 Tax Year
1 child 2 children 3 or more children
Current Law Proposal Current Law Proposal Current Law Proposal
Phase-in rate 34.00% 34.00% 40.00% 45.00%
Minimum income for maximum credit $9,720 $13,650 $13,650 $13,650
Maximum credit $3,305 $4,641 $5,460 $6,143
No change No change
Phase-out rate 15.98% 21.06% 21.06% 21.06%
Beginning income of phase-out $17,830 $17,830 $17,830 $17,830
Ending income of phase-out $38,511 $39,867 $43,756 $46,997
Sources: Urban-Brookings Tax Policy Center 2014; author’s calculations.
Note: The gray font applies to cells with values that change under the proposal.
The release of the Census SPM in 2011 provides annual reports
on the number of persons lifted out of poverty due to safety net
A New Approach
programs. The EITC lifted 3.3 million children out of poverty, Given the efficient and cost-effective reduction in poverty
more than any other program (CBPP 2014a). SNAP was the that the EITC achieves for families with children, proposals
next largest, with 2.2 million children lifted from poverty are being advanced to expand the EITC for childless
(Short 2013). Overall, the credit lifted 6.5 million individuals taxpayers and for married taxpayers. The proposal outlined
out of poverty (CBPP 2014a). in this paper to raise EITC benefits for the largest group of
recipients—one-child families—is part of this broader set of
These calculations based on the SPM are static; they calculate proposed EITC reforms.
poverty with and without the specific income source (e.g., the
EITC) but do not take into account the behavioral effects of I justify this proposal on the basis of equity and efficiency
that source on employment and earnings. To the extent that grounds: first, as discussed below, based on the principle that
the EITC leads to an increase in employment and earnings, the the credit should be equal across different family sizes in
statistics cited above are underestimates of the full antipoverty proportion to their needs, the EITC for one-child families is
effects of the EITC. below what it should be. Second, I have robust evidence based
on historical expansions that expanding the EITC provides a
Several studies have quantified benefits of the credit beyond cost-effective reduction in poverty for families with children
those on employment, earnings, and income. Dahl and by encouraging more work as the credit on income expands.
Lochner (2012) find that the increase in income through Combined, these justifications are especially important given
the EITC leads to improvements in child test scores. that real household incomes in the lower half of the income
Hoynes, Miller, and Simon (forthcoming) find the increase distribution have stagnated over the past forty years, and that
in income through the EITC leads to an improvement in the highest poverty rates for Americans are found among
infant health by reducing the incidence of low-birth-weight children (Short 2013; U.S. Census Bureau 2014).
births. 3 Evans and Garthwaite (2014) find evidence that
the expansion of the EITC improved health indicators— As presented above, using the family-size adjustment that
measured by blood and medical tests—for mothers, forms the basis of the poverty thresholds in the SPM, the
suggesting a significant relationship between increased maximum credit for families with two children should be
income and a reduction in stress. about 18.7 percent higher than the maximum credit for one-
child families, yet under current law it is 65 percent higher.
My proposal is to expand the EITC to one-child families to
be on par with the maximum credit for two-child families,
in equivalence-scale units. I keep the two-child schedule at
The Hamilton Project • Brookings 123
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 11: Building on the Success of the Earned Income Tax Credit
TABLE 11-3.
Comparison of Current Law and Earned Income Tax Credit Proposal for Hypothetical Families
One-child family One-child family
Current law (2014) Proposal
A. Minimum wage ($7.25 per hour)
Total earnings 15,080 15,080
Payroll taxes –2,307 –2,307
Federal income tax 0 0
Child Tax Credit (including refundable) 1,000 1,000
EITC 3,305 4,651
Child-care costs –1,508 –1,508
Child and Dependent Care Credit 527 527
Family disposable income 16,097 17,443
Increase in income 8%
B. 150% of minimum wage ($10.90 per hour)
Total earnings 22,672 22,672
Payroll taxes –3,469 –3,469
Federal income tax –592 –592
Child Tax Credit (including refundable) 1,000 1,000
EITC 2,531 3,621
Child-care costs –2,267 –2,267
Child and Dependent Care Credit 706 706
Family disposable income 20,581 21,671
Increase in income 5%
Source: Author’s calculations using TAXSIM (see Feenberg and Coutts 1993).
Note: The gray font applies to cells with values that change under the proposal. All figures, with the exception of the increase in income, are in dollars.
current law and as the reference schedule. The specific changes earnings on child care. Panel A considers the case where the
to the tax credit are shown in table 11-2. The phase-in rate for single woman earns the minimum wage of $7.25 per hour.
one-child families remains at 34 percent, as under current law, Under current law (column 1), after child-care expenses and
but the phase-in income range is extended to $13,650 (from payroll taxes, and after federal tax and credits, the family has
$9,720) to match the range used for two- and three-or-more– a disposable income of $16,097.4 In column 2, I show how
child credit schedules. The maximum credit increases from taxes and disposable income change with the policy proposal
$3,305 to $4,641, an increase of 40 percent. The phase-out rate (and no behavioral changes). The EITC rises to $4,651 from
increases from 15.98 percent to 21.06 percent (matching the $3,305 and disposable income rises to $17,443, an increase of
two- and three-or-more–child credit rates) and the phase-out 8 percent. Panel B considers a family where the woman earns
income range extends from $17,830 to $39,867 (compared to 150 percent of the minimum wage ($10.90 per hour). For that
$17,830 to $38,511 under current law). This proposal presents family, the proposal would increase family disposable income
an opportunity to bring the credit for families with children by 5 percent, from $20,581 to $21,691.
into a harmonized schedule, using the equivalence scale in the
SPM as the basis for harmonization. COSTS AND BENEFITS
The costs of the proposal include the federal revenue cost of
To illustrate the effect of this proposal, table 11-3 presents tax expanding the EITC. However, including all economic effects,
and income calculations for hypothetical families with a single namely higher labor supply, the EITC cost would be offset
parent with one child. Assume that a woman works full-time by the additional payroll tax revenue and (ordinary) federal
for the full year and that the family spends 10 percent of gross
124 Policies to Address Poverty in America
Hilary Hoynes
TABLE 11-4.
Simulation of Proposed Policy by Expanded Cash Income Percentile
All taxpayers Taxpayers with children
Tax units Average tax Change in Tax units Average tax Change in
with tax cut cut among after-tax income with tax cut cut among after-tax income
(in percent) beneficiaries (in percent) (in percent) beneficiaries (in percent)
(in dollars) (in dollars)
Bottom quintile 5.5 –1,029 0.4 24.9 –1,051 1.2
Second quintile 11.1 –969 0.3 36.7 –975 0.8
Middle quintile 5.0 –830 0.1 14.2 –739 0.2
Fourth quintile 0.2 –741 0.0 0.3 –714 0.0
Top quintile 0.0 0 0.0 0.0 0 0.0
All 5.0 –958 0.1 15.9 –957 0.2
Source: Urban–Brookings Tax Policy Center microsimulation model version 0613-3 (see Rohaly, Carasso, and Saleem 2005).
Note: Includes both filing and nonfiling units but excludes those that are dependents of other tax units. Tax units with negative adjusted gross income are excluded from their respective
income class but are included in the totals. For a description of expanded cash income, see Urban–Brookings Tax Policy Center (n.d.). The income percentile classes used in this table are
based on the income distribution for the entire population and contain an equal number of people, not tax units. The resulting percentile breaks are 20 percent $17,272; 40 percent $31,839;
60 percent $52,010; 80 percent $82,156; 90 percent $114,150; 95 percent $160,278; 99 percent $376,776; 99.9 percent $1,971,618 (in 2013 dollars).
income tax revenue collected with increases in employment with children, but these estimates illustrate the higher benefit
and earnings. Taxpayers benefit privately from the increase in of this proposal on this select demographic group. Roughly
after-tax income and from the reduction in poverty. Because one-quarter of taxpayers with children in the bottom quintile
the expansion in the EITC is expected to boost employment and over one-third of these taxpayers in the second quintile
and earnings of single-parent families, their income would would see an increase in after-tax income. Among taxpayers
increase through the expanded credit as well as through the with children, those in the bottom quintile would see their
predicted increase in earnings.5 The expansion may also lead after-tax income rise by an average of 1.2 percent; taxpayers
to important social benefits resulting from the increase in with children in the second income quintile would see their
income for these families. Studies find that the increase in after-tax incomes rise, on average, by 0.8 percent. The average
income could yield spillover effects by improving health and benefits for one-child families would be even higher under my
children’s cognitive skills (Dahl and Lochner 2012; Evans and proposal.
Garthwaite 2014; Hoynes, Miller, and Simon forthcoming).
The proposal would have a substantial effect on the well-being
The distributional effects of the proposal, derived from the of low-income families. Using the SPM to define poverty,
Urban–Brookings Tax Policy Center microsimulation model, CBPP (2014b) estimates that this EITC expansion would lift
follow the Joint Committee on Taxation convention of holding 410,000 people—including 131,000 children—out of poverty.
gross domestic product constant and subsequently assuming This proposal would also improve the livelihood of a large
no change in labor supply. As shown in table 11-4, the proposal number of people living below the poverty line. In total, 3
is decidedly progressive, raising after-tax income by 0.4 million people in poverty—including 1 million children—
percent for taxpayers in the bottom income quintile and 0.3 would be made less poor.
percent for taxpayers in the second quintile, with effectively
no impact on taxpayers in the top two quintiles. Tax units These estimates are conservative—that is, taking into account
benefitting from this proposal—8.1 million in total—would behavioral effects and increases in employment and earnings
each see their after-tax income rise by about $1,000. should lead to a reduction in costs (due to the offsetting payroll
and federal income taxes) and an increase in private and social
The benefits would be especially high among taxpayers benefits. The empirical research shows robust evidence that
with one child. The Urban-Brookings Tax Policy Center an increase in the EITC leads to an increase in employment
microsimulation model only shows output for all taxpayers and earnings for single filers. For single parents already in
The Hamilton Project • Brookings 125
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 11: Building on the Success of the Earned Income Tax Credit
the labor market, this proposal provides a simple income encouraging work. While the EITC now forms a central piece
transfer to those eligible for the EITC (assuming no change of the U.S. safety net for families with children, its prominence
in earnings for those already in the labor market). For single does not eliminate the need for other safety-net programs such
women currently out of the workforce, the expanded EITC is as SNAP. Critically, the EITC does not help families in the face
predicted to encourage employment and earnings, leading to of labor market weakness and job loss. I bring attention to this
an increase in after-tax income through the EITC, other tax limitation of the EITC in recent joint research with Marianne
credits, and earnings (less payroll taxes and any owed federal Bitler and Elira Kuka (Bitler, Hoynes, and Kuka 2013). The
income taxes). For married couples, the behavioral effects are implication is that in the post-welfare reform era, the Great
expected to be more muted, with minimal effects for married Recession resulted in more extreme poverty than we would
men and modest reductions in employment and earnings for have expected from experience prior to welfare reform (Bitler
married women. In sum, family resources would increase and Hoynes 2013). SNAP is essential for providing protection,
through earnings and the EITC. especially for keeping families out of extreme poverty (Parrott
2014).
The proposal also comes with costs, foremost among them
the lost revenue and expanded outlays owing to the more- How would your proposal change if you used the
generous credit. The Urban–Brookings Tax Policy Center equivalence scale implicit in the official poverty
microsimulation model estimates that the expanded EITC thresholds rather than the SPM?
would lose roughly $9 billion per year, or $92.8 billion
between 2015 and 2024. For reasons noted above, namely the The main theoretical grounding for our proposal—that
convention that labor force be held constant, this estimate the one-child schedule is too low relative to the two-child
overstates the potential cost of the expansion. A second cost schedule given the difference in family size—holds regardless
is that the expansion increases effective tax rates on those of whether we use the equivalence scale implied by the SPM or
workers whose earnings fall in the phase-out range. For these the official poverty threshold.
workers, the lost benefits for each dollar earned rise from Why not expand the schedule for childless workers?
15.98 percent to 21.06 percent—raising the disincentive to
work. This raise in effective tax rates may slightly offset the The equivalence-scale argument also extends to provide
gains to employment. justification for expanding the EITC for taxpayers without
dependents. Expanding the EITC for childless workers is
supported by many, and recently appears prominently in
Questions and Concerns President Obama’s budget. I see my proposal for the one-child
credit as part of a broader set of policies for expanding and
Given the five prior expansions in the EITC, have we
updating the EITC. I focus on the one-child credit because of
reached the limit of the employment-inducing effects
the robust employment effects found for single mothers and
of the program?
the prevailing unacceptably high child poverty rates, and in
As shown in Jim Ziliak’s proposal in this series, employment an effort to work in concert with these other proposals.
rates for single women with children have declined
Doesn’t your argument imply that the maximum credit
considerably from their peak in 2000. He shows that the
for married couples should be larger than the credit for
employment rate of single mothers with less than a high
single taxpayers?
school diploma and with a child under age thirteen has fallen
10 percentage points from 70 percent in 2000 to 60 percent Yes, it does. Families with two parents have greater needs than
in 2012; it has also fallen for single mothers with more than do families with one parent (for a given number of children),
a high school diploma from 82 percent to 72 percent. While and this is recognized by a larger equivalence scale and poverty
these are higher employment rates than were experienced on threshold. I focus my proposal on expanding the one-child
the eve of welfare reform and the expansion of the EITC in schedule for reasons of cost and in recognition of the broader
the 1990s, we have no evidence that employment rates will not policy context. In particular, there are other policies—notably
respond to the proposed expansion. The Hamilton Project proposal for a secondary-earner tax
credit (Kearney and Turner 2013)—that address the EITC
Can the EITC provide all the safety net we need for
penalty for married couples. Kearney and Turner’s proposal
low-income families?
is motivated by reducing the tax cost of entering work for low-
This proposal is based on the established track record for the and moderate-income families. This has the feature of de facto
success of the EITC in increasing after-tax income through increasing the generosity of the EITC for married couples.
126 Policies to Address Poverty in America
Hilary Hoynes
Conclusion maximum credit for one-child families to $4,641, from $3,305
under current law, for an increase of about 40 percent. This
The EITC occupies a central place in the U.S. safety net. The expansion is predicted to raise after-tax income by about
program raises 6.5 million persons, including 3.3 million $1,000 for 8.1 million working families. I view this proposal as
children, out of poverty. The only program that raises more part of the broader agenda for expanding the EITC, including
Americans above poverty is Social Security. The EITC raises the childless expansion proposed by President Obama and The
after-tax incomes at the bottom of the distribution while Hamilton Project (Scholz 2007), and expansions for married
encouraging employment. It redistributes income through the couples through a secondary-earner tax credit (Kearney and
credit as well as through increases in earnings. I propose to Turner 2013). Together, these expansions will rebalance the
expand the EITC for families with one child, the largest group EITC such that its benefits more-closely match the varying
of EITC recipients. In particular, I propose to expand the one- needs across families of different sizes and so its benefits are
child schedule to be on par with the two-child schedule, in more equitably distributed across the population.
equivalence scale–adjusted terms. The proposal expands the
The Hamilton Project • Brookings 127
128 Policies to Address Poverty in America
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 12: Encouraging Work Sharing
to Reduce Unemployment
Katharine G. Abraham
University of Maryland
Susan N. Houseman
The W. E. Upjohn Institute for Employment Research
Introduction periods of slack demand rather than lay people off. Instead of
letting twenty full-time workers go, for example, a company
During the Great Recession, millions of Americans lost their could achieve an equivalent reduction in force by reducing the
jobs as employers downsized in response to falling demand. hours of 100 employees by 20 percent. Work sharing should
A substantial body of research implies that these job losses be particularly attractive when employers expect the reduction
can lead to significant and persistent problems for affected in the demand for their products or services to be temporary,
workers, including lengthy periods of unemployment, as is often the case during a recession. By avoiding layoffs,
sustained earnings losses, serious health problems, and employers can retain valued employees and avoid screening,
other adverse outcomes (see, for example, Black, Devereux, hiring, and training costs when the economy improves and
and Salvanes 2012; Davis and von Wachter 2011; Jacobson, they want to hire more workers. By adopting work sharing,
LaLonde, and Sullivan 1993; Stevens 1997; Sullivan and von employers also may be able to avoid the adverse effects that
Wachter 2009; von Wachter, Song, and Manchester 2011). layoffs have on employee morale and productivity.
Furthermore, the adverse impacts of job loss may extend to
future generations: there is growing evidence that job loss for Work sharing has been credited with substantially reducing
a parent can lead to lower educational attainment and lower the number of layoffs and mitigating unemployment during
lifetime earnings among their children (see, for example, the recent recession in several other countries. In contrast,
Hilger 2013; Oreopoulus, Page, and Stevens 2008). during the Great Recession only seventeen U.S. states offered
a formal work-sharing option; even where available, employer
Recent public debate about the problem of unemployment— use of this option was very low. The success of work sharing in
and especially long-term unemployment—has focused to a other countries and the lingering impacts of the recession on
great extent on providing extended unemployment insurance the U.S. labor market have spurred growing interest in work
(UI) benefits to support family incomes following a job loss. sharing in this country.
Strategies for preventing layoffs have not received comparable
attention in the United States. By comparison, many other The Middle Class Tax Relief and Job Creation Act of 2012
developed countries have incorporated work sharing into is best known for extending the payroll tax cut originally
their UI systems, permitting the payment of prorated benefits introduced in 2011 and authorizing an extension of emergency
to workers who are kept on the job with reduced hours because UI compensation. But the Act also included several provisions
of slack demand. designed to encourage wider adoption and greater use of
work-sharing programs. Since the recession, an additional
If work sharing was more accessible in the United States, more nine states and the District of Columbia have incorporated
employers might be encouraged to reduce work hours during work-share programs into their UI systems. While the 2012
The Hamilton Project • Brookings 129
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 12: Encouraging Work Sharing to Reduce Unemployment
TABLE 12-1.
Percent of Employees Receiving Work-Sharing Benefits in Selected Countries, 2007–2009
All employees Manufacturing employees
2007 2008 2009 2007 2008 2009
Austria 0.00 0.03 0.63 0.00 0.17 3.41
Belgium 3.22 3.53 5.60 6.44 7.36 16.99
Canada 0.02 0.03 0.34 N/A N/A N/A
Czech Republic N/A 0.61 1.44 N/A 1.59 4.49
Finland 0.36 0.47 1.67 N/A 0.59 2.69
France 0.34 0.31 0.83 N/A 0.53 3.61
Germany 0.08 0.17 3.17 0.17 0.53 12.06
Ireland N/A 0.17 1.03 N/A 0.48 1.34
Italy 0.64 0.78 3.29 1.75 2.29 9.95
Netherlands N/A 0.20 0.75 N/A 1.39 5.01
United States 0.04 0.07 0.22 N/A N/A N/A
Source: OECD 2010b.
Note: N/A = not available. Take-up rate for the United States is computed for the subset of states with short-time compensation programs. These data were provided as a special tabulation
in an EC-OECD questionnaire.
legislation constitutes an important first step, it does not go Work-sharing programs played a substantial role in
far enough. We propose that the federal government take ameliorating the rise in unemployment in many countries
additional actions to encourage the use of work sharing as an during the most recent recession. Although employer work-
alternative to layoffs during future U.S. recessions. sharing plans typically are limited in duration, many countries
extended the permissible length of these plans as the downturn
lingered. Germany, for example, extended the maximum
The Challenge length of a work-sharing plan from six months to twenty-four
months for applications submitted in the second half of 2009
In many developed countries, when economic conditions
and to eighteen months for applications submitted in 2010
weaken, employers may choose to cut employee hours and
(Crimmann and Wiessner 2009; International Labour Office
have those workers receive prorated UI benefits in lieu of laying
[ILO] 2010). Countries also took steps to reduce employers’
workers off. Germany, which has had a work-sharing program
costs for using work sharing. Germany temporarily excused
since the 1920s, was the first to incorporate work sharing into
employers from paying a portion of the social security
its UI system. Italy and Norway introduced formal work-
contribution on hours not worked for which they otherwise
sharing programs in the 1950s; Austria, France, and Ireland
would have been liable (ILO 2010). As shown in table 12-1, when
in the 1960s; and Belgium, Canada, Denmark, Japan, and
usage peaked during 2009, 1 percent or more of the workforce
Luxembourg in the 1970s (Boeri and Bruecker 2011). Many
was collecting work-sharing benefits in six countries, and in
countries with established work-sharing programs also have
three of those countries participation in work-sharing plans
employment protection laws that mandate significant advance exceeded 3 percent. To place these numbers in perspective,
notice before a worker can be laid off and substantial severance in 2009 the number of people on work-share programs was
payments in the event a layoff occurs. Work-sharing programs 68 percent of the number of unemployed in Belgium, 38
can serve as an important complement to strong employment percent of the number of unemployed in Germany, and 39
protection legislation by facilitating reductions in the average
percent of the number unemployed in Italy. A study by OECD
weekly hours employees work (Abraham and Houseman 1993,
researchers concluded that work-sharing programs helped
1994; Boeri and Bruecker 2011).
to preserve jobs during the recent recession, with the impact
being particularly significant in Finland, Germany, Italy, and
Japan. In these countries the OECD researchers estimated
130 Policies to Address Poverty in America
Katharine G. Abraham, Susan N. Houseman
that the decline in permanent employment would have been have approached 1 million. In other words, with work-sharing
about three-quarters of a percentage point greater in the usage at European levels and assuming that work-sharing
absence of work sharing (OECD 2010a).1 Subsequent research expansions translate directly into reductions in the number of
has reached similar conclusions about the role of work sharing layoffs, as many as one in eight of the roughly 8 million jobs
in preventing employment losses during the recession (Boeri lost during the recession could have been saved (Abraham and
and Bruecker 2011; Hijzen and Martin 2013). Houseman forthcoming).
U.S. institutions and employers, in contrast to those in many Several factors beyond the relative ease and modest cost of
other countries, historically have favored the use of layoffs laying off workers are generally cited for the low use of work-
over work sharing. The United States has no requirement sharing programs in the United States. First and foremost is
that employers provide laid-off workers with severance lack of information about the availability of this option in states
payments, and advance notice provisions in U.S. law are weak. with work-sharing programs. Prior to the passage of the Middle
Employees with a sufficient work history are entitled to collect Class Tax Relief and Job Creation Act of 2012, there was some
UI benefits if they are laid off, and under states’ experience ambiguity about the legality of state work-sharing programs
rating systems, employers typically are liable for reimbursing under federal law. This may have discouraged some states from
the state UI trust fund for benefit costs received by laid-off adopting these programs or promoting their use (Balducchi
employees. Although this means that the United States UI and Wandner 2008). In addition, insufficient funding from
tax system imposes some costs on employers who engage in the federal government to administer the program may have
layoffs, these employers generally do not bear the full cost of deterred states from advertising it to employers. With some
the benefits their former employees collect.2 At the same time, exceptions, most notably Rhode Island, states with work-
support for work sharing through the payment of prorated UI sharing programs have done almost nothing to promote
benefits to employees working reduced hours has been much them; as a consequence, many employers are unaware that
less prevalent in the United States. Although seventeen states the programs even exist. In contrast, state officials in Rhode
had work-sharing programs on the books at the end of 2007 Island have promoted the program enthusiastically, and Rhode
and several more have introduced such programs in the past Island’s take-up of this option has been comparable to that of
few years, twenty-four states still do not have work-sharing some European countries (see box 12-1).
programs in operation, and usage of the programs that do exist
remained at very low levels through the recession (Abraham The administrative burden of participating in a work-
and Houseman forthcoming). Together these factors—weak sharing program also may have deterred employers from
employment protection laws, imperfect experience rating of participating. Besides submitting a work-share plan to the
UI taxes, and weak or absent work-sharing programs—help state for approval, employers must certify on a weekly or
explain U.S. employers’ much greater reliance on adjustment biweekly basis that the program is still operative, identify
of employment levels and correspondingly lower reliance on which employees are affected, and document the reductions
adjustment of average worker hours during recessions. in their hours. Cutting through this red tape is made more
difficult because the application and continued claims
Back-of-the-envelope calculations suggest that moderately processes typically are not automated.
greater use of work sharing in the United States could
have significantly reduced job loss and thereby mitigated Additionally, some states prohibit certain employers—such
unemployment during the Great Recession. Assuming that as those who have reached the maximum UI tax rate or have
hours reductions through work sharing offset hours reductions negative UI account balances because of UI benefits paid to
through layoffs on a one-to-one basis—perhaps an overly previously laid-off employees—from participating in their
generous assumption, but useful as a first approximation— work-share program. Similarly, in some states employers
work sharing by U.S. employers in 2009 reached a level who have used the UI system intensively in the past may
sufficient to have prevented the loss of only about 22,000 face a higher effective UI tax rate if they use the work-share
full-time-equivalent (FTE) jobs. Had usage in all states been program than if they lay off workers. While these provisions
as large as in Rhode Island, the state with the highest work- are designed to prevent employer abuse of the program, they
sharing rates, the average number of FTE workers on work may unnecessarily restrict access to work sharing, particularly
sharing in 2009 would have been approximately ten times during recessions.
as large as the number actually observed—in the vicinity of The Middle Class Tax Relief and Job Creation Act of 2012
220,000 FTEs rather than 22,000 FTEs. And had the average provided explicit authorization for work-sharing programs
take-up rate been similar to that in Germany or Italy in 2009, meeting certain conditions that are set out in the legislation.
the average number of FTE workers on work sharing would Under the Act, states with work-sharing programs that
The Hamilton Project • Brookings 131
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 12: Encouraging Work Sharing to Reduce Unemployment
BOX 12-1.
Work Sharing in the Great Recession: The Case of Rhode Island
Rhode Island has considerably more take-up of its work-sharing program than other states, and is often referenced as a model
program. Rhode Island’s work-sharing success was especially notable during the Great Recession. In 2007, the state made
one new work-sharing payment for roughly every twenty new standard UI payments. By the height of the Great Recession in
2009, Rhode Island paid one new work-sharing claim for roughly every six new standard UI claims (Shelton 2011).
Interestingly, the greater use of work sharing in Rhode Island can be attributed largely to factors unrelated to program
design: the parameters of Rhode Island’s work-sharing program are representative of those found in other participating
states. In contrast to other states, Rhode Island aggressively marketed work sharing to employers engaged in layoffs during
the Great Recession and made use of the media to highlight potential work-sharing benefits. According to Ray Filippone,
former UI director in Rhode Island, several strategies used in Rhode Island are critical to getting the word out to employers
about work sharing:
• Involve other agencies and stakeholders. It is important to work with other agencies and stakeholders, such as the governor’s
office, legislative offices, and chambers of commerce. Typically, the UI office is not the first point of contact for employers
who are deciding whether or not to lay off workers. Consequently, staff in these other organizations need to be familiar
with the work-sharing program so that they can inform employers about this option.
• Have a dedicated person who can answer employers’ questions about the program. Although other government representatives
or business leaders can tell employers about the work-sharing option, employers interested in utilizing the program will
need to contact the state UI office for more information. It is essential to have staff dedicated to answering employers’
questions about the program. Employers contemplating a work-sharing plan cannot wait a week or two for someone to
answer their questions.
• Have good presentation materials. Having a good presentation about the potential benefits of using work sharing over
layoffs that can be given to employer groups, workforce investment boards, or other interested parties is important.
• Contact employers engaged in layoffs. UI claims staff can flag employers making a lot of layoffs during a recession. Staff then
can contact those employers to make sure they are aware of the work-sharing option and its potential benefits.
• Automate the application and claims process. Weekly or biweekly certifications, which generally are required of employers
on a work-sharing plan, can be burdensome. An automated system reduces the administrative burden on employers and
can make the program more attractive to them.
conform to the new federal law are eligible for a share
of $100 million in grant funding to be used for program
A New Approach
implementation, such as automation of state systems for The federal government should take several additional steps to
the filing and processing of work-sharing claims, and for make work sharing more available as an option for employers
employer outreach. In addition, the law provides full federal and to encourage the use of work sharing as an alternative to
reimbursement for all of the benefits paid out under approved layoffs during future recessions.
state work-sharing laws for up to a three-year period ending in
August of 2015. While the 2012 law undoubtedly has helped MAKE WORK-SHARING PROVISIONS A
to raise the level of interest in work sharing, more needs to be REQUIREMENT FOR STATE UNEMPLOYMENT
done if work sharing is to become a significant weapon in the INSURANCE PROGRAMS
United States’ countercyclical policy arsenal. In the United States, UI is administered as a federal–state
system. Although states’ laws vary with respect to factors
such as exactly how eligibility for UI benefits is determined,
132 Policies to Address Poverty in America
Katharine G. Abraham, Susan N. Houseman
the level of benefit payments, and maximum weeks of We also recommend that states be prohibited from (a) assessing
benefits, the federal government sets minimum conditions work-sharing employers a higher UI tax rate than they would
that state law must satisfy. If state law does not conform to face if the same amount of benefits were paid to laid-off workers
the federal requirements, employers in the state are not or (b) excluding employers from participation in work sharing
eligible to receive the credit against the 6.0 percent federal based on their past use of the UI system. Among the twenty-
UI tax that is otherwise available (normally 5.4 percent, six states plus the District of Columbia that, as of this writing,
lowering the effective federal tax rate to 0.6 percent) and the have work-sharing laws in force, in three states employers who
nonconforming state is not entitled to receive federal grants choose work sharing may incur higher UI tax charges than
to cover the costs of administering its program. We propose they would if the same total benefit payments been generated
that inclusion of a work-sharing provision in the state’s law through layoffs (because of the tax schedule that is applied),
be made a conforming requirement for participation in the and in another seven states employers that already pay the
federal–state UI system. Such action would make the work- maximum tax rate or that possess negative reserve balances
sharing option available to employers and their employees in are precluded from participating in work sharing. These rules
the twenty-four states that currently do not offer the program. unnecessarily impede the use of work sharing.4
CHANGE FEDERAL REQUIREMENTS TO PROHIBIT PROVIDE STATES WITH ADEQUATE CAPACITY AND
CERTAIN PROVISIONS OF STATE WORK-SHARING FUNDING TO OPERATE AND PROMOTE THEIR WORK-
PROGRAMS THAT MAY DISCOURAGE EMPLOYER SHARING PROGRAMS
PARTICIPATION
In most states, the process for handling work-sharing claims
To make state work-sharing programs more attractive to is less automated than the process for handling regular UI
employers, we recommend that the federal government make claims. The funding provided under the 2012 federal legislation
several changes to the criteria such programs must meet. should help to address this problem, but time is running
Under current federal law, a state must require, among other out for states to access this money. States cannot apply until
things, that participating employers submit a work-sharing they have passed a new work-sharing law or amended their
plan; that the proportional reduction in hours under the preexisting work-sharing law to satisfy all of the requirements
employer plan not be less than 10 percent nor more than 60 of the Middle Class Tax Relief and Job Creation Act; in
percent; that employees be offered prorated benefits based on addition, applications for grant funding must be submitted by
the reduction in their hours; that an employee be considered the end of 2014. Most states with existing work-sharing laws
to have satisfied applicable job search requirements so long as are expected to meet this deadline, but some likely will not,
they are available to work their regular work week; and that, if and nearly half of states do not yet have a work-sharing law
health and retirement benefits are provided at the work place, in place. We recommend that, at a minimum, the deadline
the employer certify that they will not be reduced. be extended for states to submit their applications for the
federal grant funding provided in the 2012 law to help with
The last of these conditions, on health and retirement benefits, automation of state systems and outreach to employers.
may dissuade some employers from using work sharing in
lieu of layoffs. An employer who lays workers off sheds any It also will be important to ensure that concerns about the
associated health and retirement plan costs; an employer level of ongoing funding do not deter states from promoting
who uses work sharing does not. Yet there is a strong public their work-sharing programs. The allocation of funding that
interest in ensuring that individuals continue to have health states receive to administer their UI programs is based on a
insurance coverage during a recession. And, in many cases, Department of Labor formula that incorporates information
the Affordable Care Act would not permit employers to reduce about state workloads and state labor costs. States with higher
the health insurance coverage available to workers whose workloads according to this formula get a larger funding
hours are temporarily reduced even if this were permitted allocation. Work-sharing claims are counted as part of the
under a state’s work-sharing law.3 Recognizing these workload, but the workload formula does not reflect additional
concerns and complications, we propose to retain the current tasks that are necessary to operate a successful work-sharing
requirement that employers continue full health insurance program. For example, state staff must review each employer
benefits for employees who participate in work sharing. We work-share plan that is submitted, but the number of such
would, however, eliminate the requirement to maintain full reviews is not an element in the workload matrix. Anticipating
contributions to employee retirement plans. Instead, we would that work sharing will become a more important part of the
apply the less-stringent requirement that employers provide UI system in the future, we recommend that the Department
prorated retirement benefits to employees on work share based of Labor carry out a study to determine how the operation
on the fraction of regular hours their employees work. of a work-sharing program impacts a state’s administrative
The Hamilton Project • Brookings 133
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 12: Encouraging Work Sharing to Reduce Unemployment
workload and modify its funding formula accordingly. Some funding for half of state work-sharing benefits is triggered
level of funding for promoting the program to employers who whenever extended UI benefits under the Federal-State
could benefit from it ideally also would be provided. Changes Extended Unemployment Compensation Act of 1970 are
to the formula used to allocate funding to states could be made triggered. Congress also would enact legislation to add the
administratively and so would not require new legislation. following requirements for state participation in the federal–
Additional appropriations would be needed to increase the state UI compensation system: (1) the participating state
total pool of administrative funding available to be allocated. has a work-sharing program, (2) the state’s work-sharing
program does not charge employers who use work sharing
SUBSIDIZE WORK-SHARING PAYMENTS DURING higher UI tax rates than they would charge employers if the
ECONOMIC DOWNTURNS same benefits were paid to a laid-off worker, and (3) the state’s
Perhaps most importantly, we propose that mechanisms be work-sharing program does not prohibit employers from
put in place to expand support for work sharing automatically participating in work sharing based on their past usage of
during economic downturns. The existing treatment of UI the UI system. Congress would need to include provisions in
benefit durations provides a model for how this might work. this legislation stipulating that employers’ UI accounts not
Since 1970 the maximum duration of UI benefits in a state has be charged for any work-sharing benefits for which the state
risen automatically when the state experiences a significant is receiving federal reimbursement. And when Congress
increase in unemployment. The federal government ordinarily passed legislation to extend the maximum duration of UI
covers half the cost of these so-called extended benefits; during benefits, it would need to include 100 percent support for any
the recent recession, the federal government picked up their work-sharing benefits paid.
full cost. Additional increases in benefit duration, such as those
The Department of Labor would use its statutory authority
created through the emergency UI compensation program
to modify the formula for allocating UI administrative
that existed in several different forms from June 2008 through
dollars to states to ensure that adequate support is provided
December 2013, often are passed into law by Congress during
for operating a work-sharing program. Congressional action
recessions. The cost of such legislated extensions ordinarily is
to raise the total funding available for UI administrative
covered fully by the federal government.
expenses might be needed to ensure that increased funding
Given the value of keeping workers on the job during economic for the administration of work-sharing programs do not lead
downturns, similar steps should be taken to increase the to other UI operations being shortchanged and that states
support provided for work sharing during periods when have the capacity to take appropriate steps to make employers
the economy is weak. More specifically, we recommend aware of the work-sharing option.
that federal funding to cover half of the benefits paid under
COSTS AND BENEFITS
approved employer work-sharing plans be triggered whenever
extended UI benefits under the 1970 law are triggered. In Given the costs imposed by unemployment, increased
addition, we recommend that, whenever Congress enacts substitution of work sharing for job layoffs in the United
legislation to extend the maximum duration of UI benefits, States would have many benefits. The varied costs of
this legislation also include 100 percent federal support for unemployment range from diminished health of laid-off
work-sharing benefits. Finally, we recommend that employers’ workers to lower lifetime earnings, and are well-established
UI accounts not be charged for the cost of any work-sharing in the academic literature. And as noted earlier, work sharing
benefits for which the state is receiving federal reimbursement, is an especially promising remedy for unemployment; if used
thus boosting employer incentives to use work sharing in lieu at the levels seen in some European countries, work sharing
of layoffs during periods in which unemployment is already could potentially have saved up to one in eight of the jobs
high. Most states would need to make changes to their UI laws lost during the Great Recession. With fewer people losing
in order to permit noncharging of employers during future their jobs, unemployment—most importantly long-term
periods of federal funding for work-sharing benefits. To unemployment—could have been alleviated. In addition, an
facilitate these changes, we recommend that the Department increased reliance on work sharing would lower job turnover
of Labor be directed to provide model legislative language for rates and the associated firing, hiring and training costs.
the states.
Expanding work sharing in the United States could, however,
IMPLEMENTATION have significant economic costs. Work-sharing programs are
intended for businesses experiencing temporary reductions
These proposals would be enacted through a combination
in demand, as is particularly common during recessions. But
of legislation and administrative actions by the Department
recessions also serve to weed out inefficient businesses and
of Labor. Congress would enact legislation so that federal
134 Policies to Address Poverty in America
Katharine G. Abraham, Susan N. Houseman
improve the allocation of resources in the economy. A major growing enterprises and sectors (see, for example, OECD
concern about expanding work-sharing programs is that they 2010a). During a recession, however, firms typically have little
will impede needed structural adjustment in the economy difficulty in attracting new recruits, and any effect of work
(OECD 2010a). Given that there are limits on the length of sharing on the pace of economic reallocation cannot be large.
time that a work-sharing plan can be in effect, however, any
impediments to structural adjustment would likely be minor.
Also, given the large number of individuals seeking work Conclusion
during a recession, firms that are hiring during recessions
High unemployment during recessions has lasting adverse
generally will not have difficulty finding qualified workers.
effects for workers who lose their jobs and for future
On balance, in view of the high individual and social costs generations. Work sharing has been an effective policy tool in
associated with unemployment and the relatively low risk other developed countries to combat unemployment during
of significantly inhibiting structural change, the benefits recessions, but has been little-used in this country. While
associated with expanding work-share programs likely some doubt work sharing could ever be used successfully in
outweigh the costs. While work sharing may not be a panacea a country with few impediments to layoffs, many American
for reducing painful adjustment in the labor market, the employers, when faced with a temporary reduction in
United States could benefit from using it more extensively. demand, would like to retain valued employees and could
benefit from a work-sharing program. Other factors,
including lack of information about the work-sharing option
Questions and Concerns and features that tilt the UI system in the United States
toward layoffs, likely have inhibited broader adoption of
Given the relative ease of firing workers in the United States the program by states and greater use by employers in states
compared to some other countries, will U.S. employers shift where the program is available.
from using layoffs to work share in sufficient numbers to
have a noticeable impact on unemployment? Measures passed as part of the Middle Class Tax Relief and
Job Creation Act of 2012 will help reduce the barriers to work
Some might argue that efforts to promote work sharing in sharing in states. We argue, however, that stronger action at the
the United States are doomed to fail, given the relative ease federal level is needed to reduce the bias in the UI system that
of hiring and firing in this country. Part of the reason work favors layoffs instead of work sharing. These include effectively
sharing is attractive to employers in other developed countries mandating that the twenty-four states currently without work-
is that requirements for advance notice and severance sharing provisions in their UI system adopt them; changing
payments to laid-off workers make layoffs costly, which federal requirements concerning maintenance of full
increases the appeal of work sharing as an alternative. The retirement benefits, the UI tax rates assessed on work-share
much weaker notice requirements applicable to layoffs in the employers, and program eligibility under state work-sharing
United States and the fact that employers generally are not laws to mitigate existing incentives to lay off workers and
required to make severance payments to laid-off workers may increase employer take-up of these programs; and providing
be an important explanation for the very low take-up of work states with adequate funding to operate their work-sharing
sharing in this country. programs. Most importantly, we recommend that automatic
mechanisms be put in place for federal support of work-
The experience in Rhode Island, however, provides a basis for
sharing benefits during periods of high unemployment, in
optimism about what it is possible to accomplish in the United
the same way that the federal government supports extended
States, even without the changes we have recommended to
regular UI benefits for individuals who have been laid off
make work sharing more attractive to U.S. employers. The
during such periods.
level of usage in Rhode Island was similar to that in countries
such as France and the Netherlands, and suggests that other
factors are behind the low use of work sharing in this country.
By reducing layoffs, will work-share programs inhibit
needed structural adjustment?
Some express concern that such measures could cause workers
employed at declining enterprises to delay seeking alternative
employment, thereby impeding needed reallocations to
The Hamilton Project • Brookings 135
136 Policies to Address Poverty in America
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 13: Designing Thoughtful Minimum
Wage Policy at the State and Local Levels
Arindrajit Dube
University of Massachusetts Amherst
Introduction that state and local governments take into account the local
cost of living as a relevant consideration in setting a minimum
Rising wage inequality and stagnant real wages have wage, and I provide estimates of how state minimum wages
contributed to inequality in family incomes during the would vary if they reflected cost-of-living differences. I also
past three decades. While the expansion of the Earned recommend the use of regional consumer price indexes (CPIs)
Income Tax Credit (EITC) and the Supplemental Nutrition to index the local minimum wage. Finally, I propose that cities
Assistance Program (SNAP) have helped mitigate the impact and counties coordinate regional wage setting to mitigate
on low-income families (Bitler and Hoynes 2010), federal possible negative effects of local mandates.
minimum wage policy has not contributed to the solution.
The federal minimum wage has failed to keep pace with both The implementation of the state and local framework does not
the cost of living and the median wage in the labor market. override the need for reform at the federal level. Thoughtful
As a consequence, working full-time at the minimum wage reforms to the federal minimum wage can help reduce poverty
does not allow many families to escape poverty, or to attain and mitigate inequality. The federal minimum wage has been
economic self-sufficiency. the focus of substantial debate by academics and policymakers;
this proposal focuses on state and local reforms that have
State and local governments can set minimum wages in received substantially less attention. These state and local
excess of the statutory federal minimum wage.1 Indeed, state reforms can be an important part of the policy portfolio for
and local governments have played an important role in reducing the incidence of poverty and for helping low-income
establishing minimum wages across the country; as a result, families support themselves as they strive toward the middle
thirty-seven states had state minimum wages exceeding the class. In particular, although the federal minimum wage serves
federal level in 2007 prior to the most recent federal increase. as a floor in the labor market, there is some room for additional
Cities, too, have begun setting higher minimum wages, as increases in higher-wage areas.
evidenced by city-level wage minimums in Albuquerque,
San Francisco, San Jose, Santa Fe, Seattle, and Washington,
DC; other cities are actively exploring possibilities of raising The Challenge
minimum wages.
RISING INEQUALITY AND STAGNANT WAGES
In this policy memo, I propose a framework for effective state For much of the past three decades, the wages of those at the
and local minimum wage policy. First, I propose using half bottom of the wage distribution have failed to keep up with
the local-area median wage as an important gauge for setting overall economic gains. Most of the wage increase has occurred
an appropriate level of the minimum wage. Second, I propose among the top half of the wage distribution, especially since
The Hamilton Project • Brookings 137
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 13: Designing Thoughtful Minimum Wage Policy at the State and Local Levels
the 1990s. Wages in the lower half rose only during the period These concerns are exacerbated in states and localities with high
of low unemployment in the late 1990s. As a result, the 90th costs of living. In these areas, workers earning the minimum
percentile real wage grew by over 30 percent between 1973 and wage are especially challenged to pay for food and housing,
2011, while the median and 10th percentile real wages grew by or obtain other necessary goods and services. Effectively, to
less than 5 percent over the same period. escape poverty these workers must earn significantly more
than their counterparts in low-cost areas. Workers in areas
Many factors spurred this dramatic rise in wage inequality, with high median wages, which are often those with high costs
including technological change, de-unionization, increased of living, are also subjected to greater levels of local income
trade and offshoring, and deregulation (Autor, Katz, and inequality. In short, the problems associated with a stagnant
Kearney 2008; Firpo, Fortin, and Lemieux 2011; Philippon and and inadequate minimum wage are exacerbated in high-cost,
Reshef 2012). However, there is also evidence that a falling real high-wage areas.
minimum wage has contributed to this growth in inequality.
In particular, Autor, Manning, and Smith (2014) find that Low minimum wages are also problematic when they deviate
movements in the minimum wage played an important too far from the median wage because they are a reflection
(though not predominant) role in determining the 50/10 wage of the bottom of the wage distribution falling behind the
gap—a measure that highlights wage inequality in the bottom rest of the distribution. For this reason, economists often
half of the distribution by comparing how middle earners consider the ratio of the minimum to the average or median
(50th percentile) fared relative to the lowest earners (10th wage, also known as the Kaitz index. There are three reasons
percentile). The decline in the value of the minimum wage has to pay attention to this measure, especially using the median
also had a larger effect on inequality for female workers since as the reference wage. First, a comparison of the minimum
they tend to be paid less than male workers. wage to the median offers a guide for how binding a particular
minimum wage increase is likely to be, and what type of wage
A DECLINE IN THE MINIMUM WAGE the labor market can bear. When this ratio is low—say around
The federal minimum wage, which has not kept up with the 0.2—minimum wage policy is not raising the wages of many
cost of living, reached its high-water mark in 1968. While the workers. In contrast, a high ratio—say around 0.8—indicates
specific value varies with the price index used, all measures a highly interventionist policy where the minimum wage is
point toward the real minimum wage falling over time.2 Using dramatically compressing differences in wages for nearly half
the CPI-U-RS—a revised inflation index that uses current the workforce. Second, this comparison also provides us with
methods for computing inflation—the minimum wage in a natural benchmark for judging how high or low a minimum
2014 dollars stood at $9.59 per hour in 1968 and $8.58 per hour wage is across time periods or across countries that vary in
in 1979. During the 1980s, the real minimum wage declined terms of their labor markets and wage distributions. Third,
substantially, and over the intervening twenty years it has the median wage also provides a natural reference point for
largely treaded water, reaching a historical low of $6.07 per judging what is a reasonable minimum wage level: no one
hour in 2006 prior to the last federal increase. It now stands expects that the minimum wage should be set equal to the
at $7.25 per hour. median wage, but fairness may become a factor when the
minimum wage falls below, say, one-fourth or one-fifth of the
The failure of the minimum wage to keep up with inflation median wage.
means that, for workers earning the minimum wage, each
hour of labor purchases fewer goods and services. And since A natural target is to set the minimum wage to half of the
measures of poverty are indexed to inflation, an unindexed median full-time wage. This target has important historical
minimum wage means that these workers must work more precedence in the United States: in the 1960s, this ratio was
hours to stay above poverty. Recent evidence suggests that 51 percent, reaching a high of 55 percent in 1968. Averaged
workers earning close to the minimum wage are increasingly over the 1960–1979 period, the ratio stood at 48 percent.
those who rely on their earnings to support necessary Approximately half the median full-time wage is also the
household consumption, as opposed to those who are norm among all OECD countries with a statutory minimum
dependents of workers with higher earnings. For example, wage. For OECD countries, on average, the minimum wage in
between 1979 and 2011, the share of low-wage workers— 2012 (using the latest data available) was equal to 49 percent
defined as those with hourly wages of $10.00 or less in 2011 of the median wage; averaged over the entire sample between
dollars—who are younger than twenty-five fell from 47.1 1960 and 2012, the minimum stood at 48 percent of the
percent to 35.7 percent (Schmitt and Jones 2012). median (OECD 2013). In contrast, the U.S. minimum wage
now stands at 38 percent of the median wage, the third-lowest
138 Policies to Address Poverty in America
Arindrajit Dube
FIGURE 13-1.
The Ratio of Minimum to Median Full-Time Wage: United States and OECD Countries,
1960–2012
0.8
Turkey
0.7
Minimum-to-median wage ratio
France
0.6
Portugal
Australia
0.5
United Kingdom
Canada
Greece
0.4 Japan
Czech Republic
0.3
0.2
1960 1970 1980 2000 2010
United States OECD average
Sources: OECD 2013; author’s calculations.
Note: Data were not available for the full period between 1960 and 2012 for each country. For that reason, the OECD average for each year is derived using the individual country ratios that were
available for that year.
among OECD countries after Estonia and the Czech Republic international average and with the U.S. historical average
(ibid.). (See figure 13-1.) during the 1960s and 1970s. For the purpose of national and
international comparability, table 13-1 shows the value of one-
half the median full-time wage in 2012 for each state, adjusted
A New Approach to 2014 dollars. Since wages vary substantially by state, the
median-adjusted target minimum wage ranges between
Adequate state and local minimum wages play an important $12.45 (Massachusetts) and $7.97 (Mississippi). Fourteen
role in the antipoverty agenda and can compensate for states—mostly those in the Northeast and on the West
inaction at the federal level. To ensure that wages sufficiently Coast—would see their minimum wage rise above $10.00 per
support the lowest-paid workers, I propose that state and local hour with this proposal. In contrast, eighteen states would
governments gauge their minimum wage to half the local-area see their minimums set below $9.00 per hour. It is important
median wage. In addition, I propose that states consider the to note that the proposed minimum wage would exceed the
local cost of living when establishing a minimum wage, and current federal minimum of $7.25 in all states.
that the statutory minimum wage be automatically indexed
to inflation to protect against real declines in the wage floor. State-level add-ons to the minimum wage thus seem to be
Finally, I propose that local governments engage in regional a sensible strategy in these high-wage states. Indeed, many
wage setting to protect against the unintended consequences states are already doing this: as of now, eleven of the fourteen
of raising the minimum wage. states whose target minimum wage exceeds $10.00 per hour
currently have state minimums exceeding $7.25 per hour.
STATE-LEVEL POLICIES
When we factor in current and planned minimum wage
State initiatives are a sensible strategy in many places with increases by states, raising the minimum wage to half the
particularly high wages. One way to gauge what constitutes a median full-time wage in each state by 2016 would entail a
reasonable target level is to consider the ratio of the minimum 26.2 percent increase in the statutory minimum wage. (This
to the median wage: a value of 50 percent is in line with the estimate is a population-weighted average over all fifty states
The Hamilton Project • Brookings 139
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 13: Designing Thoughtful Minimum Wage Policy at the State and Local Levels
TABLE 13-1.
Target Minimum Wage by State, Adjusted Based on Median Wage and Regional Price Parity
Median Wage– Regional Price Median Wage– Regional Price
Adjusted Parity–Adjusted Adjusted Parity–Adjusted
(in dollars) (in dollars) (in dollars) (in dollars)
Massachusetts 12.45 10.45 Indiana 9.41 8.88
Connecticut 12.01 10.67 Missouri 9.35 8.59
Maryland 11.69 10.85 Iowa 9.30 8.73
New Jersey 11.45 11.12 Arizona 9.27 9.56
New Hampshire 11.20 10.35 North Dakota 9.21 8.81
Alaska 10.96 10.44 Hawaii 9.07 11.43
Rhode Island 10.96 9.62 Florida 9.06 9.63
Virginia 10.83 10.06 Nevada 8.99 9.57
Washington 10.76 10.06 New Mexico 8.96 9.24
New York 10.46 11.25 Ohio 8.96 8.70
Minnesota 10.36 9.51 Kansas 8.85 8.77
California 10.21 11.01 Texas 8.82 9.41
Colorado 10.18 9.91 Idaho 8.77 9.13
Illinois 10.07 9.81 Montana 8.71 9.18
Delaware 9.96 9.97 Nebraska 8.71 8.78
Michigan 9.96 9.20 Oklahoma 8.71 8.77
Pennsylvania 9.96 9.62 South Carolina 8.71 8.84
Utah 9.96 9.44 Tennessee 8.71 8.84
Oregon 9.69 9.63 North Carolina 8.64 8.93
Wyoming 9.62 9.40 Alabama 8.54 8.59
Wisconsin 9.60 9.06 Kentucky 8.37 8.66
West Virginia 9.54 8.64 South Dakota 8.30 8.60
Georgia 9.46 8.97 Louisiana 8.14 8.91
Maine 9.46 9.58 Arkansas 7.97 8.54
Vermont 9.46 9.84 Mississippi 7.97 8.42
Sources: Unicon Research Corporation 2012; Bureau of Economic Analysis n.d.; author’s calculations.
Note: Median wage–adjusted values are half of the median real wages (in 2014 dollars) for each state in 2012 for full-time, non-self-employed workers using the March Supplement of the
Current Population Survey. Regional price parity–adjusted wages use the Bureau of Economic Analysis regional price parity index for each state.
using the maximum of the state or federal minimum wage for Maryland, Massachusetts, New Hampshire, and Virginia)
each state.) Some states (e.g., California, Nevada, Oregon, and would require substantial increases, exceeding 50 percent.
Vermont) would need only small adjustments to their baseline When implementing as substantial an increase as in this latter
policy (under 10 percent). In contrast, higher-wage states (e.g., group of states, a longer phase-in period may be desirable.
140 Policies to Address Poverty in America
Arindrajit Dube
While the median wage is a good measure of how binding CITY-LEVEL POLICIES
a minimum wage would be, an additional consideration is While state-level minimum wages have been the most
cost of living, which tends to be greater in urban areas. To common means of allowing for regional variation, city-
provide an alternative adjustment, table 13-1 also reports level policies have become increasingly important in policy
the level of minimum wage that would prevail in a state if discussions. Since major metropolitan areas tend to have both
a $9.75 federal minimum wage—chosen because that is half higher wages and higher costs of living, minimum wage add-
the median full-time wage nationally—were adjusted using ons may make sense for large cities.
the regional price parity index for that state. To make this
an apples-to-apples comparison, both methods entail a Table 13-2 considers the twenty largest metropolitan areas in
similar overall increase in the minimum wage, letting the the country. Similarly to the state-level policies, I construct
exact pattern vary across states based on the median wage, both a median wage–adjusted and a regional price parity–
as opposed to just on the cost of living. adjusted level of the minimum wage for each of these areas.
There is considerable similarity in the target minimum wage As table 13-2 reports, DC, San Francisco, Boston, New York,
constructed using the two methods. This is to be expected and Seattle are high-wage metropolitan areas where half of the
since high-wage states also tend to have higher costs of living. 2012 full-time median wage was at least as large as $11.85 per
Nine states show up in both top ten lists, for example, and for hour in 2012 (in 2014 dollars). In another eight metropolitan
all but five states, the two methods produce a target minimum areas, half the full-time median wage exceeded $10.00 per hour.
wage that differs by less than 10 percent. These metropolitan areas represent a second tier of possible
laboratories for experimenting with local supplements. Some
The overlap is imperfect, however. For example, whereas of these cities are in areas where local wage standards are
Massachusetts has the highest median wage of all states, it preempted, but others are free to pursue policies.
ranks sixth in terms of the cost of living. Similarly, California
ranks twelfth based on median wage, but third based on cost Washington, DC and San Francisco already have local
of living. More generally, while the recommended increase in minimum wages, and Seattle recently enacted a city-wide
the minimum wage is similar under the two approaches when minimum wage policy. New York is actively exploring
averaged across all states (i.e., 26.2 percent versus 22.5 percent possibilities. The San Francisco experience has been studied
average increase in the statutory minimum wage), the regional and documented extensively (Dube, Naidu, and Reich 2007,
price adjustment produces a narrower range: between $8.42 2014). That city currently requires a minimum wage of $10.55
and $11.43 instead of between $7.97 and $12.45. per hour for all workers within city limits and this new
minimum wage has raised pay in the bottom of the distribution.
Under my proposal, state policymakers should put the greatest Yet employment growth does not appear to have been adversely
emphasis on how binding the minimum wage would be as affected in that city relative to its surrounding areas, even in a
proxied by half the median wage. This is an important metric high-impact sector like restaurants. Furthermore, Reich, Jacobs,
for gauging the extent of an intervention in the functioning and Dietz (2014) review the literature on four city minimum
of the labor market. Often this will also reflect cost-of-living wage standards, and find that they were implemented without
differences across areas. When the regional price parity– evidence of adverse effects.
adjusted minimum wage differs considerably from the median
wage–adjusted value, however, policymakers would do well A final consideration for local wage setting is regional
to also consider the regional price information—perhaps coordination. Although existing evidence does not indicate
splitting the difference between the two approaches. substantial movements of businesses across policy borders
to avoid a higher minimum wage, such movements may be
Finally, my proposal would index the state minimum wages more likely at higher levels of the minimum wage. Regional
to the regional CPI. This practice is attractive since the annual coordination in wage setting across economically connected
adjustment makes the process predictable and also responsive areas can reduce these risks.
to local conditions. Importantly, it eliminates the need for
revisiting a contentious policy issue year after year. As it One possibility is a regional collaboration in wage setting, as
stands, twelve states already have indexed their minimum exemplified in the Washington, DC metropolitan area. DC,
wages, paving the way for more to do the same. A few states, Prince George’s County (Maryland), and Montgomery County
including Nevada and Oregon, have adopted practices that (Maryland) coordinated on a simultaneous minimum wage
are very close to my recommendations: they have set the increase, though the extent of the increase varied by overall
minimum wage close to half the median wage, and have also wage levels. Similarly, in the San Francisco Bay area, the cities
indexed their wage to the CPI. of San Francisco and San Jose have both instituted citywide
The Hamilton Project • Brookings 141
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 13: Designing Thoughtful Minimum Wage Policy at the State and Local Levels
TABLE 13-2.
Target Minimum Wage by Metropolitan Area, Adjusted Based on Median Wage and Regional
Price Parity
Metropolitan Area Median Wage– Regional Price Population
Adjusted Parity–Adjusted (in millions)
(in dollars) (in dollars)
Washington, DC–Arlington–Alexandria, DC–VA–MD–WV 13.51 11.73 5.64
San Francisco–Oakland–Hayward, CA 13.37 11.81 4.34
Boston–Cambridge–Newton, MA–NH 12.85 10.87 4.55
New York–Newark–Jersey City, NY–NJ–PA 12.25 11.90 19.57
Seattle–Tacoma–Bellevue, WA 11.85 10.42 3.44
Baltimore–Columbia–Towson, MD 11.66 10.66 2.71
Philadelphia–Camden–Wilmington, PA–NJ–DE–MD 11.59 10.62 5.97
Minneapolis–St. Paul–Bloomington, MN–WI 11.23 10.03 3.35
Chicago–Naperville–Elgin, IL–IN–WI 10.79 10.38 9.46
Detroit–Warren–Dearborn, MI 10.42 9.53 4.30
San Diego–Carlsbad, CA 10.36 11.59 3.10
Los Angeles–Long Beach–Anaheim, CA 10.24 11.51 12.83
St. Louis, MO–IL 10.11 8.66 2.79
Atlanta–Sandy Springs–Roswell, GA 9.85 9.31 5.29
Riverside–San Bernardino–Ontario, CA 9.62 10.35 4.22
Dallas–Fort Worth–Arlington, TX 9.59 9.84 6.43
Houston–The Woodlands–Sugar Land, TX 9.50 9.81 5.92
Phoenix–Mesa–Scottsdale, AZ 9.39 9.71 4.19
Tampa–St. Petersburg–Clearwater, FL 9.07 9.68 2.78
Miami–Fort Lauderdale–West Palm Beach, FL 8.55 10.23 5.56
Sources: Ruggles et al. 2010; Bureau of Economic Analysis n.d.; author’s calculations.
Note: Median wage–adjusted values are half of the median real wages (in 2014 dollars) for each metropolitan area in 2010–2012 for full-time, non-self-employed workers using American
Community Survey data. Regional price parity–adjusted wages use the Bureau of Economic Analysis regional price parity index for each metropolitan area.
wages; Oakland, Berkeley, and Richmond are currently higher wages and lower poverty. The costs, such as negative
considering following suit. This type of policy coordination employment effects, are expected to be minimal.
makes both economic and political sense because it reduces
cross-jurisdictional competition and the possibility of Impact on Wages
business relocations. Under my proposal, the average minimum wage in 2016
across fifty states would rise from $7.71 per hour to $9.73 per
COSTS AND BENEFITS
hour in 2014 dollars—a 26.2 percent increase (see table 13-3).
The framework for reforming state and local minimum wages An increase in the binding minimum wage would benefit a
would have various positive economic benefits, including substantial number of workers: those whose wages would be
142 Policies to Address Poverty in America
Arindrajit Dube
TABLE 13-3.
Impact on Poverty by 2016 of Raising State Minimum Wages to Half of the State Median Wage
Estimate
Low Preferred High
Baseline statutory minimum wage (in dollars) 7.71 7.71 7.71
Statutory minimum wage under proposal (in dollars) 9.73 9.73 9.73
Change in statutory minimum wage (in percent) 26.2 26.2 26.2
Baseline nonelderly poverty rate (in percent) 15.8 15.8 15.8
Nonelderly poverty rate under proposal (in percent) 15.4 15.0 14.6
Change in poverty rate (in percentage points) –0.4 –0.8 –1.2
Change in population living in poverty (in thousands) –1,061 –2,238 –3,366
Source: Dube 2014.
Note: All dollar figures are in 2014 dollars. The statutory minimum wage in this table refers to the population-weighted average minimum wage over all fifty states using the maximum of the
state or federal minimum wage for each state. The details of the calculations are available at www.arindube.com/THP_projections.pdf.
directly raised by a higher wage floor, and those whose wages Impact on Employment
would rise through a ripple effect extending beyond the new A concern with raising the minimum wage is that businesses
wage floor by around 50 percent of the wage increase. For will respond by cutting back on hiring, thereby reducing jobs.
example, if a state raised its minimum wage by $2.00 from My review of the academic evidence suggests that this impact
$7.25 per hour to $9.25 per hour, workers earning up to $10.25 will likely be small.
per hour—$1.00 above the new minimum, or 50 percent of the
wage increase—would see their wages rise. In the 1990s, groundbreaking work by Card and Krueger
(1994, 2000) built a case-study approach to studying minimum
Rises in the minimum wage would affect many workers who wages. These authors relied on comparing adjacent states like
are not dependents of older, higher-paid workers. Estimates
New Jersey and Pennsylvania when one state increased the
of a raise in the federal minimum wage to $10.10 per hour
minimum wage. In the past decade, the Card and Krueger
indicate that the average age of the impacted worker would be approach has been generalized and refined. Dube, Lester, and
thirty-five, and that the majority (51 percent) of those impacted Reich (2010) considered all adjacent counties straddling state
by a wage increase would be aged thirty or older, while only borders for which data were available continuously for the full
13 percent would be aged twenty or younger (Cooper 2013). period between 1990 and 2006, and found no evidence of job
More than half (55 percent) of those affected by a federal losses for high-impact sectors such as restaurants and retail.
increase would be women, and about the same number (54 In follow-up work, Dube, Lester, and Reich (2013) used the
percent) would be full-time workers. While only 19 percent same cross-border methodology to study the effect on teens
of all workers have family incomes less than twice the official and found no discernible impact on their employment; Dube
poverty line, 50 percent of workers affected by a minimum and Zipperer (2014) confirm these findings using a “synthetic
wage increase would be in such families (CBO 2013). These control group approach,” which is a recent innovation in
trends at the federal level would likely persist at the state and empirical labor economics. Other researchers have obtained
local levels as well. In sum, the evidence strongly contradicts similar results. Addison, Blackburn, and Cotti (2009,
the suggestion that the typical affected worker is a teenager 2012) found that once they accounted for trends in sectoral
working for pocket money. While the minimum wage does employment, there was no evidence of job loss in the retail
not explicitly target individuals from families with very low or restaurant sectors; recent work by Hoffman (2014) finds no
incomes, most of the gains from the policy will accrue to those evidence of teen job losses using state-level case studies during
with low and moderate incomes. the 2000s.
The Hamilton Project • Brookings 143
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 13: Designing Thoughtful Minimum Wage Policy at the State and Local Levels
To be sure, some studies in the literature do suggest more- how the poverty rate responds to a higher minimum wage.
sizable job losses. These include estimates using the state- These estimates, along with state-by-state projected increases
panel approach pioneered by Neumark and Wascher (1992), as in the minimum wage, suggest that the poverty rate among
recently discussed in Neumark, Salas, and Wascher (2013). My the nonelderly would fall by anywhere between 0.4 and 1.2
own view is that this approach is less empirically compelling percentage points, representing between 1.1 and 3.4 million
than the cross-border methodology and other more- fewer individuals in poverty. The best estimate suggests that
sophisticated ways of constructing comparison groups that I the national nonelderly poverty rate would decline from 15.8
have used in my own work, as described above and discussed percent to 15.0 percent, and 2.2 million fewer people would
in Allegretto and colleagues (2013). Overall, I believe the live in poverty.
best evidence concludes that the net impact of the proposed
increase in the real statutory minimum wage would be likely
small, and likely too small to be meaningfully different from Questions and Concerns
zero. In addition, there is growing evidence that increased
What about the federal minimum wage?
minimum wages reduce job turnover (see Brochu and Green
2013 and Dube, Lester, and Reich 2013). This finding is largely The federal minimum wage plays an important role in
driven by a reduction in vacancies that result from fewer setting a nationwide standard. However, a one-size-fits-all
workers leaving jobs and the easier recruitment of workers approach creates avoidable trade-offs: states as dissimilar
into higher-paying jobs. as Massachusetts and Mississippi have different capacities
to absorb a minimum wage of, say, $11.00 per hour, and a
Impact on Poverty
single minimum wage has to balance the needs of states at
Minimum wage policies tend to increase incomes of low- and both ends of the spectrum. By allowing some variation across
moderate-income families. However, the antipoverty aspect of states, we can raise, say, the Massachusetts minimum wage
the minimum wage is limited because many families under to a reasonably high level while not putting, say, Mississippi
the poverty line do not have substantial attachment to the at risk. Leaving minimum wage setting altogether to states,
labor force. A review of past research finds that, on average, a however, will mean that patterns will reflect the vagaries
10 percent increase in the statutory minimum wage leads to a of politics across fifty states. For example, in spite of the
1.5 percent reduction in the number of individuals in poverty popularity among voters of raising the minimum wage,
(Dube 2014). state legislatures do not do so in a regular fashion, and many
states have implemented such policies only via costly ballot
My own analysis uses more and more-recent data, along
initiatives. Therefore, the lack of a federal standard can subject
with a wider range of statistical techniques than the existing
low-wage workers in many states to a substantial amount of
studies, and finds that a 10 percent increase in the minimum
risk. A moderate level of federal minimum wage, coupled with
wage would reduce the poverty rate among the nonelderly
state-level add-ons, offers a judicious balance.
population by between 1.2 and 3.7 percent, with the best
estimate suggesting a reduction of 2.4 percent (Dube 2014). Are there more-efficient or generally better ways to alleviate
In particular, robust evidence shows that an increase in the poverty?
minimum wage raises family incomes for the bottom 20
percent of the family income distribution. Strong evidence also Increases in the minimum wage have been shown to
finds that not just the incidence of poverty but also the depth substantially aid low-income families; most of the gains from
of poverty would be reduced, as measured by the poverty gap. the policy accrue to low- and moderate-income families.
At the same time, it is also true that the policy specifically
Overall, the evidence suggests that the poverty reduction targets low-wage workers and not individuals in poverty.
effects are somewhat larger in magnitude for African- Were we to assess public policies based only on their efficacy
American or Hispanic individuals, and for children under in reducing poverty, we should prefer more-targeted policies
age eighteen. The effects are somewhat smaller for single like cash transfers, SNAP, and programs that raise the
mothers and for younger adults. However, the impacts are employment rate for highly disadvantaged groups. The EITC,
larger in magnitude for young adults with no more than a in particular, is well-targeted at those with very low incomes.
high school diploma. It is important to point out, however, that as currently
structured, the EITC provides only minimal assistance to
As mentioned above, the statutory minimum wage averaged
adults without children, and may hurt some childless adults
over all fifty states would rise 26.2 percent by 2016 under
through a negative incidence on wages. Because the EITC
my proposal. Dube (2014) provides a range of estimates for
increases the labor supply, 27 cents of every dollar of EITC
144 Policies to Address Poverty in America
Arindrajit Dube
spending accrue to employers as lower wages (Rothstein 2010;
Lee and Saez 2012). Moreover, raising funds for the EITC
Conclusion
by taxing higher-income individuals also entails efficiency Minimum wage policies are not an antipoverty panacea. They
costs, which suggests an additional rationale for raising do, however, tend to raise wages for America’s lowest-paid
pretax earnings for low-wage workers (Hendren 2014). For workers—making an adequate minimum wage an important
these reasons, it makes sense to combine programs like the pillar of a national antipoverty agenda. Under my proposal, the
EITC with a minimum wage increase. poverty rate would likely decline by a little under 1 percentage
point, meaning that 2.2 million fewer individuals would live
Is there enough empirical evidence to support increasing the in poverty.
minimum wage to half the full-time median wage?
Setting the state and local minimum wages close to half the
The proposed increase of the minimum wage to half the full- median full-time wage is a well-balanced policy option. Such
time median wage does go somewhat above the range from a target is close to both U.S. experiences during the 1960s
which we can draw the best empirical evidence. This obstacle is and 1970s and to current practice in advanced industrialized
difficult to avoid given the rather low levels of minimum wages countries. While it pushes the minimum wage beyond the
since 1980. A number of additional factors make it reasonable experience over the recent period in this country, it does so in
to apply the existing estimates when evaluating this proposal, a measured fashion. In addition, states and localities should
however. First, an increase in the minimum wage from 41 consider the local cost of living when setting minimum wage
percent to 50 percent of the median full-time wage, while policy and should index wage levels for inflation. Incorporating
substantial, is still cautious. It maintains the ratio within both all of these criteria into minimum wage laws would lead to
historical and international bounds. Second, existing U.S. substantially higher wage floors in a subset of states: based on
evidence that suggests small employment effects is based on a half-median wage standard, fourteen states would have a
a number of states (e.g., Nevada, Oregon, Vermont) that have minimum exceeding $10.00 per hour, while based on cost-of-
all raised their state minimum wages to levels that surpass living considerations, ten states would do so.
46 percent of their median full-time wage. Finally, evidence
from the United Kingdom suggests that raising the minimum Possible negative impacts of a higher minimum wage can
wage close to the median full-time wage is not associated with be mitigated with regional wage coordination—localities
sizable effects on employment (Manning 2012). can cooperate to set adequate minimum wage policies. This
strategy, combined with minimum wage laws that set the
Would raising the minimum wage affect prices? wage floor based on local economic conditions, can lead
A higher minimum wage could lead to higher prices, especially to lower poverty, reduced inequality, and more-adequate
for industries that employ high levels of low-wage labor. To wages, all while mitigating the potential negative impacts
date, the clearest evidence on the effects on prices comes from on employment.
Aaronson, French, and MacDonald (2008), who find that a 10
percent minimum wage increase would raise fast-food prices
by around 0.7 percent. On average, my proposal would raise
fast-food prices by under 2 percent. While restaurant prices
will see likely increases from minimum wage increases, the
overall price level (e.g., the CPI) is unlikely to be noticeably
affected by minimum wage hikes.
The Hamilton Project • Brookings 145
146 Policies to Address Poverty in America
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 14: Smarter, Better, Faster: The Potential for
Predictive Analytics and Rapid-Cycle Evaluation to
Improve Program Development and Outcomes
Scott Cody
Mathematica Policy Research
Andrew Asher
Mathematica Policy Research
Introduction Rapid-cycle evaluation, another decision-support approach,
uses evaluation research methods to quickly determine
Public administrators have always been interested in whether an intervention is effective, and enables program
identifying cost-effective strategies for managing their administrators to continuously improve their programs by
programs. As government agencies invest in data warehouses experimenting with different interventions. Like predictive
and business intelligence capabilities, it becomes feasible to analytics, rapid-cycle evaluation leverages the data available in
employ analytic techniques used more-commonly in the administrative records. It can be used to assess large program
private sector. Predictive analytics and rapid-cycle evaluation changes, such as providing clients with a new set of services, as
are analytical approaches that are used to do more than well as small program changes, such as rewording letters that
describe the current status of programs: in both the public encourage clients to take some action. This type of formative
and private sectors, these approaches provide decision makers evaluation can be contrasted with the summative program
with guidance on what to do next. evaluations familiar to many in the policy community.
Summative program evaluations often assess whether a
Predictive analytics refers to a broad range of methods used program has an impact by comparing program participants
to anticipate an outcome. For many types of government with nonparticipants. Rapid-cycle evaluation uses similar
programs, predictive analytics can be used to anticipate techniques, but does not examine the overall impact of the
how individuals will respond to interventions, including program. Instead, it assesses the impacts of changes to the
new services, targeted prompts to participants, and program by comparing some program participants (with the
even automated actions by transactional systems. With change) to other program participants (without the change).1
information from predictive analytics, administrators can For example, rapid-cycle evaluation can determine whether an
identify who is likely to benefit from an intervention and find employment training program can use text message prompts
ways to formulate better interventions. Predictive analytics to encourage more clients to successfully complete program
can also be embedded in agency operational systems to guide activities. In this way, rapid-cycle evaluation can identify
real-time decision making. For instance, predictive analytics incremental changes that make the program more effective
could be embedded in intake and eligibility determination for its clients, increasing the likelihood that a subsequent
systems, prompting frontline workers to review suspect client summative evaluation would identify large impacts relative to
applications more-closely to determine whether income or individuals not in the program.
assets may be understated or deductions underclaimed.
The Hamilton Project • Brookings 147
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 14: Smarter, Better, Faster: The Potential for Predictive Analytics and Rapid-Cycle Evaluation to Improve Program
Development and Outcomes
We believe that these techniques can be used to help Department of Health and Human Services n.d.). The Nevada
government programs—including social service programs Reemployment and Eligibility Assessment Program increased
serving low-income individuals—to improve program services employment among participants, but only modestly: 52 percent
while efficiently allocating limited resources. We believe that of program participants were employed, which is higher—but
the use of predictive modeling and rapid-cycle evaluation— not substantially higher—than the rate in the control group, in
both individually and together—holds significant promise to which 48 percent of participants were employed (Michaelides et
improve programs in an increasingly fast-paced policy and al. 2012). In short, even programs highlighted as success stories
political environment. have room for improvement. They could benefit more clients
and they could have a larger impact on the clients they benefit.
We propose that social service agencies take two actions.
First, agency departments with planning and oversight The administrators of these and other programs are constantly
responsibilities should encourage the staff of individual seeking ways to improve outcomes. Some administrators
programs to conduct a thorough needs assessment. This seek to match clients with the right services. But without the
assessment should identify where predictive analytics right analytic tools, these administrators cannot determine
and rapid-cycle evaluation can be used to improve service if their services are targeted as effectively as possible. Other
delivery and program management. The assessment should administrators seek to test new procedures aimed at improving
also evaluate whether the benefits of adopting these tools program services. But again, without the right analytic tools,
outweigh the costs, resulting in a recommendation of these administrators may get biased results, leading them to
whether and how these tools should be deployed. Second, implement ineffective changes or to dismiss effective ones. In
federal agencies should take broad steps to promote the use the end, progress toward program improvement is slow, and
of predictive analytics and rapid-cycle evaluation across programs end up spending resources inefficiently and leaving
multiple programs. These steps include investments in data participants underserved.
quality and data linkage, as well as measures to support and
promote innovation among agency staff.
A New Approach
The Challenge Because predictive analytics and rapid-cycle evaluation have
the potential to improve program effectiveness, we believe
Our proposal is based on the simple assumption that government that social service agencies should conduct thorough needs
programs could do better. This seems self-evident: despite assessments to identify, program by program, where these
decades of antipoverty efforts, the reality is that unemployment tools can be used. The needs assessments should examine the
and underemployment, low food security, high poverty rates, quality of existing program data to determine whether they are
and related problems persist. Rigorous evaluations of federal robust enough for use in predictive analytics and rapid-cycle
social programs show that many programs have little or even evaluation. The assessments should also examine whether and
no impact on program participants. how programs can deploy predicted outcomes operationally
in a way that improves program performance. Furthermore,
In fact, even those programs held up as examples of proven, they should assess whether and to what extent experiments
evidence-based programs demonstrate that government can be conducted to test changes in program operations. In
programs could do better. For example, the Coalition for addition to conducting program-level needs assessments,
Evidence-Based Policy identifies top-tier social programs agencies should also take steps to promote the use of these
with rigorous evidence of effectiveness, such as the Nurse- tools broadly across multiple programs. These steps could
Family Partnership, Nevada’s Reemployment and Eligibility include investments to improve data systems, improve data
Assessment Program, the Transitional Care Model, and governance, and promote a willingness among program staff
other programs (Coalition of Evidence-Based Policy 2012). to test program innovations.
Multiple randomized controlled trials on each of these
programs show positive impacts on client outcomes. But To inform the needs assessment, this section begins with
even this positive evidence suggests these programs could be an explanation of how predictive analytics and rapid-cycle
more effective. A systematic review of research on the Nurse evaluation can be deployed in the administration of public
Family Partnership program concludes that there is evidence programs. These tools are not commonly used at this time.
of a positive impact on only seven of the twenty-five measures Where possible, we provide real-world examples of the
of child maltreatment, and on only five of the fifty-nine application of these tools. We supplement these examples
measures of child development and school readiness (U.S. with a discussion of potential applications. Agencies should
148 Policies to Address Poverty in America
Scott Cody, Andrew Asher
consider these real-world and potential applications when patients are likely to be readmitted, they could intervene
conducting their needs assessments. to address some of the factors contributing to the higher
likelihood of a repeat visit. This would enable the patients to
PREDICTIVE ANALYTICS avoid another hospitalization while the Medicaid program
At the individual level, predictive analytics leverages the fact would avoid paying for expensive hospital care.
that key outcomes and outputs for program clients are often
Researchers at New York University have developed such a
correlated with the client’s prior behaviors, circumstances,
predictive model to identify a combination of characteristics
and characteristics, as well as those of the client’s family,
and circumstances that indicate an elevated risk that a New
associates, service providers, and surroundings. By examining
York Medicaid beneficiary discharged from a hospital will
these correlations, predictive analytics methods can be used to
return within one year (Raven 2009; Raven et al. 2009). New
rank program clients based on the likelihood that an outcome,
York City Health and Hospitals Corporation is using this
whether positive or negative, will occur.
model within its operational systems to screen admitted
For example, an analysis predicting which participants of a job patients and identify interventions for those most likely to be
training program are likely to find employment might leverage readmitted for a preventable reason (Evans 2011).3
existing information about the clients’ education levels and
their attendance at job training sessions. The model might A similar approach could be used to prevent recipients of
tap these factors and other information to rank participants public assistance benefits from letting their eligibility lapse.
Assistance programs such as the federal Supplemental
on the likelihood that they will find employment. Using these
Nutrition Assistance Program (SNAP), formerly known as the
rankings, program administrators could decide, based on
Food Stamp Program, require beneficiaries to demonstrate
their goals and resource constraints, the exact sub-population
eligibility through a periodic recertification process. If clients
that they want to target with their additional services.
do not complete the recertification process, their benefits are
Depending on their program’s objectives, administrators
terminated. Clients often do not reapply for the program until
might focus on individuals most likely to find employment,
they realize their benefits have been terminated. This creates
or might target additional services to individuals less likely to
two problems. First, clients who are eligible for assistance forgo
find employment.2
benefits for one or two months until they reapply. Second, the
Below we describe two key uses of predictive analytics for program must bear the costs of processing a new application—
policymakers: (1) identifying program participants at risk which is more expensive than recertification. State agencies
of an adverse event and (2) predicting the optimal service that administer the federal SNAP program could use predictive
path for an individual. We then discuss deploying predictive analytics to identify clients at risk of such churning. What
analytics to impact decision making. would be required, beyond the tested and validated analytics
themselves, is that the models be built directly into the case
IDENTIFYING PROGRAM PARTICIPANTS AT RISK OF maintenance systems. Identifying these at-risk clients prior to
AN ADVERSE EVENT the redetermination would enable program administrators to
Program administrators can use predictive analytics to direct targeted, intensive communication efforts to these clients
identify clients who are at risk of an adverse outcome such as to prevent churning and help the clients maintain benefits while
unemployment, fraud, unnecessary hospitalization, mortality, saving program funds.
or recidivism. Knowing which participants are most likely to
Other potential areas for using predictive analytics include
experience an adverse outcome, program staff can provide
enforcement and fraud detection applications. For example,
targeted interventions to reduce the likelihood that such
some child support enforcement agencies are developing
outcomes will occur.
predictive models to identify noncustodial parents who will
Reducing readmission rates for certain patients discharged not make their child support payments. This information
from the hospital provides an example of how predictive can be used to triage enforcement efforts, making sure fewer
analytics can be used effectively. Reasons for unplanned resources are devoted to collection efforts against those who
readmissions can include clinical and social factors, such as will ultimately pay without enforcement and identifying those
patients’ timely access to quality primary health-care services, who are likely to pay in response to more-aggressive efforts.
their underlying conditions, whether they are homeless, and
In addition, predictive analytics can be used to identify
whether they lack social support and other factors that affect
provider, client, vendor, and billing entity fraud patterns
their ability to recuperate at home without incident (Peikes
in health-care and social service programs. In SNAP, for
et al. 2012–13). If Medicaid programs could anticipate which
instance, geographic patterns of electronic benefits transfer
The Hamilton Project • Brookings 149
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 14: Smarter, Better, Faster: The Potential for Predictive Analytics and Rapid-Cycle Evaluation to Improve Program
Development and Outcomes
redemption and historical investigative data can be used to
predict which program clients and retailers may be engaged
in benefit trafficking (exchanging SNAP benefits for cash at
a discount).
PREDICTING OPTIMAL SERVICE PATHS
Many government programs have different approaches to
working with clients to achieve the same outcome. For example,
there are multiple approaches to preventing recidivism among
juvenile offenders, encouraging preventative health care, and
boosting the parenting skills of new mothers. These paths may
differ in the services involved or the time at which the services
are offered. Under the right circumstances, predictive analytics
can be used to determine which approaches are most likely
to benefit which clients. Administrators can then identify the
optimal service path for a client among the available options.
Consider a caseworker trying to find the right jobs program
for a nineteen-year-old unemployed man with no high school
diploma. This caseworker can enroll the individual in a low-
intensity résumé support and job search program, a more-
intensive program teaching specific manufacturing skills,
or even a very intensive apprenticeship program. Each path
has a different cost, and possibly a different outcome, for this
individual. The caseworker’s job is to match the program
to the individual’s background and interests. Combining
this information—which is readily known at intake—with
a prediction, based on which programs are associated with
success for similar clients, could yield a better match between
client and services, increasing the likelihood that the client
will find employment and reducing the likelihood of wasting
funds on ineffective training. Many agencies are interested in
developing optimal service path predictions, yet in practice
few exist. We believe there is an opportunity for optimal
service path modeling to benefit the clients of public programs.
RAPID-CYCLE EVALUATION
Rapid-cycle evaluation, another tool that supports decision-
making, is increasingly used in public programs with readily
available administrative data and the ability to analyze those
data in a rapid, cost-effective manner. This type of evaluation
uses rigorous experimentation to test changes in agency
operations.* To determine any impacts from the changes,
administrators can compare client outputs and outcomes with
those for other clients who are included in the evaluation but
continue to receive regular services. The evidence from these
tests can be more reliable than other sources, such as feedback
from staff, complaints from selected clients, or anecdotes from
other agencies.
To better understand how rapid-cycle evaluations can be used
to test changes, it is useful to consider the three defining terms:
150 Policies to Address Poverty in America
1. Rapid. The “rapid” means that the impact of the intervention
will be identified quickly. To facilitate rapid identification
of results, the outcomes of interest should be observable in
administrative data. This eliminates the time-consuming
process of collecting new data. Additionally, any impacts of
the intervention should be observable within a short time
frame. For example, it would not be possible to rapidly assess
whether an intervention delivered to ninth-grade students
leads more of those students to graduate from high school.
2. Cycle. The “cycle” refers to the iterative nature of the
tests. Rapid-cycle evaluations can support a formative,
continuous improvement model in which an intervention
is tested, the results are examined, the intervention is
modified if needed, and the modified intervention is tested
again or a new intervention is tested.
3. Evaluation. The “evaluation” refers to the use of rigorous
research techniques that generate confidence that observed
changes in outcomes are due to the intervention and not to
other factors (such as differences between the group that
received the intervention and the group that did not).
This approach has been used by businesses for years to
continuously improve the match between customers and
services. For example, Capital One claims it runs more than
30,000 experiments each year to help identify the techniques
that cause customers to sign up for new credit cards as well as
techniques that encourage customers to pay Capital One back
(Davenport and Harris 2007). The company experiments with
changes in interest rates, promotional incentives, and even the
color of the envelopes used in customer mailings.
Rapid experiments are used in the public sector as well to test
a variety of program interventions, including changes in staff
procedures, the services provided to clients or customers,
and when and where those services are provided. Rapid-
cycle evaluations of experiments can assess whether the
interventions meet goals such as improving (1) the agency’s
ability to serve more clients, (2) the quality of information
agencies get from clients, (3) client outcomes, and (4) agency
efficiency. It is sometimes possible to test numerous variations
of program services simultaneously. Box 14-1 shows how
experimentation and rapid-cycle evaluation can fit into overall
program operations by presenting applications used by New
York City Human Resources Administration.
In some cases it may not be feasible to collect the necessary
outcome data. For example, target outcomes may occur
too far in the future to be examined in a rapid experiment
(e.g., the eventual graduation of ninth graders). It may still
be feasible, however, to employ rapid-cycle evaluation by
looking at impacts on intermediate outcomes (such as class
Scott Cody, Andrew Asher
BOX 14-1.
New York City Human Resources Administration
Agencies such as the New York City Human Resources Administration (HRA) have deployed rapid-cycle evaluation to
improve program services. HRA recently tested new administrative procedures to increase the establishment of child
support for children receiving cash assistance (Dinan 2013). Since court appearances are assumed to be a deterrent to
establishing child support, the agency tested new procedures that would avoid a court appearance for the noncustodial
parent. The agency’s goal was to increase the percentage of these children with established child support orders, reduce the
time needed to establish orders, and increase the proportion of noncustodial parents that comply with their established
orders. HRA staff developed a simple random assignment process for determining which cases were eligible for the
new procedures, and trained frontline workers to administer the pilot. The analysis showed the new procedures were
unsuccessful. The rate of child support order establishment for the treatment group (57.3 percent) was essentially the same
as the rate for the control group (56.5 percent). Moreover, it took longer to establish those child support orders established
through the new procedures.
HRA also used experimentation to test streamlined procedures for investigating Medicaid eligibility fraud (Weinberg 2013).
These streamlined procedures were designed to reduce the number of steps needed to investigate potentially fraudulent
Medicaid enrollees. They used a four-month random assignment experiment to evaluate the impact of these new procedures.
Although the new procedures reduced the time spent investigating cases by 12 percent, fraud investigations conducted with
the streamlined procedures were less likely to be successful. Fraud was established for 44 percent of cases investigated through
streamlined procedures, compared with 61 percent of those investigated through status quo procedures.
In both of these experiments HRA’s new procedures proved unable to achieve the desired impact. But in each case HRA
clearly and quickly established that the procedures were unsuccessful without having to implement these new procedures
program-wide.
attendance and grades), as well as program outputs (such The greatest benefit of rapid-cycle evaluations to the agencies is
as the amount and quality of services provided). Such rapid the rigorous nature of the evaluation, which can replace other,
experiments and rapid-cycle evaluations can often still help nonexperimental techniques for assessing programmatic
improve program services. changes. For example, programs may pilot new procedures
with all staff in a single location. In such cases, it is often not
Rapid-cycle evaluation also could be used to measure possible to know whether differences in outcomes are caused
real responses to potential policy changes. For example, by the new procedures or simply by the unique circumstances
programs like SNAP and Temporary Assistance for Needy of that location. This can lead program administrators to
Families (TANF) have numerous eligibility criteria and other the false conclusion that a new procedure has promise, only
regulations that are often debated by policymakers. These to learn there is no benefit once it is implemented agency
include deduction amounts, certification period lengths, wide. Alternatively, it can lead them to reject a procedure that
benefit formulas, reporting thresholds for income changes, actually has promise.
and even the required number of hours for participation
in work programs. When the changes to these regulations COMBINING PREDICTIVE ANALYTICS AND RAPID-
are discussed, policymakers debate whether these changes CYCLE EVALUATIONS
will lead to higher or lower participation rates, and whether
Predictive analytics and rapid-cycle evaluations can be
they will lead to longer program dependence or encourage
combined to help program administrators build better
employment. Rapid-cycle evaluation has the potential to
interventions. Predictive analytics allow administrators to
generate rigorous, reliable information that can take the
anticipate which individuals are most (and least) likely to
guesswork out of these policy debates. Regulatory changes
benefit from a program. These predictions can help program
can be tested to identify—and quantify—clients’ behavioral
administrators guide the formulation and scope of the
response to these changes. This information can ensure that
interventions, and determine the group or subgroups to which
regulatory changes better meet policymakers’ goals.
they would apply. By creating targeted experiments, program
The Hamilton Project • Brookings 151
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 14: Smarter, Better, Faster: The Potential for Predictive Analytics and Rapid-Cycle Evaluation to Improve Program
Development and Outcomes
FIGURE 14-1.
Combining Predictive Analytics and Rapid-Cycle Evaluation: A Simplified Example
Program
Services
1 Predictive analytics
are used to sort Program Participants
program participants
by likelihood of
benefit from the
base program. Benefit Unlikely Benefit Likely
2 A new intervention is 3 If effective, the new
developed and tested on Control Treatment treatment can be
participants unlikely to incorporated into
benefit from the base the program and the
program. cycle can repeat.
administrators can identify a series of effective, tailored hospital within one year. If program administrators want to
interventions to maximize their ability to make an impact.5 test two different interventions for these at-risk patients, they
could randomly assign the at-risk patients to one of three
Consider a program administrator seeking to test new groups—one for each of the two interventions plus a control
approaches for reaching hard-to-serve clients. Initial group—that receive the hospital’s normal discharge planning
predictive models could identify which of the program’s and other services. The team would then monitor hospital
current clients are least likely to benefit from the program, and admission rates for three months and assess whether the new
could separate the clients into treatment and control groups. interventions cause a significantly lower readmission rate.
New interventions (or potentially multiple variations of the Any successful intervention could be integrated into program
same intervention) could be tested rapidly and, if effective, operations; the unsuccessful ones could be discarded.
could be incorporated into service delivery. After the new
interventions operate for sufficient time, the entire cycle could If multiple interventions prove successful, program
be repeated (see figure 14-1) or applied to a different subgroup administrators could implement all of them or choose one
of program participants. based on cost and potential sustainability. The predictive
model could be rerun and follow-up analysis could suggest new,
The new procedures can also be tested on individuals who tailored interventions for the remaining at-risk population.
are likely to benefit from the program. Such tests can help These interventions could be formulated and tested as in the
administrators determine whether new approaches would previous cycle, evaluated, and either discarded or included in
yield even greater improvements for individuals positioned to program operations.6
benefit the most from program services. Some administrators
may view targeting those most likely to benefit as the most THE POLICY PROPOSAL
effective way to achieve gains for participants and improve the We propose that federal social service agencies take two
program’s overall success. actions. First, agency departments with planning and
For illustration, consider the hospital readmission prediction oversight responsibilities should encourage the staff of
model mentioned earlier. A predictive model could be used individual programs to conduct thorough needs assessment.
to identify at-risk patients who are most likely to return to a This assessment should identify where predictive analytics
152 Policies to Address Poverty in America
and rapid-cycle evaluation can be used to improve service
delivery and program management.
For predictive analytics, program administrators should
assess:
e Whether predictions about specific client and program
outcomes could be employed to target program services;
¢ Whether the program’s current administrative data contain
accurate, valid, and reliable measures of those outcomes—
as well as valid and reliable measures of information that
could predict those outcomes—to support predictive
modeling; and
e The magnitude of systems enhancement efforts required
to enable frontline workers to use the results of predictive
models in real-time when they interact with clients.
For rapid-cycle evaluation, program administrators should
assess:
e Whether program changes under consideration would
benefit from precise, causally-valid impact estimates
generated through rapid-cycle evaluation. The assessment
can rely not only on program staff, but also on funders
and outside experts to identify program features that they
believe would be beneficial to test but were not sure should
be implemented permanently without assessment;
e Whether program operations can be modified to facilitate
experimentation of these program changes;
¢ What types of investments in data and systems would be
required to deploy predictive analytics and rapid-cycle
evaluation together as an integrated strategy;
« What types of programmatic waivers and other policy
changes would be needed to facilitate predictive analytics
and rapid-cycle evaluation; and
e Whether it would be beneficial to use predictive analytics
to subset the program population, and to test program
changes on different types of individuals (e.g., those most
likely to benefit from current services).
The answers to each of these questions will vary by program.
The assessment also should evaluate whether the benefits of
adopting these tools outweigh the associated costs. In the end,
the assessment should contain a recommendation of whether
and how these tools should be deployed.
The second step agencies should take is to promote the
adoption of predictive analytics and rapid-cycle evaluation
Scott Cody, Andrew Asher
more broadly across programs. We recommend that agencies
take the following steps:
1. Help programs make individual-level data available for
analytics. Individual-level data provide the best foundation
for predictive analytics and rapid-cycle evaluation. These
data can be obtained through internal operational systems
maintained by the program, or through integrated data
systems that combine administrative data across programs.
A broad investment in data can facilitate predictive analytics
and rapid-cycle evaluation and can promote the use of these
tools across multiple programs.
Federal agencies can help more programs benefit from these
analytic tools by facilitating improvements to individual-
level administrative data, and ensuring that those data are
available for analytic purposes. Some agencies are already
taking the lead in this respect. For instance, the Department
of Education provided grants to promote the development of
statewide longitudinal education data, and the Centers for
Medicaid and Medicare Services funded data warehouses
to help states manage all aspects of their Medicaid and
Children’s Health Insurance Programs. Agencies can
also use their expertise to help programs identify the key
measures to track for prediction and evaluation on an
ongoing basis.
2.Improve data governance and facilitate data sharing.
Although high-quality data are necessary,
also need strong data governance policies that establish
accountability for data quality and that define the terms for
how and where data are used (see Digital Services Advisory
Group 2012). In addition, as part of data governance efforts,
agencies should work to actively support efforts to link
data across programs, which involves often-challenging
technical and legal considerations. That said, linked data
can provide a more comprehensive understanding of the
services received and circumstances faced by clients, and
provide more-accurate predictions and a more-complete
understanding of the impact of rapid-cycle experiments.
agencies
3.Encourage analytic decision making. The use of
predictive analytics and rapid-cycle evaluation requires
an organizational commitment to testing program
improvements. This means agency staff must develop
program be willing to abandon
those innovations if they prove unsuccessful. For many
program staff, this is a change in mindset from a focus
on assessment of their program (and compliance with
funder guidelines needed to properly evaluate their
programs) to a focus on how to improve the programs
and empower program administrators. Federal agencies
can help foster innovation by providing performance-
innovations—but
The Hamilton Project ¢ Brookings 153
IMPROVING SAFETY NET AND WORK SUPPORT
Proposal 14: Smarter, Better, Faster: The Potential for Predictive Analytics and Rapid-Cycle Evaluation to Improve Program
Development and Outcomes
based funding opportunities for program improvements. Agencies should compare these projected costs with the
potential benefits obtained from these tools. In many cases,
Predictive analytics and rapid-cycle evaluation can be the benefits will include long-term savings in program
effective in part because they empower frontline staff to administrative costs because the tools render the program
determine the services that best meet their clients’ needs. more efficient. Other important benefits, however, such
However, the lessons learned for individual programs as improvements to the quality, availability, and access to
can be valuable to other programs serving these same services, should also be considered.
populations. Agencies can further the effectiveness of these
tools by ensuring successful efforts are highlighted, and In the end, we believe that the benefits of predictive analytics
their lessons broadly disseminated. and rapid-cycle evaluation will be substantial for many
programs. We believe that this potential may warrant
We believe that by taking these steps, federal agencies can help significant investment in these tools for many programs. For
promote the use of these analytic tools at the federal, state, and virtually all programs, however, we believe that this potential
local levels. warrants the costs of conducting a needs assessment.
COSTS AND BENEFITS
Predictive analytics and rapid-cycle evaluation have a Questions and Concerns
number of benefits. In particular, greater use of these tools
In this section we examine some of the factors that could
would increase program effectiveness by reducing wasteful
affect an agency’s ability to adopt predictive analytics and
and inefficient spending. Even where the proposal results
rapid-cycle evaluation by posing and addressing a series of key
in an increase in direct outlays in one phase of a program’s
questions and concerns.
intervention, these outlays may generate net savings.
Moreover, these analytical innovations would allow programs What data resources are necessary for the use of these tools?
to fulfill their missions more effectively by better targeting
their intended beneficiaries and helping them continually Programs with advanced information systems that contain
identify and implement cost-effective interventions. individual-level administrative data are better suited to deploy
predictive analytics and rapid-cycle evaluation with minimal
That said, adopting these tools can require significant investment. For example, in many states sophisticated cross-
investments at a time when government budgets are under program data warehouses have been developed to support a
pressure. Developing the data and technology infrastructure wide array of Medicaid and social service program monitoring
necessary to deploy these analytical capabilities—if they needs. These systems are rapidly updated and could be easily
are not already present—is expensive, as are, to a lesser used for both predictive analytics and rapid-cycle evaluation.
extent, the resources needed to perform these analytics. For For other programs, administrative data obtained from
example, what may be considered to be the gold standard transactional systems can provide an important source
for data infrastructure—a full-featured, enterprise-wide of information. These programs would require additional
data warehouse that integrates data across programs and is investment to create analytics-ready data repositories through
refreshed on a weekly basis—can cost several million dollars the extraction, transformation, and storage of the data.
to build, and millions more annually to staff with dedicated
maintenance and analytical personnel. Less-expensive data It is important to note that predictive analytics require historic
systems, such as purpose-specific analytical datamarts within observations of key outcomes. This means that programs
existing warehouses or standalone databases focused on developing new systems may not be able to perform predictive
specific questions, may be more feasible and more appropriate analytics until the system has captured enough history.
in some cases. Similarly, programs extracting data from transactional
systems would need to extract a sufficiently large volume of
As part of their needs assessments, agencies should assess the historical data in order for predictive analytics to be effective.
costs of any changes needed to deploy predictive analytics and
rapid-cycle evaluation. In addition to data infrastructure costs, Who would implement these tools?
agencies should examine the costs associated with training
staff, as well as the costs of altering program operations to The program managers and staff in agencies directly
incorporate predictive analytics and to implement rapid-cycle responsible for program delivery would implement these
experiments and evaluations. tools. To be successful, the implementation of these tools
requires a division of labor. Program administration staff—
both program operators and those working in support of
154 Policies to Address Poverty in America
Scott Cody, Andrew Asher
them at federal and state agencies—need to determine which Second, it is important to ensure model predictions are
interventions are worth implementing and figure out how to do followed up by human judgment. Whether it is identifying
so. These program experts need to be supported by analytical clients who should receive a caseworker visit or those who
specialists who are charged with designing the predictive may be defrauding the government, predictive analytics
analytics and assessing the results of the experiments through should be used to prioritize cases; staff should make the final
the rapid-cycle evaluations. Such a partnership allows this determination. Similarly, even after optimal service paths
approach to become feasible and avoid burdening those with are predicted, clients should still have a say in the services
the pressing responsibility of running programs. they receive.
Can predictive analytics be wrong?
Yes. Predictive models detect patterns, but not every individual Conclusion
will follow that pattern. This can lead to incorrect predictions. As integrated data repositories become common in
Administrators can take several steps to minimize problems government agencies, program administrators have become
stemming from inaccuracies in predictive models. comfortable using these data to monitor their programs. Now
administrators are poised to expand the use of analytics to
First, predictive models should be subjected to extensive
better decide what to do next. Predictive analytics and rapid-
validation. For adverse event situations, such as hospital
cycle evaluation, if used individually but especially if used
readmissions or fraud, models should be deployed historically
together, can help agencies provide services where they are
so that their ability to predict known outcomes can be assessed.
needed and develop more-effective approaches for improving
Through repeated retrospective testing, use and learning, the
program outcomes.
models can be improved, often to the point where they can be
used prospectively.
The Hamilton Project • Brookings 155
156 Policies to Address Poverty in America
Authors
Katharine G. Abraham Andrew Asher
Professor, Department of Economics and Joint Program in Senior Fellow, Health Policy Division, Mathematica Policy
Survey Methodology, University of Maryland Research
Katharine Abraham is a professor in the department of Andrew Asher is a senior fellow at Mathematica and with
economics and in the Joint Program in Survey Methodology extensive experience in health care management and data
at the University of Maryland. Abraham first joined the analytics—the use of data to help agencies and health care payers
University of Maryland faculty in 1987 after holding prior strengthen the integrity of their programs, realize cost savings,
positions at Massachusetts Institute of Technology’s Sloan and improve program outcomes. He is one of the leaders of
School of Management and the Brookings Institution. She Mathematica’s data analytics initiative and his expertise includes
served two terms as Commissioner of the Bureau of Labor program integrity, health policy, health program development,
Statistics and, more recently, was a member of the President’s big data, and developing strategies to help agencies improve
Council of Economic Advisers. Among other honors and their performance. He is nationally recognized for developing
awards, she is an elected fellow of the American Statistical and implementing innovative data-driven approaches to help
Association and of the Society of Labor Economists. She health care purchasers and insurers reduce fraud and abuse
received her doctorate in economics from Harvard University. and manage their costs effectively. He rejoined Mathematica in
Abraham has studied and written about work-sharing 2013 after serving as vice president and director of government
policies, unemployment, and job openings; the operation solutions at OptumInsight from 2007 to 2013 and in various
of internal labor markets; the effects of financial aid on the leadership roles with the state of Illinois. From 2001 to 2007,
decision to attend college; the work and retirement decisions he served as the manager of the Fraud Science Team within
of older Americans; and the measurement of market and the Illinois Department of Healthcare and Family Services’
nonmarket economic activity. From 2002 to 2004, Abraham Office of Inspector General, a group whose innovative fraud
chaired a National Academy of Sciences panel on accounting detection and prevention work was featured in Government
for nonmarket activity; its report, Beyond the Market, was Accountability Office and Center for Medicare and Medicaid
published in 2005 by the National Academies Press. Services best practice reviews. Asher was a health researcher and
analyst at Mathematica from 1989 to 1994 and holds a master’s
in public policy from the University of Michigan.
Karen L. Anderson
Managing Director, The Hamilton Project
Jillian Berk
Karen Anderson is the managing director of The Hamilton Senior Researcher, Mathematica Policy Research
Project, a position she has held since 2006. From 2008 to
2009, she served as chief of staff at the President’s Council of Jillian Berk is a senior researcher and associate director
Economic Advisers. Prior to joining The Hamilton Project, of research at Mathematica Policy Research. Her research
Anderson was a vice president of State and Local Government examines employment and training programs for
Relations for Citigroup. She previously served in the White economically vulnerable populations, including dislocated
House Office of Political Affairs from 1993 to 1995. Anderson workers, older workers, workers with disabilities, and ex-
holds a master’s in public policy from the Kennedy School of offenders. Recent projects include the national evaluation
Government at Harvard University and a bachelor’s degree in of Trade Adjustment Assistance program, an evaluation of
economics from Agnes Scott College. the Virginia Community College System’s efforts to better
meet the education and training needs of adult students, an
evaluation of U.S. Department of Labor-funded youth offender
programs, and a study on the impact of the subsidization
of COBRA health insurance on the take-up of coverage and
other health and employment outcomes. She holds a doctorate
in economics from Brown University.
The Hamilton Project • Brookings 157
Authors
Elizabeth U. Cascio Arindrajit Dube
Associate Professor, Department of Economics, Dartmouth Associate Professor, Department of Economics, University of
College Massachusetts Amherst
Elizabeth Cascio is an associate professor of economics at Arindrajit Dube is an associate professor of economics at
Dartmouth College; a research associate in the programs the University of Massachusetts Amherst. His work focuses
on Education, Development of the American Economy, and on labor economics, health economics, public finance, and
Children at the National Bureau of Economic Research; and political economy. His core areas of research include minimum
a research fellow at the Institute for the Study of Labor. Her wage policies, fiscal policy, income inequality, health reform,
research is on the economics of education. A central goal of and the economics of conflict. Dube received his B.A. in
her research to date has been to understand the implications Economics and M.A. in Development Policy from Stanford
of a series of historic policy changes that began in the 1960s, University, and his Ph.D. in Economics from the University
including the large wave of state subsidization of early of Chicago. Prior to joining the University of Massachusetts
education and the introduction of progressive federal funding Amherst, he held a research economist position at the
for K–12 education. She has also studied early education Institute for Research on Labor and Employment at University
policy in the modern era. Her work has been funded by the of California, Berkeley. He is also currently a research fellow
Spencer Foundation and the National Science Foundation at IZA. During the spring semester of 2014, Dube visited
and has been published in the Quarterly Journal of Economics, the economics department at the Massachusetts Institute of
American Economic Journal: Economic Policy, The Journal Technology.
of Human Resources, and the Journal of Urban Economics,
among other outlets. She received her doctorate in economics
from the University of California, Berkeley, in 2003.
Benjamin H. Harris
Policy Director, The Hamilton Project
Scott Cody Benjamin H. Harris is the policy director of The Hamilton
Vice President and Director of Human Services Research, Project; a fellow in Economic Studies at the Brookings
Mathematica Policy Research Institution; and deputy director of the Retirement Security
Project. His primary areas of focus are tax, budget, and
Scott Cody is a vice president and director of human services retirement security. Harris has published a variety of papers
research in Mathematica’s Cambridge, Massachusetts office. He and policy briefs related to topics in public finance, and
is a national expert in the Supplemental Nutrition Assistance is regularly cited in media reports related to fiscal policy.
Program (SNAP). He led the first major in-depth study of state Previously, he was a senior research associate with the Urban
efforts to modernize enrollment procedures for programs like Institute and the Urban-Brookings Tax Policy Center, and a
SNAP, Temporary Assistance to Needy Families (TANF), and senior economist with the President’s Council of Economic
Medicaid. His research also has examined the determinants of Advisers. He has also served as a research economist at the
SNAP participation. He led an impact evaluation of efforts to Brookings Institution and as a senior economist with the
increase program participation among the elderly, and multiple Budget Committee in the U.S. House of Representatives.
studies to identify the circumstances that lead individuals to Harris has also taught as an adjunct professor at the policy
enter and exit SNAP. Cody played a central role in adapting schools at both the University of Maryland and Georgetown
Mathematica’s SNAP policy simulation model to reflect major University. He holds a Ph.D. in economics from George
changes in program policy. In addition to research on SNAP, Washington University, a master’s degree in economics
Cody is the deputy director of the What Works Clearinghouse from Cornell University, a master’s degree in quantitative
(WWC). The WWC is an initiative of the U.S. Department of methods from Columbia University, and a bachelor’s degree
Education’s Institute of Education Sciences. The WWC assesses in economics from Tufts University. In 2000, Harris was
the quality of thousands of studies of education curricula, awarded a Fulbright Scholarship to Namibia.
practices, and policies, and then summarizes the findings of
well-designed studies in reports for educators, policymakers,
and the general public. Cody holds a master’s in public policy
from the Johns Hopkins University.
158 Policies to Address Poverty in America
Harry J. Holzer research on outsourcing and offshoring also has highlighted
measurement problems in U.S. statistics. With funding from
Professor, McCourt School of Public Policy, Georgetown the Alfred P. Sloan Foundation and the Bureau of Economic
University Analysis, she has conducted research and organized
Harry Holzer is a professor at the McCourt School of Public conferences on biases in price indexes, productivity, and
Policy at Georgetown University. He is also currently an output growth, along with other measurement problems
institute fellow at the American Institutes for Research, a senior arising from the growth of globalization.
affiliate at the Urban Institute, a senior affiliate of the National
Poverty Center at the University of Michigan, a national fellow Hilary Hoynes
of the Program on Inequality and Social Policy at Harvard
University, and a research affiliate of the Institute for Research Professor of Public Policy and Economics, Haas Distinguished
on Poverty at the University of Wisconsin at Madison. He Chair in Economic Disparities, University of California,
has also been a faculty director of the Georgetown Center Berkeley
on Poverty, Inequality and Public Policy. He received his
Hilary Hoynes is a professor of public policy and economics
bachelor’s degree (1978) and doctorate (1983) from Harvard
and holds the Haas Distinguished Chair in Economic
University. Prior to coming to Georgetown, Holzer served
Disparities. She is the co-editor of the leading journal in
as chief economist for the U.S. Department of Labor and
economics, American Economic Review. Hoynes is an
professor of economics at Michigan State University. He has
economist and specializes in the study of poverty, inequality,
also been a visiting scholar at the Russell Sage Foundation and
and the impacts of government tax and transfer programs on
a faculty research fellow of the National Bureau of Economic
low-income families. Current projects include evaluating the
Research. Over most of his career, Holzer’s work has focused
impact of the Great Recession across demographic groups,
primarily on the low-wage labor market, with research on
examining the impact of Head Start on cognitive and non-
both employers and workers in that market as well as a range
cognitive outcomes, examining the impact of the Earned
of policy issues (such as workforce training, welfare reform,
Income Tax Credit on infant health, and estimating impacts
and wage supplements).
of U.S. food and nutrition programs on labor supply, health,
and human capital accumulation. In addition to her faculty
Susan N. Houseman appointment, Hoynes has research affiliations at the National
Bureau of Economic Research, the University of California,
Senior Economist, The W. E. Upjohn Institute for Employment Davis Center for Poverty Research and the Institute for Fiscal
Research Studies. She sits on the National Advisory Committee of
Susan Houseman is a senior economist at the W.E. Upjohn the Robert Wood Johnson Foundation Scholars in Health
Institute for Employment Research. Houseman chairs the Policy Research Program and the Advisory Committee for
Technical Advisory Committee to the Bureau of Labor the National Science Foundation’s Directorate for the Social,
Statistics, and is a member of the National Bureau of Economic Behavioral, and Economic Sciences.
Research Conference on Research in Income and Wealth.
Prior to joining the Upjohn Institute, Houseman was on the Ariel Kalil
faculty at the University of Maryland School of Public Policy
and was a visiting scholar at the Brookings Institution. She Professor, The Harris School of Public Policy, The University
received her doctorate in economics from Harvard University. of Chicago; Director, Center for Human Potential and Public
Houseman’s research has examined the effects of labor Policy, The University of Chicago
market policies, including policies that support work sharing,
Ariel Kalil is a professor in The Harris School of Public Policy
on workforce adjustment in response to demand changes in
at the University of Chicago, where she directs the Center for
the United States, Europe, and Japan. She currently is the
Human Potential and Public Policy. She is a developmental
principal investigator on a demonstration project funded by
psychologist who studies how economic conditions and
the Department of Labor that will test whether interventions
parents’ socioeconomic status affect child development
designed to reduce perceived barriers to work share increase
and parental behavior. Her recent projects have examined
use of this option. Houseman’s other research has focused on
the relationship between parental education and time with
employers’ use of temporary help employment, outsourcing,
children, the effects of the Great Recession on parental
and nonstandard work arrangements and their implications
behavior and child development, and the association between
for workers’ wages, benefits, and employment stability. Her
income inequality and children’s educational attainment.
The Hamilton Project • Brookings 159
Authors
Kalil received her Ph.D. in developmental psychology from classroom, including after-school programs, summer youth
the University of Michigan. Before joining Chicago Harris’ employment, and school meals. He holds a master of public
faculty in 1999, she completed a postdoctoral fellowship affairs from the Woodrow Wilson School at Princeton
at the University of Michigan’s National Poverty Center. University, and a doctorate in public administration from
Kalil has received the William T. Grant Foundation Faculty New York University’s Wagner School for Public Service.
Scholars Award, the Changing Faces of America’s Children At New York University, he was an Institute for Education
Young Scholars Award from the Foundation for Child Sciences Pre-doctoral Interdisciplinary Research Training
Development, and the National Academy of Education/ Fellow, and conducted research at NYU’s Institute for
Spencer Postdoctoral Fellowship. In 2003, she was the Education and Social Policy. Previously, Leos-Urbel worked
first-ever recipient of the Society for Research in Child with the Urban Institute, Abt Associates, The After-School
Development Award for Early Research Contributions. Corporation, and the U.S. Peace Corps.
Her recent work has been funded by the National Institute
of Child Health and Human Development and by the
MacArthur and Russell Sage Foundations. Robert I. Lerman
Professor, Department of Economics, American University;
Institute Fellow, Urban Institute
Melissa S. Kearney
Director, The Hamilton Project; Professor, Department of Robert Lerman is a professor in the department of economics
Economics, University of Maryland at American University. He researches and publishes on
employment, income support, and youth development,
Melissa S. Kearney is the director of the Hamilton Project; a especially as they affect low-income populations. In the 1970s,
senior fellow at the Brookings Institution; and a professor in he worked on reforming the nation’s income maintenance
the department of economics at the University of Maryland, programs and on youth employment policies as a staff economist
where she has been on the faculty since 2006. She is also for both the Congressional Joint Economic Committee and the
a research associate at the National Bureau of Economic U.S. Department of Labor. He was one of the first scholars to
Research. Kearney received her Ph.D. in economics from examine the patterns and economic determinants of unwed
the Massachusetts Institute of Technology in 2002 and her fatherhood and to propose a youth apprenticeship strategy in
bachelor’s degree from Princeton University in 1996. She the United States. He is currently an institute fellow at the Urban
studied on a National Science Graduate Research Fellowship Institute and a research associate at IZA in Bonn, Germany.
and a Harry S. Truman Scholarship. Past positions include Lerman is the founder and president of the American Institute
fellow at the Brookings Institution and assistant professor for Innovative Apprenticeship.
at Wellesley College. Kearney’s research focuses on issues of
social policy, poverty, and inequality and has been published
in leading economics journals. She is particularly interested
Phillip B. Levine
in the effect of government programs and economic Katharine Coman and A. Barton Hepburn Professor of
conditions on the behaviors and outcomes of economically Economics, Wellesley College
disadvantaged populations.
Phillip Levine is the Katharine Coman and A. Barton Hepburn
Professor of Economics at Wellesley College, a research
Jacob Leos-Urbel associate at the National Bureau of Economic Research,
Mary Toepelt Nicolai and George S. Blair Assistant Professor, a research affiliate of the National Poverty Center, and a
Department of Politics and Public Policy, Claremont Graduate member of the National Academy of Social Insurance. He has
University also served as a senior economist at the President’s Council
of Economic Advisers. Levine received a bachelor’s degree
Jacob Leos-Urbel is the Mary Toepelt Nicolai and George S. with honors from Cornell University in 1985 and a Ph.D. from
Blair Assistant Professor in American Politics, and teaches Princeton University in 1990. He has been a member of the
public policy courses in the department of politics and policy faculty at Wellesley since 1991. His research has largely been
at Claremont Graduate University. His research interests devoted to empirical examinations of the impact of social
focus broadly on education, and on child and youth policy. policy on individual behavior. Along with many publications
Leos-Urbel’s current research evaluates the effectiveness in academic journals and edited volumes, he is the author
of policies that aim to promote children’s education and of Sex and Consequences: Abortion, Public Policy, and the
development while operating outside the traditional school Economics of Fertility, coauthor of Reconsidering Retirement:
160 Policies to Address Poverty in America
How Losses and Layoffs Affect Older Workers, and coeditor Irma Perez-Johnson
of Targeting Investments in Children: Fighting Poverty When
Resources are Limited. Associate Director of Human Services Research,
Mathematica Policy Research
Bridget Terry Long Irma Perez-Johnson is an associate director of human services
research in Mathematica’s Princeton, New Jersey office. She
Academic Dean and Xander Professor of Education and is an expert on the design and implementation of rigorous
Economics, Harvard Graduate School of Education evaluations of employment training and K–12 education
programs, including pilots and demonstrations. For the U.S.
Bridget Terry Long is the Academic Dean and the Xander
Department of Labor, Perez-Johnson currently leads the
Professor of Education and Economics at the Harvard
evaluation of the Self-Employment Training Demonstration,
Graduate School of Education. She is also a faculty research
directs the study of Behavioral Interventions for Labor-
associate of the National Bureau of Economic Research and
Related Programs, and recently served as the project director
a research affiliate of Center for Analysis of Postsecondary
and principal investigator for the Evaluation of the Individual
Education and Employment. Her current projects examine
Training Account Experiment. She is also directing a study
the roles of information and assistance in promoting college
of Reemployment Services and Eligibility Assessments for
savings, the completion of aid applications, and college
recipients of Extended Unemployment Compensation. For
enrollment. Other work examines the effects of financial aid
the U.S. Department of Education, Institute of Education
programs and the impact of postsecondary remediation.
Sciences, she is helping design an Evaluation of Data-Driven
Instruction and until recently served as the deputy director
Sheena McConnell for the Impact Evaluation of Race-to-the-Top and School
Improvement Grants. She holds a doctorate in education
Vice President, Mathematica Policy Research
policy and evaluation from the University of Pennsylvania.
Sheena McConnell is a vice president and director of human
services research in Mathematica’s Washington, DC office.
Isabel Sawhill
Her research focuses primarily on evaluating policies to
promote employment and strong families in disadvantaged Cabot Family Chair and Senior Fellow, Brookings Institution
populations. She is currently directing a national experimental
Isabel Sawhill is the Cabot Family Chair and a Senior Fellow
evaluation of the Workforce Investment Act’s Adult and
at the Brookings Institution where she is co-director of the
Dislocated Workers programs for the U.S. Department of
Center on Children and Families. Sawhill’s areas of expertise
Labor, Employment, and Training Administration. These
are poverty, inequality, social mobility, education, fiscal
programs, which provide training, employment counseling,
policy, and the family. She is the author or editor of numerous
and other employment-related services, are the backbone of
books and articles. This policy memo is drawn in part from
the public workforce investment system in the United States.
her forthcoming book, Generation Unbound: Drifting into Sex
She also has evaluated the use of vouchers to pay for training,
and Parenthood without Marriage.
programs to assist low-income people who are interested
in starting their own business, and programs to assist low-
income couples improve their relationships. McConnell Diane Whitmore Schanzenbach
is an expert at designing and implementing experimental
evaluations of social service interventions for low-income Associate Professor, School of Education and Social Policy,
populations. Recently, she completed an experimental study Northwestern University
of the impact of teachers from Teach For America and the Diane Whitmore Schanzenbach is an associate professor in
Teaching Fellows programs on the test scores of students, most the School of Education and Social Policy at Northwestern
of whom were low-income. She received a bachelor’s of science University, a research associate at the National Bureau of
from Churchill College, Cambridge University, England, and Economic Research, and a research affiliate of the Institute
a doctorate in economics from Princeton University. for Research on Poverty. She studies issues related to child
poverty, including education policy, child health, and food
consumption. She graduated magna cum laude from Wellesley
College in 1995 with a bachelor’s degree in economics and
religion, and received a doctorate in economics in 2002 from
Princeton University. Much of her research investigates
The Hamilton Project • Brookings 161
Authors
the longer-run impacts of early life experiences, such as the Joanna Venator
impacts of receiving SNAP benefits during childhood, the
impacts of kindergarten classroom quality, and the impacts Research Assistant, Brookings Institution
of early childhood education. She recently served on the Joanna Venator is a research assistant in the Center on
Institute of Medicine’s Committee on the Examination of the Children and Families at the Brookings Institution.
Adequacy of Food Resources and SNAP Allotments. Venator’s research focuses on complex families, unintended
pregnancy and contraception, social mobility, and character
Amy Ellen Schwartz and opportunity. She received a bachelor’s degree from
Swarthmore College in 2013.
Professor, Steinhardt School of Culture, Education, and
Human Development and Robert F. Wagner Graduate School,
New York University; Director, New York University Institute James P. Ziliak
for Education and Social Policy; Daniel P. Moynihan Professor Carol Martin Gatton Endowed Chair in Microeconomics,
of Public Affairs, Maxwell School, Syracuse University Department of Economics, University of Kentucky; Director,
Amy Ellen Schwartz is professor of public policy, education, Center for Poverty Research, University of Kentucky
and economics, and director of the New York University James Ziliak holds the Carol Martin Gatton Endowed Chair
Institute for Education and Social Policy. She teaches in Microeconomics in the Department of Economics and is
courses in public finance and policy at both the Wagner founding director of the Center for Poverty Research at the
School and the Steinhardt School of Culture, Education, University of Kentucky. He served as assistant and associate
and Human Development. Her research is primarily in professor of economics at the University of Oregon, and has
applied econometrics, focusing on issues in urban policy, held visiting positions at the Brookings Institution, University
education policy, and public finance. Her current research in College London, University of Michigan, and University
K–12 education examines the relationship between student of Wisconsin. He currently serves as chair of the board of
performance and housing and neighborhood change, the overseers of the University of Michigan Panel Study of Income
role of schools and neighborhoods in shaping childhood Dynamics. His research expertise is in the areas of labor
obesity, immigration and mobility in urban schools, and the economics, poverty, food insecurity, and tax and transfer
efficacy of school reforms. Her research on urban economic policy. Recent projects include trends in earnings and income
development has included work on Business Improvement volatility in the United States, the origins of persistently poor
Districts, housing investment, school choice, and investment regions in America, the causes and consequences of hunger
in infrastructure, among other issues in public finance. among older Americans, and the effect of survey nonresponse
Schwartz has published numerous articles in academic journals on the level of and trends in poverty and inequality. He is
including the American Economic Review, Journal of Human editor of Welfare Reform and its Long Term Consequences
Resources, the Journal of Public Economics, and the Journal of for America’s Poor and Appalachian Legacy: Economic
Urban Economics. Her research has been supported by grants Opportunity after the War on Poverty.
from the Spencer Foundation, National Institute of Health,
Institute of Education Sciences, National Science Foundation,
W. T. Grant Foundation, and the John D. and Catherine T.
MacArthur Foundation, among others. Schwartz received her
doctorate in economics from Columbia University.
162 Policies to Address Poverty in America
Endnotes
Introduction 6.
The Census Bureau officially measures poverty by
comparing family income to a set of money income
thresholds that vary by family size and composition.
In 2013, the official threshold for a single nonelderly
individual was an annual income of $12,119; for a family of
two children and two adults, the threshold was $23,624. If
a family’s total income is less than the family’s threshold,
then that family and the individuals in it are considered
to be living in poverty. The income sources included in 7.
these calculations are before taxes and do not include
capital gains or noncash benefits (such as public housing,
Medicaid, and the Supplemental Nutrition Assistance
Program [SNAP]). Since 2010 the Census Bureau has
also released poverty estimates based on the SPM,
which is a more complex measure. The SPM income or
resource measure is cash income plus in-kind government
benefits (such as SNAP and housing subsidies) minus
nondiscretionary expenditures (taxes, medical expenses,
and work expenses). The SPM thresholds are adjusted for
geographic differences in the cost of living.
See John Karl Scholz’s Hamilton Project Discussion Paper
2007-14, “Employment-Based Tax Credits for Low-Skilled
Workers,” at http://www.hamiltonproject.org/papers/
employment-based_tax_credits_for_low-skilled_workers/.
See Diane Whitmore Schanzenbach’s Hamilton Project
Discussion Paper 2013-06, “Strengthening SNAP for a
More Food-Secure, Healthy America,” at http://www.
hamiltonproject.org/papers/strengthening_snap_for_a_
more_food-secure_healthy_america/.
See Melissa S. Kearney and Lesley Turner’s Hamilton
Project Discussion Paper 2013-07, “Giving Secondary
Earners a Tax Break: A Proposal to Help Low- and Middle-
Income Families,” at http://www.hamiltonproject.org/
papers/giving_secondary_earners_a_tax_break/.
See Lori G. Kletzer and Howard F. Rosen’s Hamilton Project
Discussion Paper 2006-06, “Reforming Unemployment
Insurance for the Twenty-First Century Workforce,”
at http://www.hamiltonproject.org/papers/reforming _
unemployment_insurance_for_the_twenty-first_century_
workforc/; and Jeffrey R. Kling’s Hamilton Project
Discussion Paper 2006-05, “Fundamental Restructuring
of Unemployment Insurance: Wage-Loss Insurance and
Temporary Earnings Replacement Accounts,” at http://www.
hamiltonproject.org/papers/fundamental_restructuring_of_
unemployment_insurance_wage-loss_insuranc/.
10.
For example, see Sandy Baum and Judith Scott-Clayton’s
Hamilton Project Discussion Paper 2013-04, “Redesigning the
Pell Grant Program for the Twenty-First Century,” at http://
www.hamiltonproject.org/papers/redesigning_the_pell_
grant_program_for_the_twenty-first_century/; and Caroline
M. Hoxby and Sarah Turner’s Hamilton Project Discussion
Paper 2013-03, “Informing Students about Their College
Options: A Proposal for Broadening the Expanding College
Opportunities Project,” at http://www.hamiltonproject.org/
papers/informing_students_about_their_college_options/.
See Rebecca M. Blank and Mark H. Greenberg’s Hamilton
Project Discussion Paper 2008-17, “Improving the
Measurement of Poverty,” at http://www.hamiltonproject.org/
papers/improving_the_measurement_of_poverty/.
See Michael Greenstone and colleagues’ Hamilton Project
Policy Memo, “Thirteen Economic Facts about Social Mobility
and the Role of Education,” at http://www.hamiltonproject.
org/papers/thirteen_economic_facts_social_mobility_
education/.
For example, see Derek Messacar and Philip Oreopoulos’s
Hamilton Project Discussion Paper 2012-07, “Staying in
School: A Proposal to Raise High School Graduation Rates,” at
http://www.hamiltonproject.org/papers/staying_in_school_a_
proposal_for_raising_high-school_graduation_rates/; Robert
Gordon, Thomas J. Kane, and Douglas O. Staiger’s Hamilton
Project Discussion Paper 2006-01, “Identifying Effective
Teachers Using Performance on the Job,” at http://www.
hamiltonproject.org/papers/identifying_effective_teachers_
using_performance_on_the_job/; Bradley M. Allan and
Roland G. Fryer Jr.’s Hamilton Project Discussion Paper 2011-
07, “The Power and Pitfalls of Education Incentives,” at http://
www.hamiltonproject.org/papers/the_power_and_pitfalls_of_
education_incentives/; Roland G. Fryer Jr.s Hamilton Project
Discussion Paper 2012-06, “Learning from the Successes and
Failures of Charter Schools,” at http://www.hamiltonproject.
org/papers/learning_from_the_successes_and_failures_of_
charter_schools/; and Aaron Chatterji and Benjamin F. Jones’s
Hamilton Project Discussion Paper 2012-05,” Harnessing
Technology to Improve K-12 Education, at http://www.
hamiltonproject.org/papers/harnessing_technology_to_
improve_k-12_education/.
See Louis S. Jacobson and Robert J. LaLonde’s Hamilton
«
Project Discussion Paper, “Using Data to Improve the
Performance of Workforce Training,” at http://www.
hamiltonproject.org/papers/using_data_to_improve_the_
performance_of_workforce_training/.
The Hamilton Project ¢ Brookings 163
Endnotes
11.
12.
13.
14.
See Harry J. Holzer’s Hamilton Project Discussion Paper
2011-10, “Raising Job Quality and Skills for American
Workers: Creating More-Effective Education and Workforce
Development Systems in the States,” at http://www.
hamiltonproject.org/papers/raising_job_quality_and_skills_
for_american_workers_creating_more-effe/.
See Harry J. Holzer’s Hamilton Project Discussion Paper
2007-15, “Better Workers for Better Jobs: Improving Worker
Advancement in the Low-Wage Labor Market,” at http://www.
hamiltonproject.org/papers/better_workers_for_better_jobs_
improving_worker_advancement_in_the_low/.
See David Autor’s Hamilton Project Discussion Paper, “The
Polarization of Job Opportunities in the U.S. Labor Market:
Implications for Employment and Earnings,” at http://www.
hamiltonproject.org/papers/the_polarization_of_job_
opportunities_in_the_u.s._labor_market_implica/.
See Diane Whitmore Schanzenbach’s Hamilton Project
Discussion Paper 2013-06, “Strengthening SNAP for a
More Food-Secure, Healthy America,” at http://www.
hamiltonproject.org/papers/strengthening_snap_for_a_more_
food-secure_healthy_america/.
Proposal 1
1.
See Fryer and Levitt (2013). That paper’s primary goal is to
explore the emergence of achievement gaps by race rather than
by income.
Duncan and Magnuson (2011) estimate a 1.3 standard
deviation difference in math and reading test performance
at the start of kindergarten between children in the lowest
and highest quintiles of the family income distribution.
The corresponding gaps in teacher ratings of attention and
behavior are 0.75 and 0.25 standard deviations, respectively.
These are the authors’ calculations from the 2011 October
Current Population Survey School Enrollment supplement
(NBER n.d.). The lowest family income quintile has a
maximum annual income of $17,500, while the top family
income quintile has a minimum annual income of $125,000.
The 2011-12 school year is the most recent with data available;
data are from the NIEER.
The proposal defines low- and moderate-income families
as those with income at or below 200 percent of the federal
poverty level. The Preschool for All initiative has other
elements as well, including incentives for states to implement
full-day kindergarten, a shift in the focus of Head Start toward
three-year-olds, and an expansion of the Early Head Start
program, which serves younger children.
For illustrative purposes this is drawn as a straight line, but
the actual relationship may be curved.
Recent work by Heckman, Pinto, and Savelyev (2013) finds
that the Perry program induced changes in personality skills,
which in turn explain a large portion of the improvement in
164 Policies to Address Poverty in America
10.
11.
12.
13.
adult outcomes. In a re-analysis of the Perry data, Anderson
(2008) finds that the positive impacts were found for girls
but not boys. Heckman et al. (2010) dispute the finding, and
contend that the positive benefit-to-cost ratios found in Perry
are for both boys and girls.
Ludwig and Miller (2007) take a different approach, comparing
children in counties that barely qualified for and barely missed
qualifying for special grant-writing assistance for Head Start
at the program’s inception. They find evidence that Head Start
reduces child mortality, and they find suggestive evidence that
it increases educational attainment.
In terms of benefit-to-cost ratios, the two programs appear
to be roughly equivalent, resulting in about $8.00 worth of
benefits for each $1.00 spent (Deming 2009; Heckman et al.
2010). The reason is that Head Start is relatively low cost.
In particular, the Tennessee program meets nine of the quality
benchmarks represented in figure 1-2, whereas the average
Head Start program meets only five (Espinosa 2002). Below we
discuss the potential limitations of using inputs to proxy for
quality.
This is not to suggest that preschool makes some children
worse off overall. High-socioeconomic status families
who choose to enroll their children in the public program
experience a reduction in out-of-pocket preschool spending
that offsets the decline in the learning environment.
To our knowledge, similar estimates for Georgia do not exist.
Recent results from a high-quality prekindergarten program
in Boston does find substantial short-term impacts on the
test scores of higher-income children (Weiland and
Yoshikawa 2013).
Proposal 3
1.
In an increasing number of cases these unmarried mothers
are living with the father of the child at the time of the birth,
but these cohabiting relationships are much less stable than
marriages and typically break up before the child is age five.
The term “unintended” comes from the National Survey
of Family Growth, which asks women to characterize the
intentionality of their pregnancies and births at the time they
first learned of their pregnancy. If they say the pregnancy was
unintended, they are further asked whether it was “unwanted”
or “mistimed.” An unwanted pregnancy is one the woman
did not want ever, whereas a mistimed pregnancy is one that
simply came earlier than she might have wanted—in some
cases by only a year, but in other cases by many years.
Literature on teen pregnancies suggests that most of the
correlation between having a baby as a teen and later
outcomes is due to confounding factors or unobserved traits
of the women involved. Quasi-natural experiments find
that teenagers who miscarry their pregnancy do not have
significantly better outcomes than teenagers who carry their
child to term (Hotz, McElroy, and Sanders 2005). However,
the broader literature on the effects of contraception shows
that it has increased women’s educational and labor-market
achievements quite dramatically (Bailey, Hershbein, and
Miller 2012; Goldin and Katz 2002).
An IUD isa contraceptive device that a provider inserts into
a patient’s uterus; an implant is a contraceptive device that a
provider places under a patient’s skin, typically on the arm.
Both procedures need to be done by a trained health-care
provider, usually a physician. Both last up to three years, with
some brands of IUD lasting up to twelve years.
This social marketing campaign coincided with Iowa’s
expansion of Medicaid family planning services in 2010 and a
huge increase in funding for family planning clinics starting
in 2007, so we cannot conclusively attribute this whole effect
to the social marketing campaign. However, it should be
noted that the decline in pregnancies accelerated during the
campaign. Whereas the percent of unintended pregnancies
dropped from 46.1 percent to 45.2 percent between 2007 and
2009, it dropped from 45.2 percent to 40.9 percent between
2009 and 2011.
The estimate of taxpayer savings for mistimed births does
not account for the fact that delaying a birth may result in
a woman having fewer children overall or may result in an
improvement of her living situation during the intermittent
years. It does, however, account for the fact that the present
discounted value of future benefit payouts is less than the
value of payouts now. See Monea and Thomas (2010) for more
information on how to derive this formula.
Proposal 4
1,
Earlier this year, President Obama introduced an initiative,
My Brother’s Keeper, calling for the private and philanthropic
sectors to institute mentoring programs. In his remarks he
credited those who “never gave up on me, and so I didn’t give
up on myself” (Obama 2014).
To provide a couple of examples, Child Trends identifies the
programs Fostering Healthy Futures and Parent Mentors
for Children with Asthma. These programs are targeted at
children who have suffered abuse in foster-care settings and
children who need assistance dealing with their respiratory
issues, respectively. Lawner, Beltz, and Moore (2013)
summarizes most of these programs.
The impact on GPA is only statistically significant at the
10 percent level. Given the overall strength of the results
indicating that academic ability improved and the ability to
more-easily translate GPA into subsequent wages, I conclude
that this is a meaningful effect and take this estimate
as a summary statistic of the educational impact of the
intervention. The point estimates also suggest that the effect
of mentoring on academic achievement in this experiment is
larger for girls than for boys, although these differences are
unlikely to be statistically significant (insufficient information
is provided to conduct a formal hypothesis test).
One possible explanation for the divergence in results is that
the pilot results were strongly (although not exclusively)
restricted to one of the five sites in which the program was
implemented (Levine and Zimmerman 2010). The ability of a
single administrator to make a program work and the inability
to replicate those results elsewhere is one potential weakness of
any smaller-scale intervention.
One shortcoming of this analysis is that we have access to only
the short-run effect of the Big Brothers Big Sisters intervention.
An active literature exists in other areas, such as the Head
Start program, that is concerned with test score fade-out and
the long-term impact on economic outcomes. We do not have
the ability to explore that issue more deeply in this context.
Yet the benefit-cost ratio we report here is so large that the
short-run impact would have to depreciate extensively to
substantively alter this result.
Proposal 5
1.
The program was implemented in five cities, spanning two
consecutive summers, which included summer jobs plus
academic remediation and training; evaluations found
short-term increases in reading and math scores, compared
to a comparison group that received only jobs.
We address this point later in our discussion of the costs and
benefits of summer jobs.
Seven months after the program, there were 3.7 fewer arrests
per 100 participants, a 51 percent decline.
This study sample of youth ages fourteen to twenty-four in
high crime neighborhoods included 421 participants and a
comparison group of 192 eligible youth from the waiting list.
Funds could be spent through June 2011. The one success
indicator for the program was achievement of workforce
readiness goals, which was up to local sites to define (Bellotti
et al. 2010).
An earlier SYEP was funded through the federal JTPA and
administered through the U.S. Department of Labor; see
appendix 5-B for more details.
For example, NYC’s CTE Summer Scholars summer paid
internship program rewards students who have perfect
attendance at the end of the program with a $500 bonus.
The YIEPP improvement of school enrollment rates could
be attributed to the requirement that students be enrolled in
school to participate—requiring participants to be enrolled in
the school year prior in order to be eligible might also provide
some incentive for students to stay in school.
New York State also administers a 15 percent ceiling for NYC
SYEP administrative costs.
For example, NYC’s SYEP allocates between $300 and $700
per participant for educational services, depending on the type
The Hamilton Project ¢ Brookings 165
Endnotes
10.
11.
12.
of youth and intensity of services offered.
For example, NYC’s Conditional Cash Transfer program
offered high school students a $600 incentive for each
Regents exam passed, but yielded no significant effect
(Riccio et al. 2013).
To give a sense of magnitudes of the costs of crime,
McCollister, French, and Fang (2010) estimate the societal
cost of household burglary at $6,169 in tangible costs and $321
in intangible costs, totaling $6,462; vandalism is valued at
$4,860 in tangible costs, with no intangible costs.
Focusing on youth still enrolled in school, summer
employment only, and including training closely connected to
the youth’s employment experiences separates this proposed
program from prior less-effective federal youth employment
programs. For example, the youth employment initiatives
funded by the 1982 JTPA targeted out-of-school youth who
are likely difficult to reach without intensive services and time
and had limited effects on participants (Bloom et al. 1997).
The YIEPP, a federal program operating under the
Comprehensive Employment and Training Act (CETA)
that preceded JTPA and targeted in-school youth, had small
positive impacts on school enrollment rates (Farkas, Smith,
and Stromsdorfer 1983). YIEPP did not, however, provide
training or job search assistance to students, and we believe
that a program that provides these connections has the
potential to provide greater benefits. (More details about these
programs can be found in appendix 5-B.)
Proposal 6
1,
Low income refers to the bottom 20 percent of all family
incomes, and high income refers to the top 20 percent.
The terms “remedial” and “developmental” are often used
interchangeably in the literature because some states favor
one term over the other. In this paper, both are meant to
refer to the courses and services offered to postsecondary
students below college level, including basic-skills training and
nontraditional coursework.
The question as posed by Complete College America (2012) is,
“Can an ‘open access’ college be truly open access if it denies
so many access to its college-level courses?”
In 2004, the data suggest only 27 percent of high school seniors
had completed high-level academic coursework, defined as
four years of English, three years of mathematics (including at
least one year of a course higher than Algebra II), three years
of science, three years of social studies, and two years of a
single non-English language (NCES 2010).
Greene and Foster (2003) define being minimally college
ready as (1) graduating from high school after (2) having
taken four years of English, three years of math, and two years
each of science, social science, and foreign language; and (3)
demonstrating basic literacy skills by scoring at least 265 on
the reading NAEP.
166 Policies to Address Poverty in America
10.
Those authors’ estimate is based on the number of first-time
degree-seeking fall enrollees and on assumptions about the
percent placed in remediation, the number of remedial courses
they will take, and the costs of providing a remedial course.
For example, according to Rebecca Trounson writing in the
Los Angeles Times on January 31, 2002 (“Cal State Ouster
Rate Rises Slightly”), in the fall of 2001, a California State
University campus “kicked out more than 2,200 students—
nearly 7 percent of the freshman class—for failing to master
basic English and math skills.”
The most widely used placement exams are the Computerized
Adaptive Placement Assessment and Support Systems
(COMPASS) and the Assessment of Skills for Successful Entry
and Transfer, each published by ACT, Inc., as well as the
ACCUPLACER, published by the College Board.
Moreover, she finds that the placement exam varies in how
well it predicts success in math versus English, and it does a
better job predicting who is likely to succeed rather than who
is likely to fail.
For example, see the California Early Assessment Program,
Kentucky Early Mathematics Testing Program, North
Carolina Early Mathematics Placement Testing Program,
Oklahoma Educational Planning and Assessment System, and
the Ohio Early Mathematics Placement Testing.
Proposal 7
1.
The figures come from tabulations by the author from the
March 2013 Current Population Survey (National Bureau of
Economic Research n.d.).
For a detailed look at the barriers to expanding apprenticeship
in the United States, see Lerman (2013).
Data from the combined 2001 and 2005 National Household
Education Surveys indicate that 1.5 percent of adults were in
an apprenticeship program in the prior year (National Center
for Education Statistics 2008). If these data are accurate, the
number of unregistered apprentices would far exceed the
number of registered ones.
See http://as.edu/index.html for the school’s Web site.
Proposal 8
1.
The National Fund for Workforce Solutions (http://www.
nfwsolutions.org/) is also trying to scale sectoral and career
pathway approaches in about thirty cities and regions
nationally.
Deming, Goldin, and Katz (2013) argue that the for-profit
colleges often have stronger incentives than public colleges to
keep up with evolving trends in labor demand. Rosenbaum,
Deil-Amen, and Person (2006) also argue that proprietary
occupational schools do a better job than community colleges
of having students complete vocational training and of linking
their students to jobs after graduation.
10.
The practical difficulty of measuring labor market shortages
is emphasized in Barnow, Trutko, and Piatak (2013). States
would have to decide how best to measure such shortages, or to
simply reward institutions for placing students into occupations
showing tightness or strong recent employment growth.
Up to one-fourth of the 3,000 or so Job Centers around the
country funded by the Workforce Investment Act are already
colocated on campuses. One proposal requiring all older
(defined as age twenty-five and above) Pell Grant recipients to
obtain career counseling at Job Centers appears in the College
Board (2013) recommendations for Pell Grant reform.
A range of market failures, such as imperfect information, may
also contribute to sub-optimal training by firms.
Hollenbeck’s (2008) evidence is descriptive but not rigorous,
as is earlier work by Ahlstrand, Bassi, and McMurrer (2003).
Other evidence on targeting tax credits to disadvantaged
workers using the federal Work Opportunity Tax Credit by
Hamersma (2014) suggests limited effectiveness. A number
of other studies looking at localized tax credits for employer
location or economic development (Bartik 2010; Busso,
Gregory, and Cline 2013; Faulk 2002; Ham et al. 2008) are
mixed as well, though many studies have been more positive in
the past few years. Holzer, imrohoroflu, and Swenson (1993)
also find positive effects on worker performance (as measured
by reductions in scrap rates) of a program for training grants
to small manufacturers in Michigan.
The availability of such data at the state level has been
encouraged by the State Longitudinal Data Systems grants
from the U.S. Department of Education and the Workforce
Data Quality Initiative from the U.S. Department of Labor,
as well as the Workforce Data Quality Campaign being
undertaken by the National Skills Coalition (described in
Zinn and Van Kluenen 2014).
Alternatively, the rewards might only be based on students
who remain in-state.
See Heckman, Heinrich, and Smith (2011) for a discussion of
how performance measures in workforce programs encourage
manipulation by the states of who is admitted to the workforce
system and whether they are ever counted among the program
exiters, on whose outcomes performance is measured.
Stackable credentials are a series of certifications representing
specific skills and competencies that might be more portable
than one specific occupational or industry certification.
Proposal 9
1,
While we recognize the potential importance of job search
assistance, job readiness training, and work experience, this
paper focuses on training programs that provide skills specific
to an occupation.
This includes the Job Training Partnership Act of 1982 (JTPA)
adult program, WIA Adult and Dislocated Worker programs,
H-1B Skill Training Grants, Trade Adjustment Assistance,
JTPA and WIA youth programs, YouthBuild, and Job Corps.
The U.S. Department of Labor is currently conducting a
national, experimental study of the WIA Adult and Dislocated
Worker programs. Findings on the short-term effectiveness of
the programs will be available in 2016.
Proposal 10
1.
Author’s calculations using data from two-year averages
of earnings by state in 2012 and 2013 Annual Social and
Economic Study of the Current Population Survey (U.S.
Census Bureau various years).
The CDCC and the CCDF require children under age thirteen
to be present for eligibility, while the CTC extends eligibility
to families with children under age seventeen. Employment
trends are nearly identical for mothers including this wider age
range of children.
Beginning with the 2011 wave, the Census Bureau has asked
respondents the amount of out-of-pocket child-care costs they
incur because of work. The numbers in table 10-1 and figure
10-3 pool the 2012 and 2013 survey years in order to reduce
the influence of outliers in smaller states (U.S. Census Bureau
various years).
The tax code also subsidizes child care through the employer-
provided child-care exclusion, which permits employers
to exclude up to $5,000 from an employee’s salary on a
pretax basis. There are other programs that assist with early
childhood development, such as Head Start, that are beyond
the scope of this paper.
Discretionary CCDBG grants are allocated to states based
on a formula that accounts for the state’s share of children
under age five, the state’s share of children receiving free or
reduced price lunch, and the state’s per capita income. Part of
the mandatory CCDF funds are allocated based on the state’s
funding for child-care programs authorized under the Aid
to Families with Dependent Children program in fiscal years
1994 and 1995, and part based on the state’s share of children
under age thirteen (Congressional Research Service 2012).
Since 1996 the basic TANF block grant to states totaled $16.5
billion, which had declined by about one-third in inflation-
adjusted terms by FY2012 (Congressional Research Service
2013). The state’s share of the block grant is a function of
its average expenditure on Aid to Families with Dependent
Children during FY1992-FY1994.
See https://daycare.com/states.html for links to each state’s
licensing requirements.
Proposal 11
1.
The maximum credit is the same for married and single filers.
However, the flat and phase-out regions of the credit are
expanded for married couples, in essence raising the EITC
The Hamilton Project ¢ Brookings 167
Endnotes
credit amounts for married filers with earnings over $17,000.
The SPM equivalence scale for families with one parents is
(1 + 0.8*first child + 0.5*other children)°’, which is equal to
1.50 for one-child families and 1.79 for two-child families.
Interestingly, the three-child EITC is already on par with the
two-child credit in equivalence-scale units: the three-child
equivalence scale is 2.06, suggesting a 15 percent higher
maximum benefit compared to the two-child credit; under
current law it is 12.5 percent higher.
Additionally, Baker (2008) and Strully, Rehkopf, and Xuan
(2010) find that the EITC increases average birth weight.
In this and the other calculations in this policy memo, we
assume incidence of the payroll tax is on the worker and thus
the worker “pays” the employer and employee portions of the
payroll tax. We also assume child-care costs of 10 percent of
gross earnings.
Earnings are predicted to decrease for married couples
through the modest predicted reduction in work for
secondary earners (Eissa and Hoynes 2006). Figures
calculate households’ taxes based on earnings and
demographic variables from the March Current Population
Survey, as well as Census Bureau estimates of tax filing units
and adjusted gross income. Poverty status is based on after-
tax resources of the SPM family unit.
Proposal 12
1,
Perhaps not surprisingly, the study found no effects of work-
sharing programs on the level of temporary employment
during the recession. Belgium also made heavy use of work
sharing during the downturn, but quantifying the impact on
employment is complicated by the fact that work sharing was
prevalent there even before the recession began.
Because employer repayment normally is spread out over
a number of years and states do not charge interest on the
balances employers owe, the present value of the benefits
paid out typically exceeds the present value of the employer
reimbursement. States also set minimum and maximum
UI tax rates; for employers already at these minimum or
maximum rates, the cost of an additional layoff may be
very low or zero. Additionally, in most states, if a laid-off
worker has worked recently for other employers, those
previous employers will be charged a prorated portion of the
UI benefits the worker receives. UI systems in many other
advanced countries are not experience-rated at all, meaning
that employers do not bear any of the cost of UI benefits paid
to their employees, but as already noted, employers in these
countries are typically subject to stringent advance notice and
severance pay requirements.
The Affordable Care Act requires, in essence, that all large
employers (defined as employers with fifty or more full-time
equivalent employees) offer health insurance coverage to their
full-time workers (defined as individuals who work thirty
168 Policies to Address Poverty in America
hours a week or more on average). Employer size is to be
determined by employment during the prior calendar year; in
cases where an individual employee’s hours are variable, full-
time status is based on the hours worked during a base period
and that status holds for the following six to twelve months
even if the employee’s hours change.
Some states also charge 100 percent of work-share benefits to
the work-share employer rather than prorating the charges
to all recent employers, or require all work-share employers
already at the maximum UI tax rate or with negative reserve
balances to fully reimburse the state for benefits paid. Such
provisions also likely have discouraged the use of work
sharing. The Department of Labor recently informed states
that these practices are no longer permitted, and states have
begun to amend their laws to comply with this directive, but
the Department does not believe it has the legal authority to
require the additional changes we are recommending.
Proposal 13
A statutory minimum wage is a binding, broad-based minimal
pay standard set by legal statute, as opposed to by collective
bargaining or other voluntary agreements. Some countries
(e.g., Sweden and Switzerland) do not have a statutory
minimum wage, but do have sectoral pay standards set by
collective bargaining.
Had the minimum wage been indexed to inflation in the
same manner as the IRS tax code or Social Security payments
(i.e., using the CPI-U), it would have been $10.93 per hour in
2014. The CPI-U-RS is a more reliable gauge of past cost of
living, however. Conversely, if we were to use the Personal
Consumption Expenditure deflator, the 1968 value of the
minimum wage would be $8.56 per hour. In all cases, however,
the real minimum wage has fallen since the 1960s and 1970s.
Proposal 14
1.
We recognize that there are other definitions of rapid-cycle
evaluation that will not utilize a comparison group. In this
paper, we focus on the assessment of rapid experiments using
comparison groups.
It is important to note that the performance of predictive
analytics can vary. A number of considerations, including
the extent to which strong predictors are available and the
quality of the data, can affect performance. The strength of
the underlying predictive models should be assessed before
deploying predictive analytics in high-stakes situations.
‘The estimated equation generated by this model is used as
part of an automated algorithm to find at-risk patients for
intervention among those newly admitted to the hospital. In
one early pilot, inpatient readmissions declined by 45 percent
(Raven 2009; Raven et al. 2009).
Rigorous experimental techniques include randomized
controlled trials and orthogonal research designs, and rigorous
quasi-experimental designs include regression discontinuity
research designs. These designs can be used to determine
whether an intervention caused an outcome. Like a clinical
drug trial, randomized controlled trials create randomly
formed treatment and control groups, each receiving a
different intervention. Orthogonal research designs use a
similar approach but test variation in the components of
an intervention. Regression discontinuity studies create a
treatment group with individuals above (or below) a certain
eligibility threshold (with individuals on the other side of
the threshold forming the control group), and use analysis
techniques to control for the eligibility score in the assessment
of the program.
While we are not aware of specific, published examples of the
use of these methods together, we believe the integration of
these approaches is powerful and compelling, as the discussion
that follows demonstrates.
It is noteworthy that examining multiple groups requires
additional sample observations if the same level of precision is
to be obtained. Generalizing beyond one hospital or program
likewise requires additional sample observations.
The Hamilton Project ¢ Brookings
169
170 Policies to Address Poverty in America
References
Introduction Edwards, Ashley N. 2014. “Dynamics of Economic Well-Being:
Poverty, 2009–2011.” Household Economic Studies, U.S.
Abraham, Katharine G., and Susan N. Houseman. Forthcoming. Census Bureau, Washington, DC.
“Short-Time Compensation as a Tool to Mitigate Job Evans, William, and Craig Garthwaite. 2014. “Giving Mom a Break:
Loss? Evidence on the U.S. Experience during the Recent The Impact of Higher EITC Payments on Maternal Health.”
Recession.” Industrial Relations. American Economic Journal: Economic Policy 6 (2): 258–90.
Adelman, Clifford. 2006. “The Toolbox Revisited: Paths to Degree Fox, Liana, Irv Garfinkel, Neeraj Kaushal, Jane Waldfogel, and
Completion from High School through College.” U.S. Christopher Wimer. 2013. “Waging War on Poverty:
Department of Education, Washington, DC. Historical Trends in Poverty Using the Supplemental
Almond, Douglas, Hilary W. Hoynes, and Diane Whitmore Poverty Measure.” Working Paper 13-02, Columbia
Schanzenbach. 2011. “Inside the War on Poverty: The Population Research Center, Columbia University, New
Impact of Food Stamps on Birth Outcomes.” Review of York, NY.
Economics and Statistics 93 (2): 387–403. Greene, Jay, and Greg Foster. 2003. “Public High School Graduation
Autor, David H., Lawrence F. Katz, and Melissa S. Kearney. and College Readiness Rates in the United States.” Working
2008. “Trends in U.S. Wage Inequality: Revising the Paper No. 3, Manhattan Institute, New York, NY.
Revisionists.” Review of Economics and Statistics 90 (2): Guryan, Jonathan, Eric Hurst, and Melissa Kearney. 2008.
300–323. “Parental Education and Parental Time with Children,”
Bailey, Martha J., and Susan M. Dynarski. 2011. “Inequality in Journal of Economic Perspectives 22 (3): 23-46.
Postsecondary Education.” In Whither Opportunity?: Rising Heckman, James J., Jora Stixrud, and Sergio Urzua. 2006. “The
Inequality, Schools, and Children’s Life Chances edited by Effects of Cognitive and Noncognitive Abilities on Labor
Greg J. Duncan and Richard J. Murnane, 117–131. New Market Outcomes and Social Behavior.” Working Paper
York: Russell Sage Foundation and Spencer Foundation. 12006, National Bureau of Economic Research, Cambridge,
Bishaw, Alemayehu. 2013. “Poverty: 2000 to 2012.” American MA.
Community Survey Briefs, U.S. Census Bureau, Hoynes, Hilary W., Diane Whitmore Schanzenbach, and Douglas
Washington, DC. Almond. 2012. “Long Run Impacts of Childhood Access to
Bruce, Mary, and John Bridgeland. 2014. “The Mentoring the Safety Net.” Working Paper 18535, National Bureau of
Effect: Young People’s Perspectives on the Outcomes Economic Research, Cambridge, MA.
and Availability of Mentoring.” Civic Enterprises with Hoynes, Hilary, Douglas L. Miller, and David E. Simon.
Hart Research Associates for MENTOR: The National Forthcoming. “Income, the Earned Income Tax Credit,
Mentoring Partnership, Washington, DC. and Infant Health.” American Economic Journal: Economic
Cavell, Timothy, David DuBois, Michael Karcher, Thomas Keller, Policy.
and Jean Rhodes. 2009. “Policy Brief: Strengthening Ingels, Steven J., Jeffrey A. Owings, and Phillip Kaufman. 2002.
Mentoring Opportunities for At-Risk Youth.” National “Coming of Age in the 1990s: The Eighth-Grade Class
Mentoring Center, Portland, OR. of 1988 12 Years Later.” National Center for Education
Center on Budget and Policy Priorities (CBPP). 2014. “Policy Basics: Statistics, Washington, DC.
The Earned Income Tax Credit.” Center on Budget and Kalil, Ariel, Rebecca Ryan, and Michael Corey. 2012. “Diverging
Policy Priorities, Washington, DC. Destinies: Maternal Education and the Developmental
Dahl, Gordon, and Lance Lochner. 2012. “The Impact of Family Gradient in Time with Children.” Demography 49 (4):
Income on Child Achievement: Evidence from the Earned 1361–83.
Income Tax Credit.” American Economic Review 102(5): Kearney, Melissa S., Benjamin H. Harris, Elisa Jácome, and Lucie
1927–56. Parker. 2013. “A Dozen Facts about America’s Struggling
Danziger, Sheldon, and Sandra K. Danziger. 2005. “The U.S. Social Lower-Middle Class.” Policy Memo, The Hamilton Project,
Safety Net and Poverty: Lessons Learned and Promising Brookings Institution, Washington, DC.
Approaches.” Population Studies Center, Institute for Social Knudsen, Eric I., James J. Heckman, Judy L. Cameron, and Jack
Research, University of Michigan, Ann Arbor, MI. P. Shonkoff. 2006. “Economic, Neurobiological, and
DeNavas-Walt, Carmen, Bernadette D. Proctor, and Jessica C. Behavioral Perspectives on Building America’s Future
Smith. 2013. “Income, Poverty, and Health Insurance Workforce.” Working Paper 12298, National Bureau of
Coverage in the United States: 2012.” Current Population Economic Research, Cambridge, MA.
Reports, U.S. Census Bureau, Washington, DC.
The Hamilton Project • Brookings 171
References
Rank, Mark R., and Thomas A. Hirschl. 1999. “The Likelihood of Fryer, Roland G., and Steven D. Levitt. 2013. “Testing for Racial
Poverty across the American Adult Life Span.” Social Work Differences in the Mental Ability of Young Children.”
44 (3): 201–16. American Economic Review 103 (2): 981–1005.
Shattuck, Rachel M., and Rose M. Kreider. 2013. “Social and Garces, Eliana, Duncan Thomas, and Janet Currie. 2002. “Longer-
Economic Characteristics of Currently Unmarried Women Term Effects of Head Start.” American Economic Review 92
With a Recent Birth: 2011.” American Community Survey (4): 999–1012.
Reports, U.S. Census Bureau, Washington, DC. Gordon, Robert, and Sara Mead. 2014. “How to Fix Head Start.”
Swanson, Christopher B. 2009. “Cities in Crisis 2009: Closing the Brookings Institution, Washington, DC.
Graduation Gap: Educational and Economic Conditions in Gormley, William T., and Ted Gayer. 2005. “Promoting School
America’s Largest Cities.” Editorial Projects in Education, Readiness in Oklahoma An Evaluation of Tulsa’s Pre-K
Bethesda, MD. Program.” Journal of Human Resources 40 (3): 533–58.
Waldfogel, Jane, and Elizabeth Washbrook. 2011. “Early Years Gramlich, Edward M. 1986. “Evaluation of Education Projects:
Policy.” Child Development Research 2011 (2011): 1–12. The Case of the Perry Preschool Program.” Economics of
Education Review 5 (1): 17–24.
Proposal 1 Heckman, James J., Seong Hyeok Moon, Rodrigo Pinto, Peter A.
Savelyev, and Adam Yavitz. 2010. “The Rate of Return to
Anderson, Michael L. 2008. “Multiple Inference and Gender the HighScope Perry Preschool Program.” Journal of Public
Differences in the Effects of Early Intervention: A Economics 94 (1): 114–28.
Reevaluation of the Abecedarian, Perry Preschool, and Heckman, James J., Rodrigo Pinto, and Peter Savelyev. 2013.
Early Training Projects.” Journal of the American Statistical “Understanding the Mechanisms through Which an
Association 103 (484): 1481–95. Influential Early Childhood Program Boosted Adult
Barnett, W. Steven, Megan E. Carolan, Jen Fitzgerald, and James H. Outcomes.” American Economic Review 103 (6): 2052–86.
Squires. 2012. “The State of Preschool 2012: State Preschool The Inter-university Consortium for Political and Social Research
Yearbook.” National Institute for Early Education Research, (ICPSR). n.d. “Current Population Survey Series.” The
Graduate School of Education, Rutgers University, Rutgers, Inter-university Consortium for Political and Social
NJ. Research, Institute for Social Research, University of
Bassok, Daphna. 2012. “Competition or Collaboration? Head Start Michigan, Ann Arbor. https://www.icpsr.umich.edu/
Enrollment during the Rapid Expansion of State Pre- icpsrweb/ICPSR/series/24.
kindergarten.” Educational Policy 26 (1): 96–116. Lipsey, Mark W., Kerry G. Hofer, Nianbo Dong, Dale C. Farran,
Cascio, Elizabeth, and Diane Whitmore Schanzenbach. 2013. “The and Carol Bilbrey. 2013a. “Evaluation of the Tennessee
Impacts of Expanding Access to High-Quality Preschool Voluntary Prekindergarten Program: End of Pre-K
Education.” Brookings Papers on Economic Activity, Fall, Results from the Randomized Control Design.” Research
127–92. Brookings Institution, Washington, DC. Report, Peabody Research Institute, Vanderbilt University,
Currie, Janet. 2001. “Early Childhood Education Programs.” Nashville, TN.
Journal of Economic Perspectives 15 (2, Spring): 213–38. ———. 2013b. “Evaluation of the Tennessee Voluntary
Currie, Janet, and Duncan Thomas. 1995. “Does Head Start Make Prekindergarten Program: Kindergarten and First Grade
a Difference?” American Economic Review 85 (3, June): Follow-Up Results from the Randomized Control Design.”
341–64. Research Report, Peabody Research Institute, Vanderbilt
Deming, David. 2009. “Early Childhood Intervention and Life- University, Nashville, TN.
Cycle Skill Development: Evidence from Head Start.” Ludwig, Jens, and Douglas L. Miller. 2007. “Does Head Start
American Economic Journal: Applied Economics 1 (3): Improve Children’s Life Chances? Evidence from a
111–34. Regression Discontinuity Design.” Quarterly Journal of
Duncan, Greg, and Katherine Magnuson. 2011. “The Nature and Economics 122 (1): 159–208.
Impact of Early Achievement Skills, Attention Skills, and Martin, Joyce A., Brady E. Hamilton, Michelle J. K. Osterman, Sally
Behavior Problems.” In Whither Opportunity?: Rising C. Curtin, and T. J. Mathews. 2013. “Births: Final Data for
Inequality, Schools, and Children’s Life Chances, edited by 2012.” Table 1. National Vital Statistics System, National
Greg J. Duncan and Richard J. Murnane, 47–69. New York: Center for Health Statistics, Centers for Disease Control
Russell Sage Foundation and Spencer Foundation. and Prevention, U.S. Department of Health and Human
Espinosa, Linda M. 2002. “High-Quality Preschool: Why We Services, Washington, DC. http://www.cdc.gov/nchs/data/
Need It and What It Looks Like.” NIEER Preschool Policy nvsr/nvsr62/nvsr62_09.pdf.
Matters 1, National Institute for Early Education, New National Bureau of Economic Research (NBER). n.d. “NBER CPS
Brunswick, NJ. Supplements.” National Bureau of Economic Research,
Fitzpatrick, Maria D. 2008. “Starting School at Four: The Effect Cambridge, MA. http://www.nber.org/data/current-
of Universal Pre-Kindergarten on Children’s Academic population-survey-data.html.
Achievement.” The B.E. Journal of Economic Analysis &
Policy 8 (1).
172 Policies to Address Poverty in America
National Center for Health Statistics. 2005. Vital Statistics of the
United States, 2003, Volume I, Natality. Centers for Disease
Proposal 2
Control and Prevention, U.S. Department of Health and Administration for Children and Families (ACF). 2002. “Making
Human Services, Hyattsville, Maryland. a Difference in the Lives of Infants and Toddlers and Their
Office of Head Start (OHS). Various years. “Head Start Program Families: The Impacts of Early Head Start, vol. 1 Final
Fact Sheets.” Office of Head Start, Administration for Technical Report.” U.S. Department of Health and Human
Children and Families, U.S. Department of Health and Services, Washington, DC.
Human Services, Washington, DC. http://eclkc.ohs.acf.hhs. Altintas, Evrim. 2012. “Parents’ Time with Children: Micro and
gov/hslc/mr/factsheets. Macro Perspectives.” Unpublished doctoral dissertation,
Office of the Press Secretary. 2013. “Fact Sheet President Obama’s Nuffield College, University of Oxford, UK.
Plan for Early Education for all Americans.” Office of the Aos, Steve, Roxanne Lieb, Jim Mayfield, Marna Miller, and Annie
Press Secretary, The White House, Washington, DC. Pennucci. 2004. “Benefits and Costs of Prevention and
Puma, Mike, Stephen Bell, Ronna Cook, Camilla Heid, Pam Early Intervention Programs for Youth.” Washington State
Broene, Frank Jenkins, Andrew Mashburn, and Jason Institute for Public Policy, Olympia, WA.
Downer. 2012. “Third Grade Follow-up to the Head Ashraf, Nava, Dean Karlan, and Wesley Yin. 2006. “Tying Odysseus
Start Impact Study Final Report.” OPRE Report to the Mast: Evidence from a Commitment Savings Product
# 2012-45b. Office of Planning, Research and Evaluation, in the Philippines.” Quarterly Journal of Economics 121:
Administration for Children and Families, U.S. 635–72.
Department of Health and Human Services, Atkinson, Anthony B., Thomas Piketty, and Emmanuel Saez. 2011.
Washington, DC. “Top Incomes in the Long Run of History.” Journal of
Reardon, Sean. 2011. “The Widening Academic-Achievement Economic Literature 49: 3–71.
Gap between the Rich and the Poor: New Evidence and Babcock, Elisabeth D. 2014. “Using Brain Science to Design New
Possible Explanations.” In Whither Opportunity?: Rising Pathways Out of Poverty.” Crittenton Women’s Union,
Inequality, Schools, and Children’s Life Chances, edited by Boston, MA.
Greg J. Duncan and Richard J. Murnane, 91–116. New York: Baggett, Kathleen M., Betsy Davis, Edward G. Feil, Lisa B. Sheeber,
Russell Sage Foundation and Spencer Foundation. Susan H. Landry, Judith J. Carta, and Craig Leve. 2010.
Robin, Kenneth B., Ellen C. Frede, and W. Steven Barnett. 2006. “Technologies for Expanding the Reach of Evidence-Based
“Is More Better? The Effects of Full-Day vs. Half-Day Interventions: Preliminary Results for Promoting Social-
Preschool on Early School Achievement.” NIEER Working Emotional Development in Early Childhood.” Topics in
Paper, National Institute for Early Education Research, Early Childhood Special Education 29 (4): 226–38.
New Brunswick, NJ. Bailey, Martha J., and Susan M. Dynarski. 2011. “Inequality in
Sabol, Terri J., Sandra L. Soliday Hong, Robert C. Pianta, and Postsecondary Education.” In Whither Opportunity?: Rising
Margaret R. Burchinal. 2013. “Can Rating Pre-K Programs Inequality, Schools, and Children’s Life Chances, edited
Predict Children’s Learning?” Science 341 (6148): 845–46. by Greg J. Duncan and Richard J. Murnane, 117–32. New
Schweinhart, Lawrence J., Jeanne Montie, Zongping Xiang, W. York: Russell Sage Foundation and Spencer Foundation.
Steven Barnett, Clive R. Belfield, and Milagros Nores. 2005. Belley, Philippe, and Lance Lochner. 2007. “The Changing Role of
“Lifetime Effects: The HighScope Perry Preschool Study Family Income and Ability in Determining Educational
Through Age 40.” HighScope Press, Ypsilanti, MI. Achievement.” Journal of Human Capital 1: 37–89.
Walters, Christopher. 2014. “Inputs in the Production of Early- Bierman, Karen, Brenda Heinrichs, Robert Lee Nix, Janet
childhood Human Capital: Evidence from Head Start.” Welsh, Scott David Gest, Celene Elizabeth Domitrovich,
White Paper, University of California, Berkeley, and Sukhdeep Gill. 2013. “The Head Start REDI
Berkeley, CA. Parenting Program: Extending the Impact of Classroom
Weiland, Christina, and Hirokazu Yoshikawa. 2013. “Impacts of Intervention.” Paper presented at the Society for Prevention
a Prekindergarten Program on Children’s Mathematics, Research Twenty-First Annual Meeting.
Language, Literacy, Executive Function, and Emotional Charness, Gary, and Uri Gneezy. 2005. “Incentives to Exercise.”
Skills.” Child Development 84 (6): 2112–30. Econometrica 77: 909–31.
Wong, Vivian C., Thomas D. Cook, W. Steven Barnett, and Chetty, Raj, John N. Friedman, Nathaniel Hilger, Emmanuel Saez,
Kwanghee Jung. 2008. “An Effectiveness-Based Evaluation Diane Whitmore Schanzenbach, and Danny Yagan. 2011.
of Five State Pre-Kindergarten Programs.” Journal of Policy “How Does Your Kindergarten Classroom Affect Your
Analysis and Management 27 (1): 122–154. Earnings? Evidence from Project STAR.” Quarterly Journal
Yoshikawa, Hirokazu, Christina Weiland, Jeanne Brooks-Gunn, of Economics 126: 1593–1660
Margaret R. Burchinal, Linda M. Espinosa, William T. Currie, Janet. 2001. “Early Childhood Intervention Programs.”
Gormley, Jens Ludwig, Katherine A. Magnuson, Deborah Journal of Economic Perspectives 15: 213–38.
Phillips, and Martha J. Zaslow. 2013. Investing in Our
Future: The Evidence Base on Preschool Education. Policy
Monograph, Society for Research in Child Development.
The Hamilton Project • Brookings 173
References
Davidson, Richard J., Jon Kabat-Zinn, Jessica Schumacher, Landry, Susan H., Jason L. Anthony, Paul R. Swank, and Pauline
Melissa Rosenkranz, Daniel Muller, Saki F. Santorelli, Monseque-Bailey. 2009. “Effectiveness of Comprehensive
Ferris Urbanowski, Anne Harrington, Katherine Bonus, Professional Development for Teachers of At-Risk
and John F. Sheridan. 2003. “Alterations in Brain and Preschoolers.” Journal of Educational Psychology 101 (2):
Immune Function Produced by Mindfulness Meditation.” 448–65.
Psychosom Medicine 65 (4): 564–70. Love, John M., Ellen Eliason Kisker, Christine Ross, Helen Raikes,
Deming, David. 2009. “Early Childhood Intervention and Life- Jill Constantine, Kimberly Boller, Jeanne Brooks-Gunn,
Cycle Skill Development: Evidence from Head Start.” Rachel Chazan-Cohen, Lousia Banks Tarullo, Christy
American Economic Journal: Applied Economics 1: 111–34. Brady-Smith, Allison Sidle Fuligni, Peter Z. Schochet,
Duncan, Greg J., and Katherine Magnuson. 2011. “The Nature Diane Paulsell, and Cheri Vogel. 2005. “The Effectiveness
and Impact of Early Achievement Skills, Attention Skills, of Early Head Start for 3-Year-Old Children and their
and Behavior Problems.” In Whither Opportunity?: Rising Parents: Lessons for Policy and Programs.” Developmental
Inequality, Schools, and Children’s Life Chances, edited by Psychology 41 (6): 885–901.
Greg J. Duncan and Richard J. Murnane, 47–69. New York: Ludwig, Jens, and Deborah A. Phillips. 2008. “Long-Term Effects of
Russell Sage and Spencer Foundation. Head Start on Low-Income Children.” Annals of the New
Duncan, Greg J., Ariel Kalil, Kjetil Telle, and Katherine M. Ziol- York Academy of Sciences 1136: 257–68.
Guest. 2013. “Increasing Inequality in Parent Incomes Mani, Anandi, Sendhil Mullainathan, Eldar Shafir, and Jiaying
and Children’s Completed Schooling: Correlation or Zhao. 2013. “Poverty Impedes Cognitive Function.” Science
Causation?” Paper presented at the Annual Meeting of the 341 (6149): 976–80.
Population Association of America, New Orleans, LA. Michalopoulos, Charles, Anne Duggan, Virginia Knox, Jill H.
Duncan, Greg J., Ariel Kalil, Susan E. Mayer, Robin Tepper, and Filene, Helen Lee, Emily K. Snell, Sarah Crowne, Erika
Monique R. Payne. 2005. “The Apple Does Not Fall Far Lundquist, Phaedra S. Corso, and Justin B. Ingels. 2013.
from the Tree.” In Unequal Chances: Family Background “Revised Design for the Mother and Infant Home Visiting
and Economic Success, edited by Samuel Bowles, Herbert Program Evaluation.” Report 2013-18, Office of Planning,
Gintis, and Melissa Osborne Groves, 23–79. Princeton, NJ: Research and Evaluation, Administration for Children and
Princeton University Press. Families, U.S. Department of Health and Human Services,
Eckenrode, John, Mary Campa, Dennis W. Luckey, Charles R. Washington, DC.
Henderson, Robert Cole, Harriet Kitzman, Elizabeth Milkman, Katherine L., John Beshears, James J. Choi, David
Anson, Kimberly Sidora-Arcoleo, Jane Powers, and David Laibson, and Brigitte C. Madrian. 2011. “Using
Olds. 2010. “Long-Term Effects of Prenatal and Infancy Implementation Intentions Prompts to Enhance Influenza
Nurse Home Visitation on the Life Course of Youths: Vaccination Rates.” Proceedings of the National Academy of
Nineteen-Year Follow-up of a Randomized Trial.” Archives Sciences 108 (26): 10415–20.
of Pediatric and Adolescent Medicine 164 (5): 9–15. Mullainathan, Sendhil, and Eldar Shafir. 2013. “Decision Making
Fudenberg, Drew, and David K. Levine. 2006. “A Dual-Self Model and Policy in the Context of Poverty.” In Behavioral
of Impulse Control.” American Economic Review 96 (5): Foundations of Public Policy, edited by Eldar Shafir,
1449–76. 281–300. Princeton, NJ: Princeton University Press.
Furstenberg, Frank F. 2011. “The Challenges of Finding Causal Office of Management and Budget (OMB). 2013. “Next Steps in the
Links between Family Educational Practices and Schooling Evidence and Innovation Agenda.” Office of Management
Outcomes.” In Whither Opportunity?: Rising Inequality, and Budget, Executive Office of the President, Washington,
Schools, and Children’s Life Chances, edited by Greg J. DC.
Duncan and Richard J. Murnane, 465–82. New York: Olds, David L., John Eckenrode, Charles R. Henderson, Harriet
Russell Sage Foundation and Spencer Foundation. Kitzman, Jane Powers, Robert Cole, Kimberly Sidora,
Guryan, Jonathan, Erik Hurst, and Melissa Kearney. 2008. Pamela Morris, Lisa M. Pettit, and Dennis Luckey. 1997.
“Parental Education and Parental Time with Children.” “Long-Term Effects of Home Visitation on Maternal
Journal of Economic Perspectives 22 (3): 23–46. Life Course and Child Abuse and Neglect. Fifteen-Year
Hurst, Erik. 2010. “Comments and Discussion.” Brookings Papers Follow-Up of a Randomized Trial.” Journal of the American
on Economic Activity, 177–84. Brookings Institution, Medical Association 278 (8): 637–43.
Washington, DC. Olds, David L., Charles R. Henderson, and Harriet J. Kitzman.
Kalil, Ariel, Rebecca Ryan, and Michael Corey. 2012. “Diverging 1994. “Does Prenatal and Infancy Nurse Home Visitation
Destinies: Maternal Education and the Developmental Have Enduring Effects on Qualities of Parental Caregiving
Gradient in Time with Children.” Demography 49: 1361–83. and Child Health at 25 to 50 Months of Life?” Pediatrics 93
Kamenica, Emir. 2012. “Behavioral Economics and the Psychology (1): 89–98.
of Incentives.” Annual Review of Economics 4: 427–52. The Pew Charitable Trusts. 2014. “Plenary Presentation at the
Laibson, David. 1997. “Golden Eggs and Hyperbolic Discounting.” National Summit on Quality in Home Visiting Programs.”
Quarterly Journal of Economics 112: 443–77. Public Opinion on Home Visiting. January 29–30, 2014.
174 Policies to Address Poverty in America
Price, Joseph. 2010. “The Effect of Parental Time Investments: Zaveri, Heather, Andrew Burwick, and Erin Maher. 2014. “Home
Evidence from Natural Within-Family Variation.” Visiting: The Potential for Cost Savings from Home
Unpublished paper, Brigham Young University, Provo, UT. Visiting Due to Reductions in Child Maltreatment.” Casey
Reardon, Sean F. 2011. “The Widening Socioeconomic Status Family Programs, Washington, DC.
Achievement Gap: New Evidence and Possible
Explanations.” In Whither Opportunity?: Rising Inequality
and the Uncertain Life Chances of Low-Income Children,
Proposal 3
edited by Greg J. Duncan and Richard J. Murnane. New Bailey, Martha J., Brad Hershbein, and Amalia R. Miller. 2012.
York: Russell Sage Foundation and Spencer Foundation. “The Opt-In Revolution? Contraception and the Gender
Reeves, Richard, and Kimberly Howard. 2013. “The Parenting Gap.” Gap in Wages.” Working Paper 17922, National Bureau of
Brookings Institution, Washington, DC. Economic Research, Cambridge, MA.
Rowe, Meredith L. 2008. “Child-Directed Speech: Relation to Centers for Disease Control and Prevention (CDC). 2014.
Socioeconomic Status, Knowledge of Child Development, “VitalStats–Births.” Centers for Disease Control and
and Child Vocabulary Skill.” Journal of Child Language 35: Prevention, Washington, DC.
185–205. Darney, Philip, Ashlesha Patel, Kimberly Rosen, Lena S. Shapiro,
Schweinhart, Lawrence J., Jeanne Montie, Zongping Xiang, W. and Andrew M. Kaunitz. 2009. “Safety and Efficacy of a
Steven Barnett, Clive R. Beldfield, and Milagros Nores. Single-Rod Etonogestrel Implant (Implanon): Results from
2005. Lifetime Effects: The High/Scope Perry Preschool Study 11 International Clinical Trials.” Fertility and Sterility 91
through Age 40. Ypsilanti, MI: High/Scope Press. (5): 1646–53.
Seger, Shane. 2012. “Researchers Explore Smartphone App to Dehlendorf, Christine, Kira Levy, Rachel Ruskin, and Jody
Combat Depression among First-Time, Low-Income Steinauer. 2010. “Health Care Providers’ Knowledge About
Mothers.” Yale School of Public Health, Yale University, Contraceptive Evidence: A Barrier to Quality Family
New Haven, CT. Planning Care?” Contraception 81 (4, April): 292–98.
Shah, Anuj, Mullainathan, Sendhil, and Eldar Shafir. 2012. “Some doi:10.1016/j.contraception.2009.11.006.
Consequences of Having Too Little.” Science 338 (6107): Edin, Kathryn, Paula England, Emily Fitzgibbons Shafer, and
682–85. Joanna Reed. 2007. “Forming Fragile Families: Was the
Shonkoff, Jack P. 2012. “Leveraging the Biology of Adversity Baby Planned, Unplanned, or In Between.” In Unmarried
to Address the Roots of Disparities in Health and Couples with Children, edited by Paula England and
Development.” Proceedings of the National Academy of Kathryn Edin, 25–54. New York: Russell Sage Foundation.
Sciences of the United States of America 109 (2): 17302–7. Espey, Eve, and Tony Ogburn. 2011. “Long-Acting Reversible
Stockwell, Melissa S., Elyse Olshen Kharbanda, Raquel Andres Contraceptives: Intrauterine Devices and the Contraceptive
Martinez, Celibell Y. Vargas, David K. Vawdrey, and Stewin Implant.” Obstetrics & Gynecology 117 (3): 705–19.
Camargo. 2012. “Effect of a Text Messaging Intervention on Evans, W. D., Olivia Silber-Ashley, and Jennifer Gard. 2007. “Social
Influenza Vaccination in an Urban, Low-Income Pediatric Marketing as a Strategy to Reduce Unintended Adolescent
and Adolescent Population: A Randomized Controlled Pregnancy.” The Open Communication Journal 1 (1,
Trial.” Journal of the American Medical Association 307 December 26): 1–8. doi:10.2174/1874916X00701010001.
(16): 1702–8. Farrelly, Matthew C., Kevin C. Davis, M. Lyndon Haviland, Peter
Thaler, Richard. 1991. “Some Empirical Evidence on Dynamic Messeri, and Cheryl G. Healton. 2005. “Evidence of a
Inconsistency.” In Quasi Rational Economics, edited Dose–Response Relationship between ‘Truth’ Antismoking
by Richard Thaler, 127–33. New York: Russell Sage Ads and Youth Smoking Prevalence.” American
Foundation. Journal of Public Health 95 (3): 425–31. doi:10.2105/
Thaler, Richard H., and Shlomo Benartzi. 2004. “Save More AJPH.2004.049692.
Tomorrow: Using Behavioral Economics to Increase Finer, Lawrence B., Jenna Jerman, and Megan L. Kavanaugh. 2012.
Employee Saving.” Journal of Political Economy 112: S164– “Changes in Use of Long-acting Contraceptive Methods in
S187. the United States, 2007–2009.” Fertility and Sterility 98 (4):
Villena-Rodán, Benjamín, and Cecilia Ríos-Aguilar. 2011. “Causal 893–97.
Effects of Maternal Time Investment on Children’s Frohwirth, Lori, Ann M. Moore, and Renata Maniaci. 2013.
Cognitive Outcomes.” Documentos de Trabajo 285, Centro “Perceptions of Susceptibility to Pregnancy among U.S.
de Economía Aplicada, Universidad de Chile, Santiago. Women Obtaining Abortions.” Social Science & Medicine
Wagner, Mary, Donna Spiker, and Margaret Inman Linn. 2002. 99 (December): 18–26.
“The Effectiveness of the Parents as Teachers Program Gold, Rachel Benson, Adam Sonfield, Cory L. Richards, and
with Low-Income Parents and Children.” Topics in Early Jennifer J. Frost. 2009. “Next Steps for America’s Family
Childhood Special Education 22 (2): 67–81. Planning Program: Leveraging the Potential of Medicaid
and Title X in an Evolving Health Care System.”
Guttmacher Institute, New York, NY.
The Hamilton Project • Brookings 175
References
Goldin, Claudia, and Lawrence F. Katz. 2002. “The Power of Monea, Emily, and Adam Thomas. 2011. “Unintended Pregnancy
the Pill: Oral Contraceptives and Women’s Career and and Taxpayer Spending.” Perspectives on Sexual and
Marriage Decisions.” Journal of Political Economy 110 (4, Reproductive Health 43 (2): 88–93.
August): 730–70. doi:10.1086/340778. Mosher, William D., and Jo Jones. 2010. “Use of Contraception in
Harper, Cynthia C., Maya Blum, Heike Thiel de Bocanegra, Philip the United States: 1982–2008.” Vital and Health Statistics,
D. Darney, J. Joseph Speidel, Michael Policar, and Eleanor Series 23, Data from the National Survey of Family Growth
A. Drey. 2008. “Challenges in Translating Evidence to (29) (August): 1–44.
Practice: The Provision of Intrauterine Contraception.” Ng, Alison Stewart, and Kelleen Kaye. 2012. “Teen Childbearing,
Obstetrics & Gynecology 111 (6, June): 1359–69. doi:10.1097/ Education, and Economic Well-Being. Why It Matters.”
AOG.0b013e318173fd83. National Campaign to Prevent Teen and Unplanned
Hotz, V. Joseph, Susan Williams McElroy, and Seth G. Sanders. Pregnancy, Washington, DC.
2005. “Teenage Childbearing and Its Life Cycle Peterson, Herbert B., and Kathryn M. Curtis. 2005. “Long-Acting
Consequences Exploiting a Natural Experiment.” Journal of Methods of Contraception.” New England Journal of
Human Resources 40 (3): 683–715. Medicine 353 (20): 2169–75.
Hubacher, David, Roger Lara-Ricalde, Douglas J. Taylor, Fernando Pederson, Sally. 2012. “Iowa Initiative to Reduce Unintended
Guerra-Infante, and Raymundo Guzmán-Rodríguez. 2001. Pregnancies.” Webinar presented at the Department of
“Use of Copper Intrauterine Devices and the Risk of Tubal Health and Human Services. www.hhs.gov/opa/pdfs/11-26-
Infertility among Nulligravid Women.” New England 12-iowa-initiative-transcript.pdf.
Journal of Medicine 345 (8): 561–67. Prevention First Colorado. 2009. “Develop Social Marketing and
Jones, Jo, William Mosher, and Kimberly Daniels. 2012. “Current Public Education Campaigns to Increase Contraceptive use
Contraceptive Use in the United States, 2006–2010, and and Knowledge of the Impacts of Unintended Pregnancy.”
Changes in Patterns of Use since 1995.” National Health Prevention First Colorado, Denver, CO.
Statistics Reports 60: 1–25. Sawhill, Isabel. 2012. “Are We Headed to a Permanently Divided
Karpilow, Quentin, Jennifer Manlove, Isabel V. Sawhill, and Society?” Policy Brief, Brookings Institution, Washington,
Adam Thomas. 2013. “The Role of Contraception in DC.
Preventing Abortion, Nonmarital Childbearing, and Child Sawhill, Isabel, and Joanna Venator. 2014. “Three Policies to Close
Poverty.” Paper presented at the Association for Public the Class Divide in Family Formation” Social Mobility
Policy Analysis and Management Annual Fall Research Memos (blog). http://www.brookings.edu/blogs/social-
Conference, Washington, DC. mobility-memos/posts/2014/01/21-3-policies-to-close-
Kaye, Kelleen, Katherine Suellentrop, and Corinna Sloup. 2009. family-formation-class-divide-sawhill.
“The Fog Zone: How Misperceptions, Magical Thinking, Sawhill, Isabel, Adam Thomas, and Emily Monea. 2010. “An
and Ambivalence Put Young Adults at Risk for Unplanned Ounce of Prevention: Policy Prescriptions to Reduce the
Pregnancy.” National Campaign to Prevent Teen and Prevalence of Fragile Families.” The Future of Children 20
Unplanned Pregnancy, Washington, DC. (2): 133–55. doi:10.1353/foc.2010.0004.
Kearney, Melissa S., and Phillip B. Levine. 2014. “Media Influences Secura, Gina M., Jenifer E. Allsworth, Tessa Madden, Jennifer
on Social Outcomes: The Impact of MTV’s 16 and Pregnant L. Mullersman, and Jeffrey F. Peipert. 2010. “The
on Teen Childbearing.” Working Paper 19795, National Contraceptive CHOICE Project: Reducing Barriers to
Bureau of Economic Research, Cambridge, MA. Long-acting Reversible Contraception.” American Journal
Lichter, Daniel T., and Deborah Roempke Graefe. 2001. “Finding of Obstetrics and Gynecology 203 (2, August): 115.e1–7.
a Mate? The Marital and Cohabitation Histories of Unwed doi:10.1016/j.ajog.2010.04.017.
Mothers.” In Out of Wedlock: Causes and Consequences Stoddard, Amy, Colleen McNicholas, and Jeffrey F. Peipert.
of Nonmarital Fertility, edited by Lawrence L. Wu and 2011. “Efficacy and Safety of Long-Acting Reversible
Barbara Wolfe, 317–43. New York: Russell Sage Foundation. Contraception.” Drugs 71 (8): 969–80.
Madden, Tessa, Jenifer E. Allsworth, Katherine J. Hladky, Gina Thomas, Adam. 2012. “Three Strategies to Prevent Unintended
M. Secura, and Jeffrey F. Peipert. 2010. “Intrauterine Pregnancy.” Journal of Policy Analysis and Management 31
Contraception in Saint Louis: A Survey of Obstetrician (2, March): 280–311. doi:10.1002/pam.21614.
and Gynecologists’ Knowledge and Attitudes.” Tolaymat, Lama L., and Andrew M. Kaunitz. 2007. “Long-Acting
Contraception 81 (2, February): 112–16. doi:10.1016/j. Contraceptives in Adolescents.” Current Opinion in
contraception.2009.08.002. Obstetrics and Gynecology 19 (5): 453–60.
McLanahan, Sara, and Gary Sandefur. 1994. Growing up with a U.S. Census Bureau. 2012. “Data on Families and Living
Single Parent: What Hurts, What Helps. Cambridge, MA: Arrangements.” Table FM-1. U.S. Census Bureau,
Harvard University Press. Washington, DC.
Monea, Emily, and Adam Thomas. 2010. “The Public Cost of U.S. Census Bureau. 2013. “Poverty - Experimental Measures.”
Pregnancy.” Working Paper, Brookings Institution, Table 6. U.S. Census Bureau, Washington, DC.
Washington, DC.
176 Policies to Address Poverty in America
University of California, San Francisco Bixby Center for Global Herrera, Carla, Jean Baldwin Grossman, Tina J. Kauh, Amy F.
Reproductive Health. 2014. LARC Project. http:// Feldman, Jennifer McMaken, and Linda Z. Jucovy. 2007.
bixbycenter.ucsf.edu/research/cd_and_fp/larc.html. “Making a Difference in Schools: The Big Brothers Big
Waldfogel, Jane, Terry-Ann Craigie, and Jeanne Brooks-Gunn. Sisters School-Based Mentoring Impact Study.” Public/
2010. “Fragile Families and Child Wellbeing.” The Future of Private Ventures, Philadelphia, PA.
Children 20 (2): 87–112. Karcher, Michael J. 2008. “The Study of Mentoring in the Learning
Weinreich, Nedra Kline. 1999. “The ‘Don’t Kid Yourself’ Campaign Environment (SMILE): A Randomized Evaluation of the
Case Study.” In Hands-On Social Marketing: A Step-by-Step Effectiveness of School-based Mentoring.” Prevention
Guide, edited by Nedra Kline Weinreich. Thousand Oaks, Science 9 (2): 99–113.
CA: Sage. Lawner, Elizabeth, Martha Beltz, and Kristin A. Moore. 2013.
Zolna, Mia R., and Laura Lindberg. 2012. “Unintended Pregnancy: “Research Brief: What Works for Mentoring Programs:
Incidence and Outcomes Among Young Adult Unmarried Lessons from Experimental Evaluations of Programs and
Women in the United States, 2001 and 2008.” Guttmacher Interventions.” Child Trends, Bethesda, MD.
Institute, New York, NY. Levine, Phillip B., and David J. Zimmerman. 2010. Targeting
Investments in Children: Fighting Poverty When Resources
Proposal 4 Are Limited. Chicago: University of Chicago Press.
Obama, Barack. 2014. “Remarks by the President on ‘My Brother’s
Behncke, Stefanie, Markus Frölich, and Michael Lechner. 2010. Keeper’ Initiative.” The White House, Washington, DC.
“A Caseworker Like Me: Does The Similarity between Schirm, Allen, Elizabeth Stuart, and Allison McKie. 2006. “The
the Unemployed and Their Caseworkers Increase Job Quantum Opportunity Program Demonstration: Final
Placements?” Economic Journal 120 (549):1430–59. Impacts.” Mathematica Policy Research, Washington, DC.
Bernstein, Lawrence, Catherine Dunn Rappaport, Lauren Olsho, Tierney, Joseph P., Jean Baldwin Grossman, and Nancy L. Resch.
Dana Hunt, and Marjorie Levin. 2009. “Impact Evaluation 1995. “Making a Difference: An Impact Study of Big
of the U.S. Department of Education’s Student Mentoring Brothers Big Sisters.” Public/Private Ventures, Philadelphia,
Program.” U.S. Department of Education, Washington, PA.
DC. U.S. Census Bureau. 2013. “Living Arrangements of Children.” U.S.
Big Brothers Big Sisters of America. 2012. “Annual Report 2011/12.” Census Bureau, Washington, DC. https://www.census.gov/
Philadelphia: Big Brothers Big Sisters of America. hhes/families/data/children.html.
Bruce, Mary, and John Bridgeland. 2014. “The Mentoring Wheeler, Marc E., Thomas E. Keller, and David L. DuBois. 2010.
Effect: Young People’s Perspectives on the Outcomes “Review of Three Recent Randomized Trials of School-
and Availability of Mentoring.” Civic Enterprises with Based Mentoring.” Social Policy Report 24 (3): 1–19.
Hart Research Associates for MENTOR: The National
Mentoring Partnership, Washington, DC.
Cavell, Timothy, David DuBois, Michael Karcher, Thomas Keller,
Proposal 5
and Jean Rhodes. 2009. “Policy Brief: Strengthening Bellotti, Jeanne, Linda Rosenberg, Samina Sattar, Andrea Mraz
Mentoring Opportunities for At-Risk Youth.” National Esposito, and Jessica Ziegler. 2010. “Reinvesting in
Mentoring Center, Portland, OR. America’s Youth: Lessons from the 2009 Recovery Act
Child Trends. 2014. “List of Programs.” Child Trends, Bethesda, Summer Youth Employment Initiative.” Mathematica
MD. http://www.childtrends.org/what-works/list-of- Policy Research, Washington, DC.
programs/. Bloom, Howard S., Larry L. Orr, Stephen H. Bell, George Cave,
Dee, Thomas S. 2004. “Teachers, Race and Student Achievement Fred Doolittle, Winston Lin, and Johannes M. Bos. 1997.
in a Randomized Experiment.” Review of Economics and “The Benefits and Costs of JTPA Title II-A Programs: Key
Statistics 86 (1): 195–210. Findings from the National Job Training Partnership Act
Dee, Thomas S. 2005. “A Teacher Like Me: Does Race, Ethnicity Study.” Journal of Human Resources 32 (3): 549–76.
or Gender Matter?” American Economic Review 95 (2): Bureau of Labor Statistics. n.d. “CPS Tables.” U.S. Bureau of Labor
158–65. Statistics. Washington, DC. http://www.bls.gov/cps/tables.
DuBois, David L., Nelson Portillo, Jean E. Rhodes, Naida htm.
Silverthorn, and Jeffrey C. Valentine. 2011. “How Effective Bureau of Labor Statistics. Various years. “Current Population
Are Mentoring Programs for Youth? A Systematic Survey: Employment Status of the Civilian
Assessment of the Evidence.” Psychological Science in the Noninstitutional Population by Age, Sex, and Race,
Public Interest 12 (2): 57–91. 2003–2012.” Bureau of Labor Statistics, U.S. Department of
Hahn, Andrew, Tom Leavitt, and Paul Aaron. 1994. “Evaluation of Labor, Washington, DC.
the Quantum Opportunities Program. Did the Program
Work?” Center for Human Resources, Brandeis University,
Waltham, MA.
The Hamilton Project • Brookings 177
References
Castleman, Benjamin L., Karen Arnold, and Katherine Lynk McClanahan, Wendy S., Cynthia L. Sipe, and Thomas J. Smith.
Wartman. 2012. “Stemming the Tide of Summer Melt: 2004. “Enriching Summer Work: An Evaluation of the
An Experimental Study of the Effects of Post-High Summer Career Exploration Program.” Public/Private
School Summer Intervention on Low-Income Students’ Ventures, New York, NY.
College Enrollment.” Journal of Research on Educational McCollister, Kathryn E., Michael T. French, and Hai Fang. 2010.
Effectiveness 5 (1): 1–17. “The Cost of Crime to Society: New Crime-Specific
Curnan, Susan P., and Andrew B. Hahn. 2010. “Innovating Under Estimates for Policy and Program Evaluation.” Drug
Pressure: The Story of the 2009 Recovery Act Summer Alcohol Dependence 108 (1–2): 98–109.
Youth Employment Initiative.” Center for Youth and Morisi, Teresa L. 2010. “The Early 2000s: A Period of Declining
Communities, Brandeis University, Waltham, MA. Teen Summer Employment Rates.” Monthly Labor
Dee, Thomas S. 2011. “Conditional Cash Penalties in Education: Review, Bureau of Labor Statistics, Department of Labor,
Evidence from the Learnfare Experiment.” Economics of Washington, DC.
Education Review 30 (5): 924–37. Mortimer, Jeylan T. 2003. Working and Growing Up in America.
Deming, David J., Sarah Cohodes, Jennifer Jennings, and Cambridge, MA: Harvard University Press.
Christopher Jencks. 2013. “School Accountability, New York University Institute for Education and Social Policy.
Postsecondary Attainment and Earnings.” Working Paper 2014. Steinhardt School of Culture, Education, and Human
19444, National Bureau of Economic Research, Cambridge, Development, New York University, New York, NY.
MA. http://steinhardt.nyu.edu/iesp/.
Doolittle, Fred, Steve Bell, Howard Bloom, George Cave, James Orr, Larry L., Howard S. Bloom, Stephen H. Bell, Fred
Kemple, Larry Orr, Linda Traeger, and John Wallace. 1993. Doolittle, and Winston Lin. 1996. Does Training for the
“A Summary of the Design and Implementation of the Disadvantaged Work? Evidence from the National JTPA
National JTPA Study.” MDRC, New York, NY. Study. Washington, DC: Urban Institute Press.
Entwisle, Doris R., Karl L. Alexander, and Linda Steffel Olson. Painter, Matthew A. 2010. “Get a Job and Keep It! High school
2000. “Early Work Histories of Urban Youth.” American Employment and Adult Wealth Accumulation.” Research in
Sociological Review 65 (2). Social Stratification and Mobility 28 (2): 233–49.
Farkas, George, D. Alton Smith, and Ernst W. Stromsdorfer. Rees, Albert. 1986. “An Essay on Youth Joblessness.” Journal of
1983. “The Youth Entitlement Demonstration: Subsidized Economic Literature 24 (1): 613–28.
Employment with a School Requirement.” Journal of Riccio, James A., Nadine Dechausay, David M. Greenberg, Cynthia
Human Resources 28 (4) 557–73. Miller, Zawadi Rucks, and Nandita Verma. 2010. “Toward
Foster, E. Michael. 1995. “Why Teens Do Not Benefit from Reduced Poverty across Generations: Early Findings from
Work Experience Programs: Evidence from Brother New York City’s Conditional Cash Transfer Program.”
Comparisons.” Journal of Policy Analysis and Management MDRC, New York, NY.
14 (3): 393–414. Riccio, James A., Nadine Dechausay, Cynthia Miller, Stephen
Greenberg, David H., Charles Michalopoulus, and Philip K. Nuñez, Nandita Verma, Edith Yang. 2013. “Conditional
Robins. 2003. “A Meta-Analysis of Government-Sponsored Cash Transfers in New York City: The Continuing Story of
Training Programs.” Industrial and Labor Relations Review the Opportunity NYC-Family Rewards Demonstration.”
57 (1): 31–53. MDRC, New York, NY.
Gueron, Judith M. 1984. “Lessons from a Job Guarantee: The Youth Rose, Heather. 2005. “Do Gains in Test Scores Explain Labor
Incentive Entitlement Pilot Projects.” MDRC, New York, Market Outcomes?” Economics of Education Review 25
NY. (2006): 430–46.
Heckman, James. 1998. “What Should Be Our Human Capital Ruhm, Christopher J. 1995. “The Extent and Consequences of High
Investment Policy?” Fiscal Studies 19 (2): 103–19. School Employment.” Journal of Labor Research 16 (3):
Heller, Sara. 2014. “The Effects of Summer Jobs on Disadvantaged 293–303.
Youth.” Unpublished manuscript, University of Schwartz, Amy E., Jacob Leos-Urbel, Megan Silander, and Matthew
Pennsylvania, Philadelphia, PA. Wiswall. In progress. “Making Summer Matter. The Impact
Hirsch, Barton J., Larry V. Hedges, JulieAnn Stawicki, and Megan of Youth Employment on Academic Performance.”
A. Mekinda. 2011. “After-School Programs for High Stanley, Marcus, Lawrence Katz, and Alan Krueger. 1998.
School Students: An Evaluation of After School Matters.” “Developing Skills: What We Know about the Impacts
Manuscript, Northwestern University, Evanston, IL. of American Employment and Training Programs on
Leos-Urbel, Jacob. Forthcoming. “What Is a Summer Job Worth? Employment, Earnings, and Educational Outcomes.” Paper
The Causal Impact of Summer Youth Employment on presented at G8 Economic Summit, Birmingham, UK.
Academic Outcomes: Evidence from a Large-Scale Lottery.”
Journal of Policy Analysis and Management.
Lillydahl, Jane H. 1990. “Academic Achievement and Part-Time
Employment of High School Students.” Journal of Economic
Education 21 (3): 307–16.
178 Policies to Address Poverty in America
Sum, Andrew, Ishwar Khatiwada, Joseph McLaughlin, and Sheila Bettinger, Eric, Angela Boatman, and Bridget Terry Long. 2013.
Palma. 2008. “The Collapse of the National Teen Job “Student Supports: Developmental Education and Other
Market and the Case for an Immediate Summer and Year Academic Programs.” In Cecilia Rouse, Lisa Barrow, and
Round Youth Jobs Creation Program.” Center for Labor Thomas Brock, Eds., Future of Children: Postsecondary
Market Studies Publications, Center for Labor Market Education in the U.S. 23 (1, Spring).
Studies, Northeastern University, Boston, MA. Bettinger, Eric, and Bridget Terry Long. 2005. “Remediation at the
Sum, Andrew, Mykhaylo Trubskyy, and Walter McHugh. 2013. Community College: Student Participation and Outcomes.”
“The Summer Employment Experiences and the Personal/ New Directions for Community Colleges 129: 17–26.
Social Behaviors of Youth Violence Prevention Employment ———. 2007. “Institutional Responses to Reduce Inequalities in
Program Participants and Those of a Comparison Group College Outcomes: Remedial and Developmental Courses
Center for Labor Market Studies.” Center for Labor Market in Higher Education.” In Economic Inequality and Higher
Studies, Northeastern University, Boston, MA. Education: Access, Persistence, and Success, edited by Stacy
U.S. Census Bureau. 2000. “Table 2-32: New York: Total Voting-Age Dickert-Conlin and Ross Rubenstein, 69–100. New York:
Population and Citizen Voting-Age Population by Sex, for Russell Sage Foundation.
Counties: 2000.” 2005–2009 American Community Survey 5 ———. 2009. “Addressing the Needs of Under-Prepared College
year estimates. U.S. Census Bureau, Washington, DC. Students: Does College Remediation Work?” Journal of
Walker, Gary, and Frances Vilella-Velez. 1992. “Anatomy of a Human Resources 44 (3, Summer).
Demonstration: The Summer Training and Education Boatman, Angela. 2012. “Evaluating Institutional Efforts to
Program (STEP) from Pilot through Replication and Post- Streamline Postsecondary Remediation: The Causal Effects
Program Impacts.” Public/Private Ventures, Philadelphia, of the Tennessee Developmental-Course Redesign Initiative
PA. on Early Student Academic Success.” Doctoral dissertation,
Weinstein, Meryle, and Michele C. Leardo. 2013. “CTE Summer Harvard Graduate School of Education, Cambridge, MA.
Scholars Program: Year 2 Evaluation Report.” IESP Boatman, Angela. 2014. “Beyond Ready, Fire, Aim: New Solutions
Evaluation Report, IESP, New York University, New York, to Old Problems in College Remediation.” American
NY. Enterprise Institute, Washington, DC.
Boatman, Angela, and Bridget Terry Long. 2010. “Does
Proposal 6 Remediation Work for All Students? How the Effects of
Postsecondary Remedial and Developmental Courses
Adelman, Clifford. 1999. “Answers in the Toolbox: Academic Vary by Level of Academic Preparation.” Working Paper,
Intensity, Attendance Patterns, and Bachelor’s Degree National Center for Postsecondary Research, Teachers
Attainment.” U.S. Department of Education, Washington, College, Columbia University, New York, NY.
DC. Calcagno, Juan Carlos, and Bridget Terry Long. 2008. “The
Adelman, Clifford. 2006. “The Toolbox Revisited: Paths to Degree Impact of Postsecondary Remediation Using a Regression
Completion from High School through College.” U.S. Discontinuity Approach: Addressing Endogenous Sorting
Department of Education, Washington, DC. and Noncompliance.” Working Paper 14194, National
Alliance for Excellent Education. 2006. “Paying Double: Inadequate Bureau of Economic Research, Cambridge, MA.
High Schools and Community College Remediation.” Issue Cho, Sung-Woo, Elizabeth Kopko, Davis Jenkins, and Shanna
Brief, Alliance for Excellent Education, Washington, DC. Smith Jaggars. 2012. “New Evidence of Success for
Attewell, Paul, and Thurston Domina. 2008. “Raising the Bar: Community College Remedial English Students: Tracking
Curricular Intensity and Academic Performance.” the Outcomes of Students in the Accelerated Learning
Educational Evaluation and Policy Analysis 30: 51–71. Program (ALP).” Community College Research Center,
Bailey, Thomas. 2009. “Challenge and Opportunity: Rethinking Teachers College, Columbia University, New York, NY.
the Role and Function of Developmental Education in Cloud, John. 2002. “Who’s Ready for College?” Time, October 14.
Community College.” New Directions for Community Complete College America. 2012. “Remediation: Higher
Colleges 145: 11–30. Education’s Bridge to Nowhere.” Complete College
Bailey, Thomas, Dong Wook Jeong, and Sung-Woo Cho. 2010. America, Washington, DC.
“Referral, Enrollment, and Completion in Developmental Edgecombe, Nikki. 2011. “Accelerating the Academic Achievement
Education Sequences in Community Colleges.” Economics of Students Referred to Developmental Education.”
of Education Review 29 (2): 255–70. Working Paper No. 30, Assessment of Evidence Series,
Belfield, Clive, and Peter Crosta. 2012. “Predicting Success in Community College Research Center, Teachers College,
College: The Importance of Placement Tests and High Columbia University, New York, NY.
School Transcripts.” Working Paper No. 42, Community Edgecombe, Nikki, Di Xu, Melissa, Barragan, and Shanna Smith
College Research Center, Teachers College, Columbia Jaggars. 2012. “Analysis of Chabot College’s Accelerated
University, New York, NY. Developmental English course.” Community College
Research Center, Teachers College, Columbia University,
New York, NY. Manuscript in preparation.
The Hamilton Project • Brookings 179
References
Education Commission of the States. 2012. “Instructional Delivery.” Long, Mark C., Dylan Conger, and Patrice Iatarola. 2012. “Effects
http://gettingpastgo.org/policy-levers/instructional- of High School Course-Taking on Secondary and
delivery/. Postsecondary Success.” American Education Research
Epper, Rhonda M., and Elaine Baker. 2009. “Technology Solutions Journal 49 (2): 285–322.
for Developmental Math: An Overview of Current and McCabe, Robert H. 2001. “Developmental Education: A Policy
Emerging Practices.” Report prepared for the William and Primer.” League for Innovation in the Community College
Flora Hewlett Foundation and the Bill and Melinda Gates 14 (1): 1–4.
Foundation. Martorell, Paco, and Isaac McFarlin Jr. 2011. “Help or Hindrance?
Georgia Board of Regents. 2010. “Board of Regents Policy Manual.” The Effects of College Remediation on Academic and Labor
Section 4.2.1.1: Freshman Admission Requirements. Market Outcomes.” Review of Economics and Statistics 93
University System of Georgia, Atlanta, GA. (2): 436–54.
Greene, Jay, and Greg Foster. 2003. “Public High School Graduation National Bureau of Economic Research (NBER). n.d. “Current
and College Readiness Rates in the United States.” Working Population Survey (CPS) Data at the NBER.” National
Paper No. 3, Manhattan Institute, New York, NY. Bureau of Economic Research, Cambridge, MA.
Howell, Jessica S., Michal Kurlaender, and Eric Grodsky. National Center for Education Statistics (NCES). 2003. “Remedial
2010. “Postsecondary Preparation and Remediation: Education at Degree-Granting Postsecondary Institutions
Examining the Effect of the Early Assessment Program at in Fall 2000.” U.S. Department of Education, Washington,
California State University.” Journal of Policy Analysis and DC.
Management 29 (4): 726–48. National Center for Education Statistics (NCES). 2010. “Academic
Hughes, Katherine L., and Judith Scott-Clayton. 2010. “Assessing Preparation for College in the High School Senior Class of
Developmental Assessment in Community Colleges: 2003-04: Education Longitudinal Study of 2002 (ELS:2002),
A Review of the Literature.” Working Paper No. 19, Base-year, 2002, First Follow-up, 2004, and High School
Community College Research Center, Teachers College, Transcript Study, 2004.” U.S. Department of Education,
Columbia University, New York, NY. Washington, DC.
Jenkins, Davis, Matthew Zeidenberg, and Gregory S. Kienzl. 2009. Ohio Board of Regents. 2002. “Making the transition from high
“Building Bridges to Postsecondary Training for Low-Skill school to college in Ohio 2002.” Columbus, OH: Ohio
Adults: Outcomes of Washington State’s I-BEST program.” Board of Regents.
Brief No. 42, Community College Research Center, Parsad, Basmat, Laurie Lewis, and Bernard Greene. 2003.
Teachers College, Columbia University, New York, NY. “Remedial Education at Degree-Granting Postsecondary
Logue, Alexandra W., and John Mogulescy. 2013. “CUNY Start: Institutions in Fall 2000.” Report No. NCES 2004-101,
Analysis of Student Outcomes.” Office of Academic Affairs, National Center for Education Statistics, U.S. Department
City University of New York, New York, NY. of Education, Washington, DC.
Long, Bridget Terry. 2012. “Remediation: The Challenge of Helping Quint, Janet, Shanna S. Jaggars, D. Crystal Byndloss, Asya
Underprepared Students.” In Getting to Graduation: The Magazinnik. 2013. “Bringing Developmental Education
Completion Agenda in Higher Education, edited by Andrew to Scale: Lessons from the Developmental Education
P. Kelly and Mark Schneider. Baltimore: The Johns Hopkins Initiative.” MDRC, New York, NY.
University Press. Reed, Charles B. 2010. “The Early Start Program Executive Order
Long, Bridget Terry, and Angela Boatman. 2013. “The Role of No. 1048.” Memorandum, June 11. http://calstate.edu/eo/
Remediation and Developmental Courses in Access EO-1048.html.
and Persistence.” In The State of College Access and Scott-Clayton, Judith. 2012. “Do High Stakes Placement Exams
Completion: Improving College Success for Students from Predict College Success?” Working Paper No. 41,
Underrepresented Groups, edited by Anthony Jones and Community College Research Center, Teachers College,
Laura Perna, 77–95. New York: Routledge Books. Columbia University, New York, NY.
Long, Bridget Terry, and Michal Kurlaender. 2009. “Do Community Scott-Clayton, Judith, Peter M. Crosta, and Clive R. Belfield. 2012.
Colleges Provide a Viable Pathway to a Baccalaureate “Improving the Targeting of Treatment: Evidence from
Degree?” Educational Evaluation and Policy Analysis 31 (1): College Remediation.” Working Paper 18457, National
30–53. Bureau of Economic Research, Cambridge, MA.
Long, Bridget Terry, and Erin K. Riley. 2007. “Sending Signals to Scott-Clayton, Judith, and Olga Rodriguez. 2012. “Development,
Students: The Role of Early Placement Testing in Improving Discouragement, or Diversion? New Evidence on the
Academic Preparation.” In Minding the Gap: Why Effects of College Remediation.” Working Paper 18328,
Integrating High School with College Makes Sense and How National Bureau of Economic Research, Cambridge, MA.
to Do It, edited by Nancy Hoffman, Joel Vargas, Andrea Scrivener, Susan, and Michael Weiss. 2013. “Two-Year Results from
Venezia, and Marc S. Miller, 105–12. Cambridge: Harvard an Evaluation of Accelerated Study in Associate Programs
Education Press and Jobs for the Future. (ASAP) for Developmental Education Students.” MDRC,
New York, NY.
180 Policies to Address Poverty in America
Smith, Matthew. 2012. “Choosing who Delivers: The Impact of National Bureau of Economic Research. n.d. “NBER CPS
Placing Limits on the Delivery of Remedial Education at Supplements.” National Bureau of Economic
Four-Year Institutions.” Getting Past Go Project, Education Research, Cambridge, MA. http://www.nber.org/data/
Commission of the States, Denver, CO. currentpopulation-survey-data.html.
Sommo, Colleen, Alexander K. Mayer, Timothy Rudd, and Dan National Center for Education Statistics. 2008. “Recent
Cullinan. 2012. “Commencement Day: Six Year Effects of a Participation in Formal Learning Among Working-Age
Freshman Learning Community Program at Kingsborough Adults With Different Levels of Education.” National
Community College.” MDRC, New York, NY. Center for Education Statistics, U.S. Department of
Strong American Schools. 2008. “Diploma to nowhere.” Strong Education, Washington, DC.
American Schools, Washington, DC. National Center for Education Statistics. 2013. “Table 322.20:
Twigg, Carol. 2011. “The Math Emporium: Higher Education’s Advance Release of Selected 2013 Digest Tables.” Digest
Silver Bullet.” Change: The Magazine of Higher Learning of Education Statistics, National Center for Education
(May–June): 25–34. Statistics, Washington, DC.
Venezia, Andrea, Michael Kirst, and Anthony Antonio. 2003. Olinsky, Ben, and Sarah Ayres. 2013. “Training for Success: A
“Betraying the College Dream: How Disconnected K–12 Policy to Expand Apprenticeships in the United States.”
and Postsecondary Education Systems Undermine Student Center for American Progress, Washington, DC.
Aspirations.” Stanford Institute for Higher Education Reed, Debbie, Albert Yung-Hsu Liu, Rebecca Kleinman, Annalisa
Research, Stanford, CA. Mastri, Davin Reed, Samina Sattar, and Jessica Ziegler.
Zachry, Elizabeth, and Emily Schneider. 2011. “Unlocking the 2012. “An Effectiveness Assessment and Cost–Benefit
Gate: What We Know About Improving Developmental Analysis of Registered Apprenticeship in 10 States.”
Education.” MDRC, New York, NY. Mathematica Policy Research, Washington, DC.
Skills Funding Agency. 2014. “Breakdown by Geography, Equality
Proposal 7 & Diversity and Sector Subject Area: Starts 2002/03
to 2012/13.” Skills Funding Agency, Department for
Bauernschuster, Stefan, Olivier Falck, and Stephan Heblich. 2009. Business Innovation & Skills, London, UK. https://www.
“Training and Innovation.” Journal of Human Capital 3 (4): gov.uk/government/statistical-data-sets/fe-data-library-
323–53. apprenticeships--2.
Cellini, Stephanie. 2012. “For-Profit Higher Education: An Workforce Training and Education Coordinating Board. 2014.
Assessment of Costs and Benefits.” National Tax Journal 65 “2014 Workforce Training Results by Program.” Olympia,
(1): 153–80. WA. http://wtb.wa.gov/WorkforceTrainingResults.asp.
Evans, Stephen, and Gerhard Bosch. 2012. “Apprenticeships in
London: Boosting Skills in a City Economy with Comment
on Lessons from Germany.” Organisation for Economic
Proposal 8
Co-operation and Development, Paris, France. Ahlstrand, Amanda, Laurie Bassi, and Daniel McMurrer. 2003.
Halpern, Robert. 2009. The Means to Grow Up: Reinventing Workplace Education for Low-Wage Workers. Kalamazoo:
Apprenticeship as a Developmental Support in Adolescence. W. E. Upjohn Institute for Employment Research.
New York: Routledge. Bailey, Martha, and Susan Dynarski. 2011. “Inequality in
Hollenbeck, Kevin. 2008. “State Use of Workforce System Net Postsecondary Education.” In Whither Opportunity?: Rising
Impact Estimates and Rates of Return.” Paper presented at Inequality, Schools, and Children’s Life Chances, edited
the Association for Public Policy Analysis and Management by Greg J. Duncan and Richard J. Murnane, 117–32. New
(APPAM) Conference, Los Angeles, CA. York: Russell Sage Foundation and Spencer Foundation.
Lerman, Robert I. 2013. “Expanding Apprenticeship in the United Barnow, Burt, John Trutko, and Jaclyn Schede Piatak. 2013.
States: Barriers and Opportunities.” In Contemporary Occupational Labor Shortages: Concepts, Causes,
Apprenticeship: International Perspectives on an Evolving Consequences and Cures. Kalamazoo: W. E. Upjohn
Model of Learning, edited by Alison Fuller and Lorna Institute for Employment Research.
Unwin. New York: Routledge. Bartik, Timothy. 2010. “Not All Job Creation Tax Credits Are
Lerman, Robert I. 2014. “Do Firms Benefit from Apprenticeship?” Created Equal.” Economic Policy Institute, Washington,
IZA World of Labor. http://wol.iza.org/articles/do-firms- DC.
benefit-from-apprenticeship-investments. Becker, Gary. 1996. Human Capital (2nd ed.). Chicago: University
Lerman, Robert I., Lauren Eyster, and Kate Chambers. 2009. “The of Chicago Press.
Benefits and Challenges of Registered Apprenticeship: Bettinger, Eric, Angela Boatman, and Bridget Terry Long. 2013.
The Sponsors’ Perspective.” The Urban Institute Center on “Student Supports: Developmental Education and Other
Labor, Human Services, and Population, Washington, DC. Academic Programs.” The Future of Children 23 (1): 93–115.
The Hamilton Project • Brookings 181
References
Bound, John, Michael Lovenheim, and Sarah Turner. 2009. “Why Hollenbeck, Kevin. 2008. “Is There a Role for Public Support of
Have College Completion Rates Declined? An Analysis of Incumbent Worker On-the-Job Training?” W. E. Upjohn
Changing Student Preparation and Collegiate Resources.” Institute for Employment Research, Kalamazoo, MI.
Working Paper 15566, National Bureau of Economic Hollenbeck, Kevin. 2012. “Returns on Investment in Workforce
Research, Cambridge, MA. Development Programs.” Working Paper, W.E. Upjohn
Busso, Matias, Jesse Gregory, and Patrick Cline. 2013. “Results of Institute for Employment Research, Kalamazoo, MI.
the Federal Empowerment Zone Program.” Kalamazoo: Holzer, Harry, and Erin Dunlop. 2013. “Just the Facts Ma’am:
Upjohn Institute for Employment Research. Postsecondary Education and Labor Market Outcomes in
Choitz, Vicky. 2013. “A Framework for Measuring Career Pathways the U.S.” CALDER Working Paper, American Institutes for
Innovation: A Working Paper.” Center for Law and Social Research, Washington, DC.
Policy, Washington, DC. Holzer, Harry. 2013. “Workforce Development Programs” In
College Board. 2013. “Rethinking Pell Grants.” Rethinking Pell Legacies of the War on Poverty, edited by Martha J. Bailey
Grants Study Group, College Board Advocacy and Policy and Sheldon Danziger. New York: Russell Sage Foundation.
Center, New York, NY. Holzer, Harry, Richard Block, Marcus Cheatham, and Jack Knott.
Davidson, Carl, and Stephen A. Woodbury. 1990. “The 1993. “Are Training Subsidies for Firms Effective? The
Displacement Effect of Reemployment Bonus Programs.” Michigan Experience.” Industrial and Labor Relations
Working Paper No. 90-02, W.E. Upjohn Institute for Review 46 (4): 625–636.
Employment Research, Kalamazoo, MI. Holzer, Harry, and Robert Lerman. Forthcoming. “Work-Based
Deming, David, Claudia Goldin, and Lawrence Katz. 2013. “For Learning to Expand Jobs and Occupational Qualifications
Profit Colleges.” The Future of Children 23 (1): 137–63. for Youth.” Challenge 37 (4): 3–16.
Faulk, Dagney. 2002. “Do State Economic Development Incentives Jacobson, Louis, and Christine Mokher. 2009. “Pathways
Create Jobs? An Analysis of State Employment Tax to Boosting the Earnings of Low-Income Students
Credits.” National Tax Journal 55 (2): 263–80. by Increasing their Educational Attainment.” CNA
Greenstone, Michael, and Adam Looney. 2011. “Where is the Best Corporation, Washington, DC.
Place to Invest $102,000—In Stocks, Bonds, or a College Jacobson, Louis, Robert LaLonde, and Daniel G. Sullivan. 2005.
Degree?” The Hamilton Project, Brookings Institution, “Estimating The Returns To Community College Schooling
Washington, DC. For Displaced Workers.” Journal of Econometrics 125 (1–2):
Greenstone, Michael, Adam Looney, Jeremy Patashnik, and 271–304.
Muxin Yu. 2013. “Thirteen Economic Facts about Social Lerman, Robert, Signe-Mary McKernan, and Stephanie Riegg.
Mobility and the Role of Education.” The Hamilton Project, 2004. “The Scope of Employer-Provided Training in the
Brookings Institution, Washington, DC. United States: Who, What, Where and How Much.” In Job
Ham, John, Ayşe İmrohoroğlu, and Charles Swenson. 2010. Training Policy in the United States, edited by Christopher
“Government Programs Can Improve Local Labor J. O’Leary, Robert A. Straits, and Stephen A. Wandner.
Markets: Evidence from State Enterprise Zones, Kalamazoo: W. E. Upjohn Institute for Employment
Federal Empowerment Zones and Federal Enterprise Research.
Communities.” University of Southern California, Los Maguire, Sheila, Joshua Freely, Carol Clymer, Maureen Conway,
Angeles, CA. and Deena Schwartz. 2010. “Tuning In to Local Labor
Hamersma, Sarah. 2014. “Employment Subsidies to Firms and Markets: Findings from the Sectoral Employment Study.”
Workers: Key Distinctions Between the Effects of the Public/Private Ventures, Philadelphia, PA.
Work Opportunity Tax Credit and the Earned Income Tax Martin, Vanessa, and Joseph Broadus. 2013. “Enhancing GED
Credit.” In What Works for Workers?, edited by Stephanie Instruction to Prepare Students for College and Careers.”
Luce, Jennifer Luff, Joseph A. McCartin, and Ruth Policy Brief, MDRC, New York, NY.
Milkman, 186–214. New York: Russell Sage Foundation. National Conference of State Legislatures. 2014. “Performance-
Heckman, James J., Carolyn J. Heinrich, and Jeffrey Smith. 2011. Based Funding for Higher Education.” National Conference
“Do Short-Run Performance Measures Predict Long-Run of State Legislatures, Washington, DC.
Impacts?” In The Performance of Performance Standards, National Governors Association. 2013. “State Sector Strategies
edited by James J. Heckman, Carolyn J. Heinrich, Pascal Coming of Age: Implications for State Workforce
Courty, Gerald Marschke, and Jeffrey Smith, 273–304. Policymakers.” Center for Best Practices, National
Kalamazoo: W.E. Upjohn Institute for Employment Governors Association, Washington, DC.
Research. National Student Clearinghouse Research Center. 2011. “National
Heckman, James, Robert Lalonde, and Jeffrey Smith. 1999. “The Post-Secondary Enrollment Trends: Before, During
Economics and Econometrics of Active Labor Market and After the Great Recession.” National Student
Policies.” In Handbook of Labor Economics, edited by Orley Clearinghouse, Herndon, VA.
Ashenfelter and David Card, volume III. Amsterdam: Office of Management and Budget. 2014. “Fiscal Year 2015 Budget
Elsevier Science B.V. of the U.S. Government.” Office of Management and
Budget, Executive Office of the President, Washington, DC.
182 Policies to Address Poverty in America
Reardon, Sean. 2011. “The Widening Academic Achievement D’Amico, Ronald, Kate Dunham, Jennifer Henderson-Frakes,
Gap between the Rich and the Poor: New Evidence and Deborah Kogan, Vinz Koller, Melissa Mack, Micheline
Possible Explanations.” In Whither Opportunity?: Rising Magnotta, Jeffrey Salzman, Andrew Wiegand, Dan
Inequality, Schools, and Children’s Life Chances, edited by Weissbein, and Gardner Carrick. 2004. “The Workforce
Greg J. Duncan and Richard J. Murnane, 91–116. New York: Investment Act After Five Years: Results from the National
Russell Sage Foundation and Spencer Foundation. Evaluation of the Implementation of WIA.” Employment
Reed, Debbie, Albert Yung-Hsu Liu, Rebecca Kleinman, Annalisa and Training Occasional Paper 2004-05, U.S. Department
Mastri, Davin Reed, Samina Sattar, and Jessica Ziegler. of Labor, Washington, DC.
2012. “An Effectiveness Assessment and Cost–Benefit Fein, David J. 2012. “Career Pathways as a Framework for Program
Analysis of Registered Apprenticeship in 10 States.” Design and Evaluation: A Working Paper from The
Mathematica Policy Research, Washington, DC. Innovative Strategies for Increasing Self-Sufficiency (ISIS)
Robst, John. 2007. “Education and Job Match: The Relatedness of Project.” Report 32012-30, Office of Planning, Research and
College Major and Work.” Economics of Education Review Evaluation, Administration for Children and Families, U.S.
26 (4): 397–407. Department of Health and Human Services, Washington,
Roder, Anne, and Mark Elliott. 2012. “A Promising Start: Year Up’s DC.
Initial Impact on the Earnings of Young Adults.” Economic Goldin, Claudia, and Lawrence F. Katz. 2012. “The Future of
Mobility Coroporation, New York, NY. Inequality. The Other Reason Education Matters So Much.”
Rosenbaum, James, Regina Deil-Amen, and Ann E. Person. 2006. The Milken Institute Review (July 2009): 26–33.
After Admission: From College Access to College Success. Goldrick-Rab, Sara, and Kia Sorenson. 2010. “Unmarried Parents in
New York: Russell Sage Foundation. College.” Fragile Families 20 (2): 179–203.
Scott-Clayton, Judith. 2011. “The Shapeless River: Does a Lack Heinrich, Carolyn. 2013. “Targeting Workforce Development
of Structure Inhibit Student Progress at Community Programs: Who Should Receive What Services? And
Colleges?” Working Paper, Community College Research How Much?” Paper prepared for Workforce Development
Center, Teachers College, Columbia University, New York, Needs in the Global Economy: US and Cross-National
NY. Perspectives, School of Public Policy, University of
Webber, Douglas A., and Ronald G. Ehrenberg. 2010. “Do Maryland, College Park, MD
Expenditures Other Than Instructional Expenditures Heinrich, Carolyn, Peter Mueser, Kenneth Troske, Kyung
Affect Graduation and Persistence Rates in American Seong-Jeon, and Daver C. Kahveciogly. 2013. “Do
Higher Education?” Economics of Education Review 29 Public Employment and Training Programs Work?
(6): 947–58. Nonexperimental Net Impact Estimates of the Workforce
Zeidenberg, Matthew, Sung-Woo Cho, and Davis Jenkins. 2010. Investment Act Program.” IZA Journal of Labor Economics
“Washington State’s I-BEST Program: New Evidence 2 (6): 1–23.
of Effectiveness.” Working Paper, Community College Hollenbeck, Kevin. 2009. “Workforce Investment Act (WIA) Net
Research Center, Columbia University, New York, NY. Impact Estimates and Rates of Return.” W. E. Upjohn
Zinn, Rachel, and Andy Van Kluenen. 2014. “Making Workforce Institute for Employment Research, Kalamazoo, MI.
Data Work: How Improved Education and Workforce Data Holzer, Harry, and Erin Dunlop. 2013. “Just the Facts, Ma’am:
Systems Could Help the U.S. Compete in the 21st Century Postsecondary Education and Labor Market Outcomes
Economy.” Workforce Data Quality Campaign, National in the U.S.” IZA Discussion Paper Series, No. 7319, Bonn,
Skills Coalition, Washington, DC. Germany.
Holzer, Harry, Julia Lane, David Rosenblum, and Fredrik
Proposal 9 Andersson. 2011. Where Are All the Good Jobs Going? What
National and Local Job Quality and Dynamics Mean for
Berk, Jillian. 2012. “Understanding the Employment Outcomes U.S. Workers. New York: Russell Sage Foundation.
of Trainees in the Trade Adjustment Assistance (TAA) Horn, Laura, and Thomas Weko. 2009. “On Track To Complete?
Program under the 2002 TAA Program.” Submitted to the A Taxonomy of Beginning Community College Students
U.S. Department of Labor by Mathematica Policy Research, and Their Outcomes Three Years After Enrolling: 2003–04
Princeton, NJ. Through 2006.” National Center for Education Statistics,
Bloom, Howard, Larry Orr, George Cave, Steve Bell, and Fred Washington, DC.
Doolittle. 1993. “The National JTPA Study: Title II-A Jacobson, Louis, and Robert LaLonde. 2013. “Using Data to
Impacts on Earnings and Employment at 18 Months.” Abt Improve the Performance of Workforce Training.” The
Associates, Bethesda, MD. Hamilton Project, Brookings Institution, Washington, DC.
Card, David, Jochen Kluve, and Andrea Weber. 2010. “Active Jacobson, Louis, Robert LaLonde, and Daniel G. Sullivan. 2005.
Labour Market Policy Evaluations: A Meta-Analysis.” “Estimating the Returns to Community College Schooling
Economic Journal 120 (548): F452–F477. for Displaced Workers.” Journal of Econometrics 125 (1–2):
271–304.
The Hamilton Project • Brookings 183
References
Johnson, Terry, Mark Gritz, Russell Jackson, John Burghardt, Carol Schochet, Peter Z., Ronald D’Amico, Jillian Berk, Sarah Dolfin, and
Boussy, Jan Leonard, and Carlyn Orians. 1999. “National Nathan Wozny. 2012. “Estimated Impacts for Participants
Job Corps Study: Report on the Process Analysis.” Battelle in the Trade Adjustment Assistance (TAA) Program
Memorial Institute, Seattle, WA. Under the 2002 Amendments” and “Methodological
Kemple, James, Fred Doolittle, and John Wallace. 1993. “The Notes on the Impact Analysis.” Final reports submitted
National JTPA Study: Final Implementation Report.” to the U.S. Department of Labor, Employment and
MDRC, New York, NY. Training Administration by Mathematica Policy Research,
Laird, Elizabeth, and Pamela Holcomb. 2011. “Effective Case Princeton, NJ.
Management: Key Elements and Practices from the Field.” U.S. Government Accountability Office (GAO). 1996. “Job Training
Issue brief, U.S. Department of Labor, Washington, DC. Partnership Act: Long-Term Earnings and Employment
Lane, Julia. 2000. “The Role of Job Turnover in the Low-Wage Labor Outcomes.” GAO/HEHS-996-40, Government
Market.” In The Low-Wage Labor Market: Challenges and Accountability Office, Washington, DC.
Opportunities for Self-Sufficiency, edited by Kelleen Kaye ———. 2011. “Employment and Training Programs: Opportunities
and Demetra Smith Nightingale, 185–198. Washington, Exist for Improving Efficiency.” GAO-11-506T, Government
DC: Urban Institute. Accountability Office, Washington, DC.
Maguire, Sheila, Joshua Freely, Carol Clymer, Maureen Conway, U.S. Bureau of Labor Statistics. 2014. “Employment Projects.” U.S.
and Deena Schwartz. 2010. “Tuning In to Local Labor Bureau of Labor Statistics, U.S. Department of Labor,
Markets: Findings from the Sectoral Employment Impact Washington, DC. http://www.bls.gov/emp/ep_chart_001.
Study.” Public/Private Ventures, Philadelphia, PA. htm.
Manyika, James, Susan Lund, Byron Auguste, and Sreenivas U.S. Department of Labor. 2002. “One-Stop Innovations: Leading
Ramaswamy. 2012. “Help Wanted: The Future of Work in Change under the WIA One-Stop System, Final Report.”
Advanced Economies.” Discussion Paper, McKinsey Global John J. Heldrich Center for Workforce Development, New
Institute, New York, NY. Brunswick, NJ.
Meyer, Daniel, and Geoffrey L. Wallace. 2009. “Poverty Levels and ———. 2014. “Training and Employment Programs,
Trends in Comparative Perspective.” In Changing Poverty, Summary of Budget Authority, FY 1984 to 2012, By
Changing Policies, edited by Maria Cancian and Sheldon Year of Appropriation.” Employment and Training
Danziger, 1–33. New York: Russell Sage Foundation. Administration, U.S. Department of Labor, Washington,
Mikelson, Kelly, and Demetra Smith Nightingale. 2004. “Estimates DC. http://www.doleta.gov/budget/bahist.cfm.
of Public and Private Expenditures on Job Training.” The Vollman, Jim. 2011. “Brookings Institution LMI Forum Real
Urban Institute, Washington, DC. Time Labor Market Information.” Brookings Institution,
Organisation for Economic Co-operation and Development Washington, DC.
(OECD). 2008. OECD Glossary of Statistical Terms, 299. Woolsey, Lindsey, and Kenneth Groves. 2010. “State Sector
Paris: OECD Publishing. Strategies Coming of Age: Implications for State Workforce
———. 2013. Public expenditure and participant stocks on Policymakers.” National Governors Association Center for
LMP.” Labour Market Programmes, OECD Stat Extracts, Best Practices, Washington, DC.
OECD, Paris, France. http://stats.oecd.org/Index. Zeidenberg, Matthew, Sung-Woo Cho, and Davis Jenkins. 2010.
aspx?DataSetCode=LMPEXP. “Washington State’s Integrated Basic Education And
Perez-Johnson, Irma, Quinn Moore, and Robert Santillano. 2011. Skills Training Program (I-BEST): New Evidence Of
“Improving the Effectiveness of Individual Training Effectiveness.” Working Paper, Community College
Accounts: Long-Term Findings from an Experimental Research Center, Teachers College, Columbia University,
Evaluation of Three Service Delivery Models.” Final report New York, NY.
submitted to the U.S. Department of Labor by Mathematica
Policy Research, Princeton, NJ.
Richburg-Hayes, LaShawn. 2008. “Helping Low-Wage Workers
Proposal 10
Persist in Community College.” MDRC, New York, NY. Akerlof, George. 1978. “The Economics of ‘Tagging’ as Applied
Schochet, Peter Z., John A. Burghardt, and Sheena M. McConnell. to the Optimal Income Tax, Welfare Programs, and
2008. “Does Job Corps Work? Impact Findings from the Manpower Planning.” American Economic Review 68 (1):
National Job Corps Study.” American Economic Review 68 8–19.
(5): 1864–86. Bastani, Spencer, Soren Blomquist, and Luca Micheletto. 2013. “The
Welfare Gains of Age-Related Optimal Income Taxation.”
International Economic Review 54 (4): 1219–49.
Baum II, Charles. 2002. “A Dynamic Analysis of the Effect of Child
Care Costs on the Work Decisions of Low-Income Mothers
with Infants.” Demography 39: 139–64.
184 Policies to Address Poverty in America
Berger, Mark, and Dan Black. 1992. “Child Care Subsidies, Quality Heckman, James, Jora Stixrud, and Sergio Urzua. 2006. “The Effects
of Care, and the Labor Supply of Low-Income, Single of Cognitive and Noncognitive Abilities on Labor Market
Mothers.” The Review of Economics and Statistics 74: Outcomes and Social Behavior.” Journal of Labor Economics
635–42. 24 (3): 411–82.
Bernal, Raquel, and Michael P. Keane. 2011. “Child Care Choices IRS. 2014. “EITC Participation Rate by States.” IRS, Washington,
and Children’s Cognitive Achievement: The Case of Single DC. http://www.eitc.irs.gov/EITC-Central/Participation-
Mothers.” Journal of Labor Economics 29 (3): 459–512. Rate.
Blank, Rebecca, and Brian Kovak. 2008. “Helping Disconnected Kearney, Melissa, and Lesley Turner. 2013. “Giving Secondary
Single Mothers.” Center on Children and Families Brief Earners a Tax Break: A Proposal to Help Low- and Middle-
#38, The Brookings Institution, Washington, DC. Income Families.” Discussion Paper 2013-07, The Hamilton
Blau, David, and Janet Currie. 2006. “Preschool, Day Care and Project, Brookings Institution, Washington, DC.
After-School Care: Who’s Minding the Kids.” In Handbook Kimmel, Jean. 1995. “The Effectiveness of Child-Care Subsidies in
of Economics of Education, edited by Eric Hanushek and Encouraging the Welfare-To-Work Transition of Low-
Finis Welch, 1163–1278. Amsterdam: North Holland. Income Single Mothers.” American Economic Review 85:
Bollinger, Christopher, Luis Gonzalez, and James P. Ziliak. 271–75.
2009. “Welfare Reform and the Level and Composition Laughlin, Lynda. 2013. Who’s Minding the Kids? Child Care
of Income.” In Welfare Reform and its Long-Term Arrangements: Spring 2011. U.S. Census Bureau, 70–135.
Consequences for America’s Poor, edited by James P. Ziliak, http://www.census.gov/prod/2013pubs/p70-135.pdf.
59–103. Cambridge: Cambridge University Press. Maag, Elaine. 2013. “Taxation and the Family: How Does the Tax
Child Care Aware of America. 2013. Child Care in America: System Subsidize Child Care Expenses?” Urban-Brookings
2013 State Fact Sheets. Child Care Aware of America, Tax Policy Center, Washington, DC.
Arlington, VA. Maag, Elaine, and Adam Carasso. 2013. “Taxation and the Family:
Congressional Research Service. 2012. “The Child Care and What is the Child Tax Credit?” Urban-Brookings Tax
Development Block Grant: Background and Funding.” Policy Center, Washington, DC.
RL30785, by Karen Lynch. Mankiw, N. Gregory, Matthew Weinzierl, and Danny Yagan. 2009.
Congressional Research Service. 2013. “The Temporary Assistance “Optimal Taxation in Theory and Practice.” Journal of
for Needy Families (TANF) Block Grant: A Primer on Economic Perspectives 23 (4): 147–74.
TANF Financing and Federal Requirements.” RL32748, by Meyer, Bruce, and Dan Rosenbaum. 2001. “Welfare, the Earned
Gene Falk. Income Tax Credit, and the Labor Supply of Single
Currie, Janet, and Firouz Gahvari. 2008. “Transfers in Cash and Mothers.” Quarterly Journal of Economics 116 (3): 1063–
In-Kind: Theory Meets the Data.” Journal of Economic 1114.
Literature 46 (2): 333–83. Morris, Pamela, Lisa Gennetian, Greg Duncan, and Aletha Huston.
Dahl, Gordon, and Lance Lochner. 2012. “The Impact of Family 2009. “How Welfare Policies Affect Child and Adolescent
Income on Child Achievement: Evidence from the Earned School Performance: Investigating Pathways of Influence
Income Tax Credit.” American Economic Review 102 (5): with Experimental Data.” In Welfare Reform and its Long-
1927–56. Term Consequences for America’s Poor, edited by James P.
Eissa, Nada, and Hilary Hoynes. 2004. “Taxes and the Labor Ziliak, 255–89. Cambridge: Cambridge University Press.
Market Participation of Married Couples: The Earned Romich, Jennifer, and Thomas Weisner. 2000. “How Families View
Income Tax Credit.” Journal of Public Economics 88 (9–10): and Use the EITC: Advance Payment versus Lump Sum
1931–58. Delivery.” National Tax Journal 53 (4): 1245–65.
Fox, Liana, Wen-Jui Han, Christopher Ruhm, and Jane Waldfogel. Schulman, Karen, and Helen Blank. 2013. “Pivot Point: State Child
2013. “Time for Children? Trends in the Employment Care Assistance Policies 2013.” National Women’s Law
Patterns of Parents, 1967–2009.” Demography 50 (1): 25–49. Center, Washington, DC.
Gao, Qin, Neeraj Kaushal, and Jane Waldfogel. 2009. “How Have Smeeding, Timothy, Katherin Ross Phillips, and Michael O’Connor.
Expansions in the Earned Income Tax Credit Affected 2000. “The EITC: Expectation, Knowledge, Use, and
Family Expenditures?” In Welfare Reform and its Long- Economic and Social Mobility.” National Tax Journal 53 (4):
Term Consequences for America’s Poor, edited by James P. 1187–1209.
Ziliak, 104–171. Cambridge: Cambridge University Press. Tekin, Erdal. 2007. “Child Care Subsidies, Wages, and the
Government Accountability Office (GAO). 2007, August. Employment of Single Mothers.” Journal of Human
“Advanced Earned Income Tax Credit.” Report to the Resources 42: 453–87.
Joint Committee on Taxation. http://www.gao.gov/ U.S. Census Bureau. Various years. “Current Population Survey.”
assets/270/265295.pdf. U.S. Census Bureau, Washington, DC.
Grogger, Jeffrey. 2003. “The Effects of Time Limits, the EITC, and www.census.gov/cps.
Other Policy Changes on Welfare Use, Work, and Income
Among Female-Headed Families.” Review of Economics
and Statistics 85 (2): 394–408.
The Hamilton Project • Brookings 185
References
U.S. Department of Agriculture. 2014. “Supplemental Nutrition Eissa, Nada, and Jeffrey Liebman. 1996. “Labor Supply Response
Assistance Program Participation Rates: Fiscal Years 2010 to the Earned Income Tax Credit.” Quarterly Journal of
and 2011.” Office of Policy Support, Food and Nutrition Economics 111: 605–37.
Service, U.S. Department of Agriculture, Washington, DC. Evans, William, and Craig Garthwaite. 2014. “Giving Mom a Break:
U.S. Department of Health and Human Services (DHHS). The Impact of Higher EITC Payments on Maternal Health.”
2012. “TANF Financial Data FY2012.” Office of Family American Economic Journal: Economic Policy 6(2): 258–90.
Assistance, Office of the Administration for Children and Feenberg, Daniel Richard, and Elizabeth Coutts. 1993. “An
Families, U.S. Department of Health and Human Services, Introduction to the TAXSIM Model.” Journal of Policy
Washington, DC. Analysis and Management 12 (1): 189–194.
———. 2013. “Temporary Assistance for Needy Families: http://www.nber.org/taxsim/.
Tenth Report to Congress.” Office of Family Assistance, Hoynes, Hilary, Douglas L. Miller, and David E. Simon.
Office of the Administration for Children and Families, Forthcoming. “Income, the Earned Income Tax Credit,
U.S. Department of Health and Human Services, and Infant Health.” American Economic Journal: Economic
Washington, DC. Policy.
Weinzierl, Matthew. 2011. “The Surprising Power of Age-Dependent IRS. 2013. “Statistics of Income Tax Year 2011.” Table 2.5. IRS,
Taxes.” Review of Economic Studies 78 (4): 1490–1518. Washington, DC. http://www.irs.gov/uac/SOI-Tax-Stats-
Ziliak, James P., Charles Hokayem, and Bradley Hardy. 2008. Individual-Income-Tax-Returns-Publication-1304-
“Child Care Subsidies and the Economic Well-Being (Complete-Report).
of Recipient Families: A Survey and Implications for Kearney, Melissa, and Leslie Turner. 2013. “Giving Secondary
Kentucky.” Technical Report Prepared for the Kentucky Earners a Tax Break: A Proposal to Help Low- and Middle-
Youth Advocates. University of Kentucky Center for Income Families.” Discussion Paper 2013-07, The Hamilton
Poverty Research, Lexington, KY. Project, Brookings Institution, Washington, DC.
MDRC. 2014. “Paycheck Plus: Expanded Earned Income Tax
Proposal 11 Credit for Single Adults.” http://www.mdrc.org/project/
paycheck-plus-expanded-earned-income-tax-credit-single-
Baker, Kevin. 2008. “Do Cash Transfer Programs Improve Infant adults#featured_content.
Health: Evidence from the 1993 Expansion of the Earned Meyer, Bruce D., and Dan T. Rosenbaum. 2000. “A Making Single
Income Tax Credit?” Mimeo, University of Notre Dame, Mothers Work: Recent Tax and Welfare Policy and its
South Bend, IN. Effects.” National Tax Journal, Vol. 53: 1027–1062.
Bitler, Marianne, and Hilary Hoynes. 2013. “The More Things Meyer, Bruce D., and Dan T. Rosenbaum. 2001. “Welfare, the
Change, the More They Stay the Same: The Safety Net, Earned Income Tax Credit, and the Labor Supply of Single
Living Arrangements, and Poverty in the Great Recession.” Mothers.” Quarterly Journal of Economics 116(3): 1063–
Working Paper 19449, National Bureau of Economic 1114.
Research, Cambridge, MA. Parrott, Sharon. 2014. “The EITC Works Very Well—But It Is
Bitler, Marianne, Hilary Hoynes, and Elira Kuka. 2013. “Do In- Not a Safety Net by Itself.” Center for Budget and Policy
Work Tax Credits Serve As A Safety Net?” Working Paper Priorities, Washington, DC.
19785, National Bureau of Economic Research, Cambridge, Rohaly, Jeffrey, Adam Carasso, and Mohammed Adeel
MA. Saleem. 2005. “The Urban-Brookings Tax Policy
Center on Budget and Policy Priorities (CBPP). 2014a. “Policy Center Microsimulation Model: Documentation and
Basics: The Earned Income Tax Credit.” Center on Budget Methodology for Version 0304.” Urban-Brookings Tax
and Policy Priorities, Washington, DC. Policy Center, Washington, DC. http://www.urban.org/
Center on Budget and Policy Priorities (CBPP). 2014b. UploadedPDF/411136_documentation.pdf.
“Unpublished Center on Budget and Policy Priorities Scholz, John Karl. 2007. “Employment-Based Tax Credits for Low-
analysis of March 2013 Current Population Survey and Skilled Workers.” Discussion Paper 2007-14, The Hamilton
Supplemental Poverty Measure (SPM) public use files.” Project, Brookings Institution, Washington, DC.
Center on Budget and Policy Priorities, Washington, DC. Short, Kathleen. 2013. “The Research Supplemental Poverty
Dahl, Gordon, and Lance Lochner. 2012. “The Impact of Family Measure: 2012.” Current Population Report P60-247,
Income on Child Achievement: Evidence from the Earned U.S. Census Bureau, U.S. Department of Commerce,
Income Tax Credit.” American Economic Review, 102(5): Washington, DC.
1927–56. Strully, Kate, David H. Rehkopf, and Ziming Xuan. 2010. “Effects
Eissa, Nada, and Hilary Hoynes. 2004. “Taxes and the Labor of Prenatal Poverty on Infant Health: State Earned Income
Market Participation of Married Couples: The Earned Tax Credits and Birth Weight.” American Sociological
Income Tax Credit.” Journal of Public Economics 88: 9–10. Review 75: 534–562.
Eissa, Nada, and Hilary Hoynes. 2006. “Behavioral Responses to Urban–Brookings Tax Policy Center. 2014. “Tax Facts, Historical
Taxes: Lessons from the EITC and Labor Supply.” Tax EITC Parameters.” Urban-Brookings Tax Policy Center,
Policy and the Economy 20: 74–110. Washington, DC.
186 Policies to Address Poverty in America
Urban–Brookings Tax Policy Center. n.d. “Tax Topics: Income ———. 2010b. “Moving Beyond the Jobs Crisis—Supporting
Measure Used in Distributional Analyses by the Tax Policy Material for Chapter 1 of the 2010 OECD Employment
Center.” Urban-Brookings Tax Policy Center, Washington, Outlook.” Paris: OECD Publishing.
DC. http://www.taxpolicycenter.org/TaxModel/income. Shelton, Alison. 2011. “Compensated Work Sharing Arrangements
cfm. (Short-Time Compensation) as an Alternative to Layoffs.”
U.S. Census Bureau. 2014. “Historical Income Tables: Income Congressional Research Service Report 40689, Washington,
Inequality.” U.S. Census Bureau, Washington, DC. DC.
Stevens, Ann Huff. 1997. “Persistent Effects of Job Displacement:
Proposal 12 The Importance of Multiple Job Losses.” Journal of Labor
Economics 15 (1): 165–88.
Abraham, Katharine G., and Susan N. Houseman. 1993. Job Sullivan, Daniel, and Till von Wachter. 2009. “Job Displacement
Security in America: Lessons from Germany. Washington: and Mortality: An Analysis using Administrative Data.”
Brookings Institution. Quarterly Journal of Economics 124 (3): 1265–1306.
———. 1994. “Does Employment Protection Inhibit Labor Market von Wachter, Till, Jae Song, and Joyce Manchester. 2011. “Long-
Flexibility? Lessons from Germany, France and Belgium.” Term Earnings Losses Due to Mass Layoffs during the 1982
In Social Protection vs. Economic Flexibility: Is There A Recession: An Analysis Using Longitudinal Administrative
Trade Off?, edited by R. Blank, 59–93. Chicago: University Data from 1974 to 2008.” Unpublished working paper.
of Chicago Press.
———. Forthcoming. “Short-Time Compensation as a Tool to
Mitigate Job Loss? Evidence on the U.S. Experience during
Proposal 13
the Recent Recession.” Industrial Relations. Aaronson, Daniel, Eric French, and James MacDonald. 2008. “The
Balducchi, David E., and Stephen A. Wandner. 2008. “Work Minimum Wage, Restaurant Prices, and Labor Market
Sharing Policy: Power Sharing and Stalemate in American Structure.” Journal of Human Resources 43 (3): 688–719.
Federalism.” Publius: The Journal of Federalism 38 (1): 1–27. Addison, John T., McKinley L. Blackburn, and Chad D. Cotti.
Black, Sandra E., Paul J. Devereux, and Kjell G. Salvanes. 2012. 2009. “Do Minimum Wages Raise Employment? Evidence
“Losing Heart? The Effect of Job Displacement on Health.” from the U.S. Retail-Trade Sector.” Labor Economics 16 (4):
Working Paper 18660, National Bureau of Economic 397–408.
Research, Cambridge, MA. ———. 2012. “The Effects of Minimum Wages on Labor Market
Boeri, Tito, and Herbert Bruecker. 2011. “Short-Time Work Benefits Outcomes: County-Level Estimates from the U.S.
Revisited: Some Lessons from the Great Recession.” Restaurant and Bar Sector.” British Journal of Industrial
Economic Policy 26 (68, October): 697–765. Relations 50 (3): 412–35.
Crimmann, Andreas, and Frank Wiessner. 2009. “Verschnaufpause Allegretto, Sylvia, Arindrajit Dube, Michael Reich, and Ben
dank Kurzarbeit.” IAB Kurzbericht. Zipperer. 2013. “Credible Research Designs for Minimum
Davis, Steven J., and Till von Wachter. 2011. “Recessions and the Wage Studies.” IZA Discussion Paper 7638, Institute for the
Costs of Job Loss.” Brookings Papers on Economic Activity Study of Labor, Bonn, Germany.
43 (2, Fall): 1–72. Autor, David H., Lawrence F. Katz, and Melissa S. Kearney.
Hijzen, Alexander, and Sebastien Martin. 2013. “The Role of Short- 2008. “Trends in U.S. Wage Inequality: Revising the
Time Work Schemes During the Global Financial Crisis Revisionists.” Review of Economics and Statistics 90 (2):
and Early Recovery: A Cross-Country Analysis.” IZA 300–323.
Journal of Labor Policy 2 (5): 1–31. Autor, David H., Alan Manning, and Christopher L. Smith. 2014.
Hilger, Nate. 2013. “How Does Family Income Affect Enrollment?” “The Contribution of the Minimum Wage to U.S. Wage
Unpublished working paper. Inequality over Three Decades: A Reassessment.” Working
International Labour Office (ILO). 2010. Germany’s Response to Paper, Massachusetts Institute of Technology, Cambridge,
the Crisis. G20 Country Brief prepared for the Meeting of MA.
Labour and Employment Ministers, Washington, DC. Bitler, Marianne P., and Hilary W. Hoynes. 2010. “The State of
Jacobson, Louis S., Robert J. LaLonde, and Daniel G. Sullivan. the Social Safety Net in the Post-Welfare Reform Era.”
1993. “Earnings Losses of Displaced Workers.” American Brookings Papers on Economic Activity Fall (2): 71–127.
Economic Review 83 (4): 685–709. Brochu, Pierre, and David A. Green. 2013. “The Impact of
Oreopoulos, Philip, Marianne Page, and Ann Huff Stevens. 2008. Minimum Wages on Labour Market Transitions.”
“The Intergenerational Effects of Worker Displacement.” Economic Journal 123 (573): 1203–35.
Journal of Labor Economics 26 (3): 455–83. Bureau of Economic Analysis. n.d. “Regional Data: GDP & Personal
Organisation for Economic Co-operation and Development Income.” Bureau of Economic Analysis, U.S. Department of
(OECD). 2010a. “Moving Beyond the Jobs Crisis.” Commerce, Washington, DC.
Employment Outlook. Paris: OECD Publishing.
The Hamilton Project • Brookings 187
References
Card, David, and Alan Krueger. 1994. “Minimum Wages and Neumark, David, J. M. Ian Salas, and William Wascher. 2013.
Employment: A Case Study of the Fast-Food Industry in “Revisiting the Minimum Wage-Employment Debate:
New Jersey and Pennsylvania.” American Economic Review Throwing Out the Baby with the Bathwater?” Working
48 (4): 772–93. Paper 18681, National Bureau of Economic Research,
———. 2000. “Minimum Wages and Employment: A Case Study Cambridge, MA.
of the Fast-Food Industry in New Jersey and Pennsylvania: Organisation for Economic Co-operation and Development
Reply.” American Economic Review 90 (5): 1397–1420. (OECD). 2013. ““Earnings: Minimum wages relative to
Congressional Budget Office (CBO). 2013. “The Effects of median wages.” OECD Employment and Labour Market
a Minimum-Wage Increase on Employment and Statistics (database). doi: 10.1787/data-00313-en (Accessed
Family Income.” Report, Congressional Budget Office, on 27 May 2014).
Washington, DC. Philippon, Thomas, and Ariell Reshef. 2012. “Wages and Human
Cooper, David. 2013. “Raising the Federal Minimum Wage to Capital in the US Finance Industry: 1909–2006.” Quarterly
$10.10 Would Lift Wages for Millions and Provide a Modest Journal of Economics 127 (4): 1551–1609.
Economic Boost.” Briefing Paper no. 371, Economic Policy Reich, Michael, Ken Jacobs, and Miranda Dietz. 2014. Making
Institute, Washington, DC. Mandates Work: Raising Labor Standards at the Local Level.
Dube, Arindrajit. 2014. “Minimum Wages and the Distribution of Los Angeles: University of California Press.
Family Incomes.” Unpublished manuscript. Rothstein, Jesse. 2010. “Is the EITC as Good as an NIT? Conditional
Dube, Arindrajit, T. William Lester, and Michael Reich. 2010. Cash Transfers and Tax Incidence.” American Economic
“Minimum Wage Effects Across State Borders: Estimates Journal: Economic Policy 2 (1): 177–208.
Using Contiguous Counties.” Review of Economics and Ruggles, Steven, J. Trent Alexander, Katie Genadek, Ronald
Statistics 92 (4): 945–64. Goeken, Matthew B. Schroeder, and Matthew Sobek. 2010.
———. 2013. “Minimum Wage Shocks, Employment Flows and Integrated Public Use Microdata Series (IPUMS), American
Labor Market Frictions.” IRLE Working Paper 149-13, Community Survey (ACS): Version 5.0 [Machine-readable
Institute for Research on Labor and Employment, Berkeley, database]. Minneapolis: University of Minnesota.
CA. Schmitt, John, and Janelle Jones. 2012. “Where Have All the Good
Dube, Arindrajit, Suresh Naidu, and Michael Reich. 2007. “The Jobs Gone?” Center for Economic and Policy Research,
Economic Effects of a Citywide Minimum Wage.” Washington, DC.
Industrial and Labor Relations Review 60 (4): 522–43. Unicon Research Corporation. 2012. “Current Population Survey.”
———. 2014. “Labor Market Impacts of San Francisco’s Minimum https://www.unicon.com/.
Wage.” In Making Mandates Work: Raising Labor
Standards at the Local Level, edited by Michael Reich, Ken
Jacobs, and Miranda Dietz, 47–69. University of California
Proposal 14
Press, Los Angeles. Coalition of Evidence-Based Policy. 2012. “Social Programs that
Dube, Arindrajit, and Ben Zipperer. 2014. “Pooled Synthetic Work.” Coalition for Evidence-Based Policy, Washington,
Control Estimates for Recurring Treatments: An DC.
Application to Minimum Wage Laws.” Unpublished Davenport, Thomas, and Jeanne Harris. 2007. Competing on
manuscript. Analytics: The New Science of Winning. Boston: Harvard
Firpo, Sergio, Nicole M. Fortin, and Thomas Lemieux. 2011. Business School Publishing.
“Occupational Tasks and Changes in the Wage Structure.” Digital Services Advisory Group, Federal Chief Information
Unpublished manuscript. Officers Council, and Federal Web Managers Council.
Hendren, Nathaniel. 2014. “The Inequality Deflator: Interpersonal 2012. “Digital Service Governance Recommendations.”
Comparisons without a Social Welfare Function.” http://www.whitehouse.gov/digitalgov/digital-services-
Unpublished manuscript. governance-recommendations.
Hoffman, Saul D. 2014. “Are the Effects of Minimum Wage Dinan, Kinsey. 2013. “Local Agency Lessons On Implementing
Increases Always Small? A Re-Analysis of Sabia, Random Assignment: An Example from NYC’s Child
Burkhauser, and Hansen?” Unpublished manuscript. Support Program.” Presentation to the Association for
Lee, David, and Emmanuel Saez. 2012. “Optimal Minimum Wage Public Policy Analysis and Management, Washington, DC.
Policy in Competitive Labor Markets.” Journal of Public Evans, Melanie. 2011. “Healthcare’s ‘Moneyball’: Predictive
Economics 96 (9–10): 739–49. Modeling Being Tested in Data-Driven Effort to Strike Out
Manning, Alan. 2012. “Minimum Wage: Maximum Impact.” Hospital Readmissions.” Modern Healthcare 41 (41): 28–30.
Resolution Foundation. Michaelides, Marios, Eileen Poe-Yamagata, Jacob Benus, and
Neumark, David, and William Wascher. 1992. “Employment Effects Dharmendra Tirumalasetti. 2012. “Impact of the
of Minimum and Subminimum Wages: Panel Data on Reemployment and Eligibility Assessment (REA) Initiative
State Minimum Wage Laws.” Industrial & Labor Relations in Nevada.” Impaq International LLC, Columbia, MD.
Review 46 (1): 55–81.
188 Policies to Address Poverty in America
Peikes, Deborah, Rebecca Sweetland Lester, Boyd Gilman, and
Randall Brown. 2012–13. “The Effects of Transitional
Care Models on Re-Admissions: A Review of the Current
Evidence.” GENERATIONS—Journal of the American
Society on Aging 36 (4, Winter): 44–55.
Raven, Maria C. 2009. “Improving Care and Cutting Costs for
High-Risk Medicaid Patients.” New York State Health
Foundation, New York, NY.
Raven, Maria C., John C. Billings, Lewis R. Goldfrank, Eric D.
Manheimer, and Marc N. Gourevitch. 2009. “Medicaid
Patients at High Risk for Frequent Hospital Admission:
Real-Time Identification and Remediable Risks.” Journal of
Urban Health 86 (2): 230–41.
U.S. Department of Health and Human Services (HHS).
n.d. “Home Visiting Evidence of Effectiveness.” U.S.
Department of Health and Human Services, Washington,
DC.
Weinberg, Stephen. 2013. “Using RCTs to Test Procedural
Changes: Evidence From Medicaid Fraud Investigations.”
Presentation to the Association for Public Policy Analysis
and Management, Washington, DC.
The Hamilton Project • Brookings 189
190 Policies to Address Poverty in America
The Hamilton Project • Brookings 191
192 Policies to Address Poverty in America
ADVISORY COUNCIL
GEORGE A. AKERLOF TIMOTHY GEITHNER ALICE M. RIVLIN
Koshland Professor of Economics Former U.S. Treasury Secretary Senior Fellow, The Brookings Institution
University of California at Berkeley Professor of Public Policy
RICHARD GEPHARDT Georgetown University
ROGER C. ALTMAN President & Chief Executive Officer
Founder & Executive Chairman Gephardt Group Government Affairs DAVID M. RUBENSTEIN
Evercore Co-Founder & Co-Chief Executive Officer
ROBERT GREENSTEIN The Carlyle Group
ALAN S. BLINDER President
Gordon S. Rentschler Memorial Professor Center on Budget and Policy Priorities ROBERT E. RUBIN
of Economics & Public Affairs Co-Chair, Council on Foreign Relations
Princeton University MICHAEL GREENSTONE Former U.S. Treasury Secretary
3M Professor of Environmental Economics
JONATHAN COSLET Massachusetts Institute of Technology LESLIE B. SAMUELS
Senior Partner & Chief Investment Officer Senior Counsel
TPG Capital, L.P. GLENN H. HUTCHINS Cleary Gottlieb Steen & Hamilton LLP
Co-Founder
ROBERT CUMBY Silver Lake SHERYL SANDBERG
Professor of Economics Chief Operating Officer
Georgetown University JIM JOHNSON Facebook
Chairman
JOHN DEUTCH Johnson Capital Partners RALPH L. SCHLOSSTEIN
Institute Professor President & Chief Executive Officer
Massachusetts Institute of Technology LAWRENCE F. KATZ Evercore
Elisabeth Allison Professor of Economics
CHRISTOPHER EDLEY, JR. Harvard University ERIC SCHMIDT
The Honorable William H. Orrick, Jr. Executive Chairman
Distinguished Professor; MARK MCKINNON Google Inc.
Faculty Director, Chief Justice Earl Warren Former Advisor to George W. Bush
Institute on Law and Social Policy Co-Founder, No Labels ERIC SCHWARTZ
Boalt School of Law, University of California, 76 West Holdings
Berkeley ERIC MINDICH
Chief Executive Officer THOMAS F. STEYER
BLAIR W. EFFRON Eton Park Capital Management Business Leader & Investor
Founding Partner
Centerview Partners LLC SUZANNE NORA JOHNSON LAWRENCE SUMMERS
Former Vice Chairman Charles W. Eliot University Professor
JUDY FEDER The Goldman Sachs Group, Inc. Harvard University
Professor & Former Dean
McCourt School of Public Policy PETER ORSZAG PETER THIEL
Georgetown University Vice Chairman of Global Banking Technology Entrepreneur, Investor, and
Citigroup, Inc. Philanthropist
ROLAND FRYER
Robert M. Beren Professor of Economics RICHARD PERRY LAURA D’ANDREA TYSON
Harvard University Managing Partner & Chief Executive Officer S.K. and Angela Chan Professor of Global
CEO, EdLabs Perry Capital Management, Haas School of Business
University of California, Berkeley
MARK T. GALLOGLY MEEGHAN PRUNTY EDELSTEIN
Cofounder & Managing Principal Senior Advisor
Centerbridge Partners The Hamilton Project MELISSA S. KEARNEY
Director
TED GAYER ROBERT D. REISCHAUER
Vice President & Director Distinguished Institute Fellow and
of Economic Studies President Emeritus
The Brookings Institution The Urban Institute
Introduction
Melissa S. Kearney, Benjamin H. Harris, and Karen L. Anderson
Section 1. Promoting Early Childhood Development
1. Expanding Preschool Access for Disadvantaged Children
Elizabeth U. Cascio and Diane Whitmore Schanzenbach
2. Addressing the Parenting Divide to Promote Early Childhood
Development for Disadvantaged Children
Ariel Kalil
3. Reducing Unintended Pregnancies for Low-Income Women
Isabel Sawhill and Joanna Venator
Section 2. Supporting Disadvantaged Youth
4. Designing Effective Mentoring Programs for Disadvantaged Youth
Phillip B. Levine
5. Expanding Summer Employment Opportunities for Low-Income Youth
Amy Ellen Schwartz and Jacob Leos-Urbel
6. Addressing the Academic Barriers to Higher Education
Bridget Terry Long
Section 3. Building Skills
7. Expanding Apprenticeship Opportunities in the United States
Robert I. Lerman
8. Improving Employment Outcomes for Disadvantaged Students
Harry J. Holzer
9. Providing Disadvantaged Workers with Skills to Succeed in the Labor Market
Sheena McConnell, Irma Perez-Johnson, and Jillian Berk
Section 4. Improving Safety Net and Work Support
10. Supporting Low-Income Workers through Refundable Child-Care Credits
James P. Ziliak
11. Building on the Success of the Earned Income Tax Credit
Hilary Hoynes
12. Encouraging Work Sharing to Reduce Unemployment
Katharine G. Abraham and Susan N. Houseman
13. Designing Thoughtful Minimum Wage Policy at the State and Local Levels
Arindrajit Dube
14. Smarter, Better, Faster: The Potential for Predictive Analytics and Rapid-
Cycle Evaluation to Improve Program Development and Outcomes
Scott Cody and Andrew Asher
HAMILTON
1775 M husetts Ave., NW 202) 797-6484
Ieshianiien, 0G 2062 wwnhemiltonprojectorg BRO OKI NG )
&) Printed on recycled paper.
File and source
- File
- proposal-12-work-sharing-to-reduce-unemployment.pdf
- Size
- 4,637,616 bytes
- SHA-256
- a07ce474a7078b6a6e71ff78a16b9d188216b01ddffe7174b4992e61f6a8de42
- Original
- www.brookings.edu