Listen to thisSummary
Many U.S. policymakers support changing the “culture" of poor parents to
encourage marriage, work, and religion as a means to end the
intergenerational transmission of poverty. In this article Jens Ludwig
and Susan Mayer review and evaluate research on how parental work,
marriage, and religion affect children's socioeconomic status as adults,
as well as on the likelihood that changing these indicators of parental
behavior will reduce poverty in the next generation. They conclude that
even if policymakers were able to ensure that all children had married,
working, and religious parents, the result would be a far smaller
reduction in poverty among the children's generation than many people
believe. The explanation for this “poverty-prevention paradox," say
Ludwig and Mayer, is that the poverty rate in the children's generation
depends not only on how many poor children grow up to be poor adults,
but also on how many nonpoor children grow up to be poor adults.
Reducing the chances that poor children become poor adults will
dramatically lower future poverty rates only if most poor adults begin
life as poor children. But most poor adults grow up as nonpoor children
in the type of “pro-social" households that policymakers are pushing to
attain. Moreover, little good evidence supports the idea that such
parental behaviors as marriage, work, and religious adherence have
strong causal effects on children's long-term economic success. The
authors argue that encouraging positive social behaviors in the parents
of poor children is a worthwhile goal in its own right. But they stress
that policymakers should recognize the limits of this strategy for
reducing poverty among future generations. There may be no substitute
for a system of social insurance and income transfers for those children
who do wind up poor as adults. A primary goal of contemporary U.S.
social policy is to reform the “culture" of poor parents to make it less
likely that their children will grow up to be poor. Because the
indicators of culture that policymakers usually emphasize are marriage,
work, and religion, in this paper we assess how much poverty rates in
the children's generation would fall if all children had married,
working, and religious parents. Research in this area is less
informative than one might expect. Little empirical research estimates
the relationship between parents” marriage, work, and religion and
children's eventual income as adults, and in our view much of that
research is seriously flawed. Our own estimates show that changing these
three aspects of family culture will reduce poverty in the children's
generation much less than many policymakers, policy analysts, and voters
seem to believe. The public’s concern about intergenerational economic
mobility appears to spring from a desire to see that children are not
condemned to a lifetime of poverty just because they were born to poor
parents. Many public discussions assume (as the quote above suggests)
that reducing poverty among future generations and reducing the
intergenerational transmission of poverty are equivalent goals. They are
not. The poverty rate in the children's generation depends not only on
how many poor children grow up to be poor adults, but also on how many
nonpoor children grow up to be poor adults. Reducing the chances that
poor children become poor adults will dramatically lower future poverty
rates only if most poor adults begin life as poor children. Even if
parental work, marriage, and religion improve children's economic future
as adults, getting all parents to work, marry, and attend religious
services would not cause poverty to plunge in the next generation
because most poor adults do not grow up in families headed by parents
who are unmarried, do not work, and do not attend religious services.
Epidemiologists often encounter a similar problem in fighting disease.1
As Geoffrey Rose puts it, “a large number of people at a small risk may
give rise to more cases of disease than the small number who are at high
risk," which limits what might be accomplished by focusing on high-risk
cases.2 Public policies that
seek to end tomorrow’s poverty by changing today’s parental culture
encounter a similar problem. We call this the poverty-prevention
paradox.
Public Policy and the Intergenerational Transfer of Poverty
Policy proposals to reduce the intergenerational transfer of poverty focus on three broad areas: schools, neighborhoods, and families. Cecilia Elena Rouse and Lisa Barrow discuss the role of schools in their article in this volume. We briefly discuss research on the relationship between children's neighborhoods and their economic success as adults before turning to the role of families, particularly the relationship between parents” marital status, work, and religion and their children's chances of being poor as adults.
Neighborhoods
Surprisingly little is known about the effect of children's
neighborhoods on their economic status as adults. Most researchers focus
on how neighborhoods affect child or adolescent outcomes“Outcomes that
are only modestly correlated with the children's economic status as
adults.3 In addition most of
the research is “nonexperimental"–that is, its subjects are not randomly
assigned to different neighborhood environments. Instead, the research
simply compares children whose families live in different neighborhoods.
Such an analytical approach, however, risks finding links between the
children's outcomes and the neighborhoods themselves when the outcomes
are instead due to something about the parents who choose to live in
certain neighborhoods.4 A
different approach taken by some researchers is to consider the
correlation, or similarity, in outcomes of children growing up within a
few blocks of one another. One study found that the correlation in the
years of schooling of children living in the same neighborhood is around
0.1 (where perfect correlation would be 1), which suggests a fairly
limited role for neighborhood influence–at least on average across a
nationally representative sample.5
Even if neighborhoods have little effect on average, the consequences of
growing up in America’s most disadvantaged urban communities could be
important for those at the bottom of the income distribution–a
possibility that has generated particular concern in public discussions.6
Yet even here the influence of neighborhoods on many
childhood outcomes may be surprisingly modest, at least according to
available data from the U.S. Department of Housing and Urban
Development’s Moving to Opportunity (MTO) demonstration. The MTO program
overcomes the self-selection problem by randomly assigning low-income,
mostly minority families into groups that are offered different forms of
treatment. It thereby creates substantial differences in neighborhoods
among otherwise comparable groups of poor parents and children. Although
evaluations of MTO after four to seven years find that moving to less
disadvantaged communities reduces risky and criminal behavior in girls,
they find that such moves on balance increase these behaviors in boys
and have no detectable effects on children's academic performance, such
as achievement test results or the chance of dropping out of high
school.7 It is possible that
the benefits of moving away from very disadvantaged neighborhoods may
become greater over time, or that the benefits may be more pronounced
among children who were very young when they moved.8
But there is as yet no strong evidence that moving poor families to less
disadvantaged areas will substantially change children's life chances.9
Families
If moving children to better neighborhoods is unlikely to improve their
lives significantly, it is reasonable to contemplate making changes in
the families in which children are raised. Historically, recognizing
that serious material deprivation could hurt children's life chances,
antipoverty policy focused on changing the family environment by
providing income transfers to poor families.10
But as the number of single mothers increased and welfare rolls began to
grow, policymakers turned their attention to trying to discourage out-of
wedlock childbearing and to encourage mothers to work outside the home.
The welfare reform legislation signed by President Bill Clinton in 1996
explicitly bases the need for attention to these two issues on their
presumed effect on children's life chances.11
As Douglas Besharov notes, for most Americans welfare reform “was about
reducing the deep-seated social and personal dysfunction associated with
long-term dependency, thereby ultimately reducing poverty. For welfare
reform to be a success on this measure will depend on whether the
low-paying jobs taken by many women leaving welfare lead to better jobs,
whether the household arrangements (and other sources of support) that
have allowed mothers to leave welfare without working prove supportive
and nurturing, and whether the eventual result is less dysfunctional
behavior among parents and better outcomes for children."12
More recently many commentators aiming to change the “culture of
poverty" have added another goal: religious adherence. Senator Rick
Santorum, chairman of the Senate Republican Conference, has said apropos
of faith-based social welfare programs, “The whole idea of funding
people of faith is not just to provide good human services. It’s also to
provide good human services with that additional touch, if you will,
with that aspect of healing that comes through that spiritual
interaction. ... You can’t ignore the importance of the spiritual part
of someone’s life and say you’re going to solve their problems. you’re
throwing good money after bad."13
The importance of religion in addressing poverty, like the promotion of
work and marriage, inspires bipartisan agreement. In discussing
faith-based social welfare programs several years ago, then–Vice
President Al Gore argued that their “religious character ... is so
often key to their effectiveness," although he noted that the
government’s job would not be to promote “a particular religious view."14
Commentator William Raspberry notes, “Many of the problems that are most
difficult to get at in our society today have to do with changing
attitudes. ... There are people who do these things and some of the
most successful ones are those who go to changing the person from the
inside. Religious organizations may be better equipped than most
organizations to do that kind of thing."15
Public debates on antipoverty policy devote more attention
to changing parental culture than to approaches that would change
specific parenting practices.16
Such debates also give relatively little attention to another
family- oriented strategy that is beyond our focus: replacing rather
than changing the child's home environment, for example, by sending the
child to preschool and before- and afterschool programs.17
As David Brooks notes, recently conservatives “have had free rein to
offer their own recipe for social renewal: churches that restrain male
selfishness, decency standards that check hedonism, social norms that
discourage childbearing outside of wedlock. ... Daniel Patrick
Moynihan observed that the core conservative truth is that culture
matters most [for poverty], and that the core liberal truth is that
government can reshape culture."18
In the next section we discuss in more detail the ways in which parental
work, marriage, and religion–the most common indicators of “culture"–may
affect children's educational attainment and future income, and we
review what researchers have found about the likelihood that changing
each of these indicators will reduce poverty in the next generation.
Consistent with the beliefs of many policymakers, some research has
documented large correlations between whether children lived in a
two-parent family“Or lived with at least one parent who worked–and their
schooling and adult income. The simple correlation between parents” religious participation and children's schooling and income is less
strong and less consistent. There is, however, little evidence to date
that any of these relationships are causal–that, for example, living in
a two-parent family in and of itself raises a child's educational
attainment. We then consider what would happen to poverty in the
children's generation if one could substantially change these indicators
of culture among today’s parents. Finally we discuss the implications
for public policy.
The Effects of Parental Work, Marriage, and Religious Adherence on Children
Research on the relationship between parents” religious adherence, family structure, and employment and their children's schooling and future income is hampered by three problems that make definitive conclusions impossible. First, little research directly examines the relationship between these parental characteristics and the adult income of children. Policymakers often generalize from links between these parental characteristics and young children's outcomes to the links between these characteristics and children's schooling or income as adults, even though the relationships between young children's outcomes and adult schooling or income are generally modest. Second, little research exists about the causal rather than the correlational links between these parental characteristics and children's adult outcomes, and studies that do try to assess causality produce conflicting results. Third, in policy debates and in some research, religion, family structure, and employment are treated as though they were dichotomous–religious or not, married or not, working or not. But each is much more complicated. For example, among the many alternatives to the twomarried- parent household are a household headed by a single female, a household with a stepfather, or a long-term cohabiting relationship. Policymakers and researchers must identify carefully the terms of their comparisons in their claims about what is good for children.
Do Religious Parents Increase Children's
Schooling and Adult Income?
Policymakers who encourage religion or religiosity among parents as a
way to help children avoid poverty in adulthood believe that the moral
content of religion leads both children and parents to behave in ways
that promote children's future economic well-being. According to this
view, religious belief provides a moral compass that leads children away
from teenage childbearing, delinquency, drug and alcohol use, and other
behaviors that can lead to a life of poverty. At the same time religion
instills a work ethic, honesty, and other characteristics valued by
employers, making it more likely that when children grow up they will
get and keep a job. Religious parents will stay married and avoid
behaviors that lead to the kinds of family dysfunction that hurt
children's life chances. Thus, parents” religion can enhance children's
future economic well-being by encouraging parents to behave in ways that
benefit the children and by passing on religious attitudes to the
children. Ideally, to assess the relationship between parents” religion
and children's adult income or educational attainment one needs a
meaningful measure of parents” religion and a longterm data set that
allows one to link a child's religious environment with his or her adult
outcomes. One must also be able to control all the factors that are
correlated with parental religion that might also affect children's
eventual income and schooling. For example, parents who are more
motivated and able to attend religious services may also be more likely
to participate in school events or other activities that enhance their
children's outcomes. No research achieves this standard. An even more
basic issue with which research in this area must grapple is determining
what aspect of parents” religion–affiliation, denomination, or
intensity–might have the greatest effect on their children's outcomes.
As yet there is no agreement on this point. Thus it is not always clear
exactly what policymakers mean when they talk about religion as a way to
end poverty. But clarity on this point is important, because trends in
religious belief, denominational membership, and intensity of religious
observance are not the same, and they may have quite different effects
on children.19 Right now,
the main policy mechanism for encouraging religion or religiosity seems
to be public funding for religious organizations to provide social
services. Religious groups have always played an important role in
aiding the poor in the United States, although that support has
historically been mostly separate from government aid. The federal
government’s direct support of religious social services started with
the “charitable choice" measures signed into law by President Clinton
between 1996 and 2000 to keep religious organizations from being
excluded from competition for federal funds simply because they are
religious.20 President
George W. Bush’s “faith-based initiative" aims to expand the ability of
faith-based organizations to compete for federal funding. One effect of
these policy changes is to reduce the “cost" to individuals of religious
participation: it requires less effort for people to receive religious
messages when they are already at the local church, mosque, or synagogue
for job training, drug abuse treatment, or other services. Although
policymakers say that they uphold the value of religious choice, and
hence that they do not mean to influence the denomination that
individuals choose, denomination is still an important aspect of
religion and it may be inadvertently influenced if the government funds
religious social services. Low-income Americans are already more likely
than high-income Americans to identify with a religious denomination,
attend church, and pray often. In the General Social Survey, respondents
whose income is below the mean for all Americans are more likely to
identify with a religion than respondents whose income is above the
mean; in fact, 91.2 percent of respondents whose family income is below
the mean claim a religious affiliation. Respondents whose income is
below the mean attend religious services at about the same rate as those
whose income is above the mean. 21
African Americans are more likely to identify with a religion and
to have greater religious intensity than whites. Yet African Americans
have incomes that average only about half the incomes of whites.22
These descriptive statistics raise questions about the extent to which “finding religion" would reduce poverty in the next generation, although
they are of course far from conclusive evidence about the causal effects
of religion on poverty. Two studies estimate the relationship between
parents” religious denomination and children's adult income. One, by
Nigel Tomes, finds that parental religious denomination has no effect on
children's income as adults once parental education is accounted for.23
The other, by Todd Steen, finds that net of their own education and
labor market experience, adult sons raised in Jewish and Catholic homes
earn more than children raised in Protestant homes, and that adult
children raised in families with no religious affiliation earn about the
same as those raised in Protestant homes.24
Several studies consider the relationship between an adult’s religious
affiliation and his or her own income. The incomes of adults who
identify with any religion are no higher than those of adults who
identify with none.25 This
conclusion, which holds even after taking into account the effects of
family background, is consistent with the simple descriptive statistics
already noted. But Jews have higher incomes and more schooling than
members of other denominations, while members of theologically
conservative Christian denominations (fundamentalist, Pentecostal, or
sectarian religions) have lower incomes and less schooling.26
One study finds that whether boys attended religious services at least
twice a month had a small and statistically insignificant association
with their eventual earnings.27
The religiosity of adults also appears to be linked with their own
income and schooling, at least in some studies. Bruce Sacerdote and
Edward Glaeser find that people with more schooling attend church more
often than those with less schooling: 50 percent of college graduates,
but only 36 percent of high school dropouts, attend church “several
times a year."28 But members
of the most highly educated denominations attend church least often, and
members of the least educated denominations attend most often. Sacerdote
and Glaeser explain this puzzle by noting that education reduces the
intensity of religious belief, so that people sort into less fundamental
religions as they get more schooling. This observation does not tell us
whether children raised in families that adhere to fundamental religions
get less schooling, but it does suggest, consistent with Tomes’s findings regarding religious mobility, that as such children get more
schooling they are likely to switch to less fundamental denominations.29
If adults sort across denominations based on their
schooling, the correlation between adults” education and their religious
affiliation cannot be used to infer the relationship between parental
religion and children's education. The research discussed so far is
unable to project what would happen to children's eventual income if
some outside force caused their parents to become more religious,
because it is unable to isolate the causal influence of religion. Many
factors that cause parents to be religious–motivation, sociability, good
physical and mental health, access to good public or their own private
transportation” can also affect their own income and their children's
future income. Jonathan Gruber tries to assess the causal influence of
religion by looking not at the effect of a family’s own religion but
instead at the effect of living in geographic areas where religion is
concentrated.30 Gruber uses
ancestry in an area as a predictor of religious density. Thus an Italian
living in an area with many Poles lives with a greater density of
Catholics than an Italian living in an area with many Swedes. Gruber
finds that adults are much more likely to participate in religious
activities when they live in areas where a larger fraction of people
share their religion. People who live in such areas are also more likely
to graduate from college and to have higher incomes. As Gruber
acknowledges, the fact that greater religious density appears to lead to
better outcomes among adults does not mean that the individual’s religion or religious intensity leads to the better outcomes.
Characteristics of more homogeneous communities or the social benefits
provided by religious institutions may lead to these better outcomes.
For example, people trust people who are like themselves more than they
trust people different from themselves. It may be trust and other
aspects of community, not religion, that lead to greater integration and
better outcomes for adults and children. Gruber’s work suggests that a
highly religiously segregated society would produce better outcomes than
a society that is equally religious but less segregated. But religious
segregation might be costly in other ways, especially because religion
is correlated with race and income. Gruber also finds evidence that high
religious density leads to more intense religious beliefs, an effect
that recent history suggests may not be unambiguously beneficial. In
sum, because people can choose whether to follow a religion, what
religion to follow, and how enthusiastically to follow it, religion may
be both a cause and an effect of many individual and family
characteristics that affect children's lives. It is hard to separate the
effects of factors that cause not only parents” religious affiliation
and intensity but also their children's outcomes from the effect of
religion itself. Almost no research does a good job of this. Moreover
most children are already raised in families that identify with some
religion, poor families are no less likely than rich families to
identify with a religion, and children who are raised in families with
no religious affiliation are likely to grow up to identify with some
religion.31
Do Married Parents Improve Children's
Schooling and Adult Income?
Growing up with married parents has long been correlated with a litany
of positive child outcomes–a fact often cited by supporters of policies
that promote marriage.32
Many recent studies take into account the effects of family background
to try to establish more closely the causal relationship between
parents” marital status and children's outcomes. That analytical
approach sometimes diminishes the link between parents” marital status
and children's outcomes–a finding often cited by observers skeptical of
policies that promote marriage. Little research estimates the
relationship between parents” marital status and their children's adult
income. Moreover, researchers have no reliable way to estimate causal
models of the effect of parents” marital status, and they have not been
consistent in defining and measuring marital status. As a result there
is little good empirical basis for estimating the likely outcome of
policies that encourage marriage. The one study that does estimate the
relationship between parents” marital status and adult poverty of their
children finds that having unmarried parents during childhood has no
statistically significant effect on the probability of being poor at age
twenty-four, regardless of whether the parents were divorced, separated,
or never married.33 The
model attempts to isolate the effect of marital status by taking into
account effects caused by parents” education, religion, and national
origin. But such models do not necessarily provide better estimates
about the causal effects of parents” marital status than simple
correlations do, because they can still leave out many factors that are
related to both parents” marital status and children's outcomes, thus
biasing the estimates either up or down.34
In addition, these models often take into account the effects of factors
that are arguably caused by the parents” marital status. Family income
is a case in point; but income often declines when families break up. If
low family income is a result of parents” marital status, its effect
should not be excluded in estimating the overall effect of parents” marital status on children. Doing so probably biases the effect of
single parenthood downward. When Charles Manski and several colleagues
estimated the effect of family structure during childhood on high school
graduation, using a variety of methods to take into account the way
adults self-select into different types of family arrangements, they
found that the estimates depended on how they took into account this
self-selection.35 Some
researchers have used parental death as a quasiexperiment to examine
whether parental absence hurts children's outcomes. Parental death is
assumed to be random or at least less correlated than divorce with
unobserved parental characteristics. These studies typically find that a
parent’s death has effects that are modest or nonexistent, and certainly
much smaller than those that have been estimated from parental divorce.36
There are many types of family structures, and families can transition
from one to another. Thus to argue that one type of family structure is
beneficial to children requires clearly specifying the alternative
family structure that one has in mind. Most people seem to agree that
living with married biological parents is ideal for children. Research
supports the claim that children raised by only one biological parent
fare worse than children raised by both biological parents in many
areas, including their schooling and mental health.37
Anne Case, I-Fen Lin, and Sara McLanahan find that in a
sample of children living in married-couple families, step-, adopted,
and foster children averaged one year less of schooling than the
biological children of the same mothers.38
On the other hand, many policymakers seem to believe that living with a
never-married single mother is the worst family structure for children.
But several studies suggest that being raised by a step-parent may be
worse for children's educational attainment than being raised by a
single mother, regardless of whether the mother has never married or is
divorced.39 Although these
studies take into account the effects of several important
characteristics of parents and children, such as parent education or
race and ethnicity, none tries to take into account the possibility of
unobserved characteristics that may confound the link between family
structure and child outcomes. From the research on family structure, we
draw two conclusions. First, on average children who grow up with two
married biological parents get more schooling and earn higher wages than
children who grow up in any other family type. And second, children
raised in families with a step-parent get no more schooling and may get
less schooling than those raised by their single biological mother. But
the degree to which these patterns reflect the causal impacts of the
family structure on children's life chances is not well understood.
Do Working Parents Improve Children's
Schooling and Adult Income?
The intent of the welfare reforms of the 1990s was to get single mothers
into the labor market to increase family income, reduce reliance on
welfare, and provide a role model for children's eventual work habits.
Some people also argued that work would be a disciplining and
integrating experience, making these mothers happier and more
selfconfident and thus also benefiting their children. Others thought
that for the most disadvantaged children, outside child care would
provide a better environment for social and cognitive development than
their own home. At the same time, many supporters of welfare reform and
work requirements argue as a general principle that maternal work
outside the home will impair the social and cognitive development of at
least young children. Thus many policymakers hold that young children
should live in a home with married parents, one of whom works (the
father) and one of whom stays home with the children (the mother)–the
idealized traditional family of the past. But when that ideal cannot be
met, the custodial parent should work both because it is better for
children to have one working parent–because work and selfsufficiency are
morally desirable–and because it lowers the cost to governments that
might otherwise have to support the family. According to this reasoning,
children in married-couple families will be worse off if their mothers
work than if they do not, and children in single-mother families will be
better off if their mothers work than if they do not. And children of
working single mothers will do worse than children of married mothers
who do not work. It is not clear whether children of working single
mothers would do better or worse than children of married mothers who
work.40 We could find no
study that estimates the effects of maternal work on children's
schooling or eventual income. In fact, we could find no research that
estimates the effect of parental or maternal employ- ment on children's
income as adults. Most of the research estimates how maternal employment
affects young children's cognitive skills.41
Although the results are mixed, the most recent and usually the best of
these studies find that maternal employment during a child's first year
slightly impairs the child's cognitive skills.42
The findings for maternal work during the child's second and third years
are less conclusive.43 The
evidence on the relationship between maternal employment and young
children's cognitive test scores is not strong, and there is little
consensus about whether effects found for preschoolers are maintained as
children get older.44
Nonetheless, one can use this evidence to do a back-of-theenvelope
calculation of the potential importance of maternal employment for wages
in the next generation to demonstrate how a relationship found for young
children translates into an adult outcome. If we assume that the
higher-end estimates of the effect of maternal employment on a child's
cognitive skills during his or her first year of life are correct, that
these effects are causal, and that they are maintained into the child's
adulthood, we can calculate that mothers” employment in the child's
first year of life, through its effect on cognitive skill, would
decrease future wages for white children by about 4 percent and leave
unchanged the wages of black and Hispanic children.45
Thus maternal employment is not likely to have any noticeable effect on
poverty in the next generation through its effect on cognitive test
scores. One study that assesses the relationship between parental and
children's work ethic finds that parents and children work similar
numbers of hours and that the similarity is only partially accounted for
by similar labor market conditions.46
Parents pass on preferences for work hours through their
modeling of work behavior or through other factors. This research is at
least consistent with the idea that children model their work habits on
their parents” work habits. Two public policy changes during the 1990s
provide a useful “natural experiment" for examining the effects of an
externally imposed increase in employment among single mothers. The
first, the 1996 welfare reforms, implemented work requirements for
welfare recipients. In principle, the experimental evaluations of these
work requirements” that is, the evaluations that sort study participants
by random assignment and thus are the gold standard of research–should
provide the most direct evidence about how policies that encourage (or
require) work affect disadvantaged children. The second change, the
increasing generosity of the federal earned income tax credit, also
encouraged many single mothers to increase their work effort.47
Research documenting how these policies affect the work
activity of single mothers is extensive.
48 Far less research, however, documents how changes in the
mother’s work activity change child outcomes. The most recent and most
comprehensive synthesis, by Jeffery Grogger and Lynn Karoly, examines
sixty-seven studies based on either randomassignment experiments or
econometric techniques intended to control for unobserved
characteristics that might confound the link between work activity and
child outcomes” in other words, techniques intended to approximate the
rigor of the random-assignment experiments. For our purposes, the two
outcomes of most interest are academic achievement and behavior problems
in adolescents. 49 Grogger
and Karoly’s synthesis suggests that welfare reforms that increased
mother’s work effort either increased the behavior problems of
adolescents and reduced their academic achievement or had mixed effects,
depending on the specific policy. (These same reforms sometimes had
opposite effects on younger children's outcomes.) This evidence should
be interpreted cautiously both because the number of studies examined
was modest and because all evaluated relatively short-term effects
before welfare time limits were implemented. (These limits restrict the
total amount of time women can spend on welfare, and so may increasingly
push a different, and particularly disadvantaged, population of
low-income women into the labor market.) Nonetheless this research
suggests that policies that encourage poor single mothers to work
probably do not improve the behavior or school achievement of their
adolescent children. Overall we can draw three conclusions about the
relationship between maternal work and children's achievements. First,
the evidence does not yet allow us to assess whether mothers” employment
has different effects on children's schooling or future income depending
on whether the mothers are married or single parents. Second, most
evidence about how maternal employment affects children's outcomes
suffers from methodological problems serious enough to limit its
usefulness for policy purposes. And third, studies that evaluate
programs to encourage work among low-skilled single mothers suggest that
increasing their employment is not unambiguously beneficial for their
children's schooling or behavior.
The Poverty-Prevention Paradox
Even if parents” culture as indicated by marriage, work, and religion had a causal effect on children's schooling and adult income” which is uncertain–encouraging parents to marry, work, and become religious would do far less to reduce poverty among future generations of American adults than most policymakers believe. The reason for this conclusion is that children who grow up with parents who are married, working, and religious also face some risk of experiencing poverty as adults. Therefore even successful efforts to change parental behavior or culture among “high-risk" families will have surprisingly modest effects on poverty in the next generation. We term this the poverty-prevention paradox. To illustrate the paradox, we estimate the expected poverty rates in adulthood of children who grow up with their biological fathers and those who grow up apart from them. We use data from the U.S. Department of Education's National Education Longitudinal Study (NELS) of 1988, which interviewed a nationally representative sample of eighth graders in 1988 and then again in 2000, when participants would have been around twenty-five years old. According to these data, eighth graders who were living apart from their biological fathers had an expected poverty rate of 16.6 percent when they were twenty-five. In contrast, the poverty rate for eighth graders who were living with their fathers was 9.9 percent. The powerful association between growing up without one’s father and being poor as an adult has led many people to conclude that addressing “culture" is the key to reducing poverty in America. However, the contribution of fathers” absence to overall poverty in the next generation depends not only on the difference in the poverty rates but also on the relative size of these two groups of children. According to the NELS, in 1988, 28 percent of eighth graders were living apart from their biological fathers and 72 percent were living with their fathers. To determine the contribution of each group of children to the overall poverty population in the next generation, we multiply the proportion of children in each family type by the poverty rate for each group by the total number of children in the population. For example, consider a population of 10,000 children (2,800 of whom are raised apart from their biological father and 7,200 of whom are raised with their father). Using the numbers from the NELS data, we see that about 465 children from father-absent families would be poor at age twenty-five (0.28 “0.166 “10,000) as compared with 713 children from father-present families (0.72 “0.099 “10,000). Note that despite their higher poverty rate, children from father-absent families would account for only 39.5 percent of poor adults (465 divided by 1,178). Next, we calculate what the size of the poverty population would be if all eighth graders had lived with two biological parents in 1990. Using the same formulas as above and assuming that the poverty rate for all children was the same as the poverty rate of those living with both biological parents, we find that about 990 adults would be poor in the next generation (10,000 “0.099), a reduction of 188 poor people. Despite the powerful link between family structure and children's chances of being poor as an adult, ensuring that every single eighth grader in the United States lived with his or her biological father would eliminate only around 16 percent (188 divided by 1,178) of poverty in the children's generation–even assuming that the effect of living with one’s father is entirely causal, which it almost certainly is not. This perhaps surprising conclusion holds for our other measures of parent culture as well. For example, similar calculations using NELS data show that ensuring that all eighth graders lived in a household with at least one working adult would lower poverty in the children's generation by 7 percent. Ensuring that all eighth graders participated in religious services would cut poverty in their generation by 9 percent. If eighth graders lived with their biological parents, at least one parent worked, and the child attended religious services, poverty in their generation would fall by 22 percent, assuming that the effects of these indicators of culture on children's poverty status are entirely causal. One potential limitation of these estimates is that the indicators of culture are measured only in the eighth grade. But even when we use data from the Panel Study of Income Dynamics (which follows children for a longer period than the NELS) to measure these family environments over several years rather than just at a single point in time, poverty in the children's generation does not decline much. It is possible to generate calculations indicating that changing “culture" would cut future poverty dramatically. If no child ever spent time in a household that did not have two married adults, at least one of whom was working, and that did not attend weekly religious services–and if these factors all had a causal effect on the child's chance of being poor–poverty in the children's generation would fall by as much as three-quarters. But setting such a high bar for the “pro-social" ideal would mean that the large majority of American households would not meet this standard and would assume that government interventions were capable of completely eliminating, without exception, such features of modern American life as divorce, job changes, and the choice not to attend religious services each and every week.50
Policy Implications
Many policymakers believe that they could largely eliminate poverty in America if only they could increase parental attachment to mainstream institutions such as work, marriage, and religion. They assume that something about parental engagement with these institutions is associated with the capacity of families to instill in children the cognitive and noncognitive skills crucial for long-term economic success. Changing parental behavior in this way serves as an omnibus strategy for changing all the other parenting practices that matter for children. Yet policymakers” faith that the key to reducing poverty is to encourage marriage, work, and religion among poor parents rests on a shaky empirical foundation. Little good evidence supports the idea that these parental behaviors have strong causal effects on children's long-term economic success. Most studies on this topic focus on short-term effects on children, such as schooling, rather than on adult earnings or employment, and few studies convincingly solve the selfselection problem that plagues all research in this area. Parents who choose to get married (or work or go to church frequently) are likely to be systematically different from parents who choose not to do these things, in ways that are difficult for social scientists to observe with available data sets. As a consequence, convincingly separating out the ways in which parental marriage, work, or religion affect children from the ways in which a parent’s propensity to be connected with these mainstream institutions does so is quite difficult. Equally important is our observation that most adults who experience some economic poverty were brought up in the sort of “Ozzie and Harriet" households that many policymakers wish to make universal. The risk of adult poverty is indeed much lower among adults brought up in such households, but these “pro-social" households are so numerous that they wind up accounting for most poor adults. Targeting interventions to the highest-risk children is, in our view, a worthwhile goal in its own right, but policymakers should recognize the limits of this strategy for reducing poverty among future generations” the poverty-prevention paradox. We are not arguing that a focus on high-risk behavior will never make a large dent in the overall scale of a social problem. The size of the dent that it will make will depend on how widespread the behavior is and on the strength of its association with the problem.51 If in the future there is a substantial increase in the share of households that never attend religious services or that lack two parents or that lack at least one worker, then targeting parent culture could become more important for ending poverty in America than our estimates suggest. Although the share of births to unmarried women has been increasing over the long term, this trend has flattened out over the past ten years.52 Recent years have also seen substantial increases in labor force participation rates among single or never-married mothers, while rates for married mothers have changed little.53 And although the share of Americans who attend religious services has been in steady decline, as in other industrialized countries, the United States still has one of the world's highest rates of religious attendance.54 More important for present purposes, measures of religiosity are weakly correlated with poverty. Nor are we arguing that interventions focused on parental employment, marriage, or religious participation are necessarily a bad idea. There might be many reasons for changing these behaviors apart from the chance that doing so will reduce poverty in the next generation. The argument we make here is not about the social benefits of an Ozzie-and-Harriet approach in relation to the costs“Which is the relevant comparison for decisions about whether to support specific policy interventions–but rather about its benefits in relation to the scale of the overall poverty problem. Many policymakers believe that changing parental work, marriage, or religiosity can end poverty in America. Based on the available evidence, that prospect does not seem likely. Our final observation about the poverty-prevention paradox that arises with the Ozzie-and-Harriet approach is in some sense similar to the conclusion of Christopher Jencks and his colleagues regarding the role of economic poverty among children.55 If ensuring that no child is raised in economic or moral poverty cannot eliminate poverty in the future, or even make a large dent in the problem, then what can? The answer that Jencks and colleagues offered thirty years ago seems as relevant today. To reduce poverty among future generations, there may be no substitute for a system of social insurance and income transfers for those children who end up poor as adults.
Notes
1. Thanks to Philip Cook for this point.
2. Geoffrey Rose, “Sick Individuals and Sick Populations," International Journal of Epidemiology 14 (1985): 37.
3. For example, previous research suggests that a 1 standard deviation increase in cognitive test scores is associated with an increase in earnings of between 3 and 27 percent, with the best estimates in the 15 to 20 percent range; Christopher Winship and Sanders Korenman, “Economic Success and the Evolution of Schooling and Mental Ability," in Earning and Learning: How School Matters, edited by Susan E. Mayer and Paul E. Peterson (Brookings, 1999), pp. 49-78. Choosing the midpoint of the best estimates (18 percent), an intervention that increases children's cognitive test scores by 0.20 standard deviations, would increase adult earnings by 0.20*18 = 3.6 percent. Child characteristics other than test scores, such as self-esteem, efficacy, and depression, have even weaker effects on future income than cognitive test scores; Greg J. Duncan and others, “School Readiness and Later Achievement," manuscript, Northwestern University (2004).
4. For good reviews of this literature, see Christopher Jencks and Susan Mayer, “The Social Consequences of Growing Up in a Poor Neighborhood," in Inner-City Poverty in the United States, edited by Laurence Lynn and Michael McGeary (Washington: National Academy Press, 1991), pp. 111–86; and Ingrid Gould Ellen and Margery Austin Turner, “Do Neighborhoods Matter and Why?" in Choosing a Better Life, edited by John Goering and Judith Feins (Brookings, 2003), pp. 313–38.
5. Gary Solon, Marianne E. Page, and Greg J. Duncan, “Correlations between Neighboring Children in their Subsequent Educational Attainment," Review of Economics and Statistics 82, no. 3 (2000): 383–92.
6. Alex Kotlowitz, There Are No Children Here (New York: Anchor Books, 1992); LeAlan Jones and Loyd Newman, Our America: Life and Death on the South Side of Chicago (New York: Washington Square Press, 1998).
7. Technically the estimates discussed in this paragraph describe the effects of offering families the chance to move to less distressed neighborhoods through MTO–known in the evaluation literature as the “intent-totreat" effect (ITT)–since not all families assigned to one of the MTO treatment groups move under the program. But if assignment to one of the MTO treatment groups has no (or at least only modest) effects on families who do not actually move under MTO, then the ITT estimates will be proportional to the effects of MTO moves on those who do make such moves–that is, the “effects of treatment on the treated," or TOT–by the fraction of families assigned to the treatment group who move under MTO. For more details about these results and methods see Jeffrey R. Kling, Jeffrey B. Liebman, and Lawrence F. Katz, “Experimental Estimates of Neighborhood Effects," Working Paper 11577 (Cambridge, Mass.: National Bureau of Economic Research, 2005); or Jeffrey R. Kling, Jens Ludwig, and Lawrence F. Katz, “Neighborhood Effects on Crime for Female and Male Youth: Evidence from a Randomized Housing Voucher Experiment," Quarterly Journal of Economics 120, no. 1 (2005): 87-130; and Lisa Sanbonmatsu and others, “Neighborhoods and Academic Achievement: Results from the Moving to Opportunity Experiment," Journal of Human Resources (forthcoming).
8. These are speculations for which there is currently not much empirical evidence. On the other hand, data for MTO participants show that moving to a less disadvantaged neighborhood reduces arrests for violent crime among male youth in the first few years after randomization, an effect that dissipates by three to four years after assignment and gives way to a positive treatment-control difference in property crime arrests; Kling, Ludwig, and Katz, “Neighborhood Effects" (see note 7). There is no detectable evidence for age heterogeneity in MTO effects on scores on achievement tests conducted four to seven years after randomization for children aged six to twenty at the time of the tests; Sanbonmatsu and others, “Neighborhoods and Academic Achievement" (see note 7). There is also not much evidence for age heterogeneity in MTO impacts on arrests among people who were fifteen to twenty-five in the period four to seven years after randomization; Kling, Ludwig, and Katz, “Neighborhood Effects" (see note 7); see also Bernard Harcourt and Jens Ludwig, “Broken Windows Policing: New Evidence from New York City and a 5-City Social Experiment," University of Chicago Law Review 73 (2006): 271–320.
9. Data from the Gautreaux program in Chicago, which moved African American families in public housing to other suburban or urban neighborhoods in the Chicago area, yield suggestive evidence that mobility might have more beneficial effects over a longer term than has been observed for MTO to date; see, for example, the summary in Leonard S. Rubinowitz and James E. Rosenbaum, Crossing the Class and Color Lines: From Public Housing to White Suburbia (University of Chicago Press, 2001). One difference between Gautreaux and MTO is that the former generates more racial integration than does the latter. However, strong conclusions about Gautreaux’s effect on families are complicated by the fact that assignment of families to neighborhoods may not have been random, and the set of families who remained in lower poverty, more racially integrated suburban communities at the time outcomes were measured may not have been a representative sample of all families initially placed in these areas.
10. The effects of increased income transfers to poor families on the life chances of poor children remain unclear; see Susan Mayer, What Money Can’t Buy (Harvard University Press, 1997). A more recent study finds that randomized welfare-to-work experiments that offer income supplements together with work requirements yield bigger gains in children's achievement scores than do work-only programs; Pamela Morris, Greg J. Duncan, and Christopher Rodrigues, “Does Money Really Matter? Estimating Impacts of Family Income on Children's Achievement with Data from Random-Assignment Experiments," Working Paper (New York: MDRC, 2004) These results are not necessarily inconsistent with those from Mayer, since the two studies consider different interventions (cash assistance versus cash assistance plus work requirements). Additional income with work may have beneficial effects by forcing (and enabling) low-income mothers to put their children into more structured child care settings, whereas extra cash in isolation may be devoted to things that improve family well-being but not necessarily child development. Unfortunately this hypothesis cannot be tested directly within the context of the welfare-to-work experiments examined by Morris and coauthors, since no experimental program changes income without changing labor supply (so experimental assignment cannot be used to instrument for an interaction of income with maternal employment). However, suggestive support for this interpretation comes from the fact that Morris and others find combined income and work effects only on children aged two to five and not on those already of school age (six to fifteen).
11. For example, the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 begins as follows: “The Congress makes the following findings: (1) Marriage is the foundation of a successful society. (2) Marriage is an essential institution of a successful society which promotes the interests of children. (3) Promotion of responsible fatherhood and motherhood is integral to successful child rearing and the well-being of children."
12. Douglas J. Besharov, “Welfare Reform–Four Years Later," Public Interest (Summer 2002) (emphasis added).
13. Michael Sokolove, “The Believer," New York Times Magazine, May 22, 2005.
14. Quoted in Dennis R. Hoover, “Charitable Choice and the New Religious Center," Religion in the News (Greenberg Center for the Study of Religion in Public Life, Trinity College, Hartford, Conn.) 3, no 1 (2000).
15. To be clear, Raspberry goes on to note, “The difficulty for me is at the intersection of these [religious] things with government funding," but his remarks quoted above speak to the larger view that religious belief per se has important implications for people’s life chances. Pew Forum on Religion and Public Life, “With Ben Franklin's Blessing: A Primer on “Faith-Based Initiatives,–" www.pewforum.org/events/ index.php?EventID=83 (accessed January 25, 2006).
16. For a good review of this literature, see Jeanne Brooks-Gunn and Lisa Markman, “The Contribution of Parenting to Ethnic and Racial Gaps in School Readiness," Future of Children 15, no.1 (2005): 139–68.
17. For a good review of early childhood programs, see Katherine A. Magnuson and Jane Waldfogel, “Early Childhood Care and Education: Effects on Ethnic and Racial Gaps in School Readiness," Future of Children 15, no.1 (2005): 169–96. Evidence on early childhood programs” long-term effects during adulthood includes Eliana Garces, Duncan Thomas, and Janet Currie, “Longer-Term Effects of Head Start," American Economic Review 92, no. 4 (2002): 999–1012; Frances Campbell and others, “Early Childhood Education: Young Adult Outcomes from the Abecedarian Project," Applied Developmental Science 6, no. 1 (2002): 42–57; Jens Ludwig and Douglas Miller, “Does Head Start Improve Children's Life Chances? Evidence from a Regression Discontinuity Design," Working Paper 11702 (Cambridge, Mass.: National Bureau of Economic Research, 2005); and Lawrence J. Schweinhart and others, Lifetime Effects: The High/Scope Perry Preschool Study through Age 40 (Ypsilanti, Mich.: High/Scope Press, 2005). On tutoring programs, most of what is known comes from a randomized experimental evaluation of perhaps the most famous such program, Big Brothers/Big Sisters; Joseph P. Tierney and Jean Baldwin Grossman, Making a Difference: An Impact Study of Big Brothers/Big Sisters (Philadelphia: Public / Private Ventures, 2000). While this study found encouraging differences in self-reported behavior between tutored and control youth, there remains some uncertainty about whether this evaluation captures tutoring effects on youths” willingness to engage in or self-report problem behaviors. Very little is known about the effects of such tutoring programs on the long-term adult economic outcomes of participants.
18. David Brooks, “Dollars and Sense," New York Times, January 26, 2006, p. A23.
19. For example, the share of Americans claiming that they do not believe in God increased from 1.5 percent in 1988 to 2.9 percent in 2000, but the proportion who have no doubt that God exists also increased, from 64 to 66.2 percent, and the share reporting that they had no religion or denominational preference has been largely unchanged for a decade; Michael Hout and Claude Fischer, “Why More Americans Have No Religious Preference: Politics and Generations," American Sociological Review 67, no. 2 (2002): 165–90.
20. www.whitehouse.gov/government/fbci/guidance/charitable.html.
21. Calculations from the General Social Survey, http://sda.berkeley.edu/7507/quicktables.
22. This review of the research on the relationship between religion and economic well-being relies heavily on Lawrence Iannaconne, “Introduction to the Economics of Religion," Journal of Economic Literature 36, no. 3 (1998): 1465–95, which cites an extensive literature. We do not repeat the citations from that article but add relevant and more recent research.
23. Nigel Tomes, “The Effects of Religion and Denomination on Earnings and the Returns to Human Capital," Journal of Human Resources 19, no. 4 (1984): 472–88.
24. Todd P. Steen, “Religion and Earnings: Evidence from the NLS Youth Cohort," International Journal of Social Economics 23, no. 1 (1996): 47–58.
25. Iannaconne, “Introduction to the Economics of Religion" (see note 22).
26. Ibid.; Alfred Darnell and Darren E. Sherkat, “The Impact of Protestant Fundamentalism on Educational Attainment," American Sociological Review 62, no. 2 (1997): 306–15.
27. Steen, “Religion and Earnings" (see note 24).
28. Bruce Sacerdote and Edward Glaeser, “Education and Religion," Working Paper 8080 (Cambridge, Mass.: National Bureau of Economic Research, 2001). A large body of research shows that social group membership increases with education and that religion is one of the most common forms of social membership. Greater education is not associated with private forms of religious expression, such as prayer and church attendance, but it is positively correlated with other forms of social membership. Thus any relationship between religion and economic outcomes is probably attributable to the fact that more social people are more likely to be religious and more social people are more likely to be successful, not to religion per se.
29. Tomes, using the General Social Survey, shows that about 60 percent of children raised in families with no religion claim to be members of a religion once they are adults, and about 10 percent of individuals who grew up in a family with a religion claim to have no religion once they are adults. This reduces our ability to conclude that the correlation between adults” religion and their income is a good proxy for the correlation between parents” religion and children's income. Tomes, “The Effects of Religion" (see note 23).
30. Jonathan Gruber, “Religious Market Structure, Religious Participation, and Outcomes: Is Religion Good for You?" Working Paper 11377 (Cambridge, Mass.: National Bureau of Economic Research, 2005).
31. As noted above, many policymakers believe that religious parents will set a better example for their children because they will be more likely to marry and remain married. Evidence suggests that individuals who report no religious affiliation have higher divorce rates than individuals reporting any affiliation; Linda Waite and Evelyn Lehrer, “The Benefits from Marriage and Religion in the United States: A Comparative Analysis," Population and Development Review 29, no. 2 (2003): 255–75. Today, which denomination adults adhere to has little relationship with marriage or divorce rates, although historically fewer Catholics and Jews divorced than Protestants. Waite and Lehrer also present evidence that children's own religious participation is associated with a lower probability of substance abuse and juvenile delinquency, less depression, and delayed sexual debut. However, the relatively small effect of religion on these outcomes is unlikely to translate into much of an effect on income in adulthood.
32. This brief review of the literature draws from a review by Donna Ginther and Robert Pollak, “Family Structure and Children's Educational Outcomes: Blended Families, Stylized Facts, and Descriptive Regressions," Demography, November 2004.
33. Eng Seng Loh, “Changes in Family Structure, Attained Schooling and Adult Poverty Status," Social Science Quarterly 77, no. 1 (1996): 145–58.
34. For example, an influential study by Wallerstein and various colleagues (Judith Wallerstein and Sandra Blakeslee, Second Chances: Men, Women and Children a Decade after Divorce (Boston: Houghton Mif- flin, 1996), revised edition; Judith Wallerstein, Julia Lewis, and Sandra Blakeslee, The Unexpected Legacy of Divorce: The 25 Year Landmark Study (Hyperion, 2001) claimed that divorce was extremely detrimental to children, reducing their psychological well-being, their educational attainment, and their ability to form good relationships. This claim was based on a small sample of families in which the parents divorced and whom the authors followed over many years. They did not follow a control group of families in which the parents did not divorce, but they claimed that before the divorce their sample was representative of married-couple families. However, Andrew Cherlin notes that half the fathers and nearly as many of the mothers in this study suffered from a mental health problem for which they had been treated before the study began. These mental health problems probably contributed to both the divorce and the problems experienced by the children and controlling parents” mental health would have reduced the estimated effect of divorce. Of course there could have been other differences between the families in the sample and the average married-couple family that also could have contributed to both divorce and children's outcomes, and these too would have to be controlled to produce unbiased estimates of the effect of divorce. Andrew Cherlin, “Going to Extremes: Family Structure, Children's Well-Being and Social Science," Demography 36, no. 4 (1999): 421-28.
35. Charles Manski and others, “Alternative Estimates of the Effect of Family Structure during Adolescence on High School Graduation," Journal of the American Statistical Association 87, no. 417 (March 1992): 25–37.
36. Timothy Biblarz and Greg Gottainer, “Family Structure and Children's Success: A Comparison of Widowed and Divorced Single-Mother Families,–" Journal of Marriage and Family 62 (2000); Kevin Lang and Jay L. Zagorsky, “Does Growing Up with a Parent Absent Really Hurt?" Journal of Human Resources 36, no. 2 (2001): 253–73; Miles Corak, “Death and Divorce: The Long-Term Consequences of Parental Loss on Adolescents," Journal of Labor Economics 19, no. 3 (2001): 682–715.
37. P. R. Amato and B. Keith, “Parental Divorce and the Well-Being of Children: A Meta-Analysis," Psychological Bulletin (1991) 100: 26–46; Timothy J. Biblarz and Adrian E. Raftery, “Family Structure, Educational Attainment, and Socioeconomic Success: Rethinking the “Pathology of Matriarchy,–" American Journal of Sociology 105, no. 2 (1999): 321–65; Sara McLanahan and Gary Sandefur, Growing Up with a Single Parent: What Hurts, What Helps (Harvard University Press, 1994); Andrew J. Cherlin, P. Lindsay Chase- Lansdale, and C. McRae, “Effects of Divorce and Mental Health through the Life Course," American Sociological Review 63 (1998), 239–49.
38. Anne Case, I-Fen Lin, and Sara McLanahan, “Educational Attainment of Siblings in Stepfamilies," Evolution and Human Behavior 22, no. 4 (2001): 269–89.
39. McLanahan and Sandefur, Growing Up with a Single Parent (see note 37); Biblarz and Raftery, “Family Structure" (see note 37); Roger Wojtkiewicz, “Simplicity and Complexity in the Effects of Parental Structure on High School Graduation," Demography 30, no. 4 (1993): 701–17; Scott Boggess, “Family Structure, Economic Status, and Educational Attainment," Journal of Population Economics 11, no. 2 (1998): 205–22.
40. Put another way, we would expect the interaction of indicator variables for married mother and employed mother to be negative, and the coefficient on married mother and employed mother to be positive. Estimates like these would, of course, require one to address all the problems associated with the meaning of “married" and “single" parents discussed above.
41. A few studies estimate the effect of maternal employment on children's educational outcomes during adolescence. One finds that among white children in married-couple families, “upper-class" and “middle- class" boys got lower grades when their mothers worked, and their grades were lower still if the mother worked when the boy was of preschool age. The effects of maternal work were much less for girls and for children of lower social classes. See Karen Bogenschneider and Laurence Steinberg, “Maternal Employment and Adolescents” Academic Achievement: A Developmental Analysis," Sociology of Education 67, no.1 (1994): 60–77. Another finds no relationship between maternal employment and the school achievement of a small sample of eighth-grade children who lived with two (not necessarily biological) parents; see Sharon Paulson, “Maternal Employment and Adolescent Achievement Revisited: An Ecological Perspective," Family Relations 45, no. 2 (1996): 201–08. Controlling parents” marital status, another study finds that children whose mothers were not employed had higher math scores than children whose mothers were employed full time. Children whose mothers worked part time had higher math scores than children whose mothers were not employed or who were employed full time. See Chandra Muller, “Maternal Employment, Parental Involvement, and Mathematics Achievement among Adolescents," Journal of Marriage and Family 57 (1995): 85–100. These studies all control some family background characteristics, but none tries to control unobserved heterogeneity or to distinguish the effect of maternal employment for married and unmarried mothers.
42. Suzanne James-Burdumy, “The Effect of Maternal Labor Force Participation on Child Educational Attainment" (Mathematica Policy Research, December 1999); Wen-Jui Han, Jane Waldfogel, and Jeanne Brooks- Gunn, “The Effects of Early Maternal Employment on Later Cognitive and Behavioral Outcomes," Journal of Family and Marriage 63 (2001): 336–54; Jennifer Hill, Jane Waldfogel, and Jeanne Brooks-Gunn, “Assessing the Differential Impacts of High-Quality Child Care: A New Approach for Exploiting Post-Treatment Variables," Journal of Policy Analysis and Management 21, no. 4 (2002): 601–27; Jeanne Brooks- Gunn, Wen-Jui Han, and Jane Waldfogel, “Maternal Employment and Child Cognitive Outcomes in the First Three Years of Life: The NICHD Study of Early Child Care," Child Development 73, no. 4 (2002): 1052–72; Jane Waldfogel, Wen-Jui Han, and Jeanne Brooks-Gunn, “The Effects of Early Maternal Employment on Child Cognitive Development," Demography 39, no. 2 (2002): 369–92; Charles L. Baum, “Does Early Maternal Employment Harm Child Development? An Analysis of the Potential Benefits of Leave Taking," Journal of Labor Economics 21, no. 2 (2003): 409–48; Christopher J. Ruhm, “Parental Employment and Child Cognitive Development," Journal of Human Resources 39, no. 1 (2004): 155–92.
43. Some studies find that mother’s work is associated with higher cognitive test scores; see Debrah Vandell and Janaki Ramanan, “Effects of Early and Recent Maternal Employment on Children from Low-Income Families," Child Development 63, no. 4 (1992): 938–49; Toby Parcel and Elizabeth G. Menaghan, “Early Parental Work, Family Social Capital, and Early Childhood Outcomes," American Journal of Sociology 99, no. 4 (1994): 972–1009; Kristin Moore and Anne K. Driscoll, “Low-Wage Maternal Employment and Outcomes for Children: A Study," Future of Children 7, no. 1 (1997): 122–27. Others find that it is associated with lower scores; see Arleen Leibowitz, “Parental Inputs and Children's Achievement," Journal of Human Resources 12, no. 2 (1977): 242–51; Frank Stafford, “Women's Work, Sibling Competition, and Children's School Performance," American Economic Review 77, no. 5 (1987): 972–80; Frank Mott, “Developmental Effects of Infant Care: The Mediating Role of Gender and Health," Journal of Social Issues 47, no. 2 (1991): 139–58; Jay Belsky and David Eggebeen, “Early and Extensive Maternal Employment and Young Children's Socioemotional Development," Journal of Family and Marriage 53 (1991): 1083–110. Some of the studies find that the estimates differ depending on the timing of work or the specific group or outcome analyzed; see Sonalde Desai, P. Lindsay Chase-Lansdale, and Robert T. Michael, “Mother or Market? Effects of Maternal Employment on the Intellectual Ability of 4-Year-Old Children," Demography 26, no. 4 (1989): 545–61; Nazli Baydar and Jeanne Brooks-Gunn, “Effects of Maternal Employment and Child-Care Arrangements on Preschoolers” Cognitive and Behavioral Outcomes: Evidence from the Children of the National Longitudinal Survey of Youth," Developmental Psychology 27, no. 6 (1991): 932–45; Francine Blau and Adam J. Grossberg, “Maternal Labor Supply and Children's Cognitive Development," Review of Economics and Statistics 74, no. 3 (1992): 474–81; Parcel and Menaghan (cited above in this note); Theodore Greenstein, “Are the “Most Disadvantaged” Children Truly Disadvantaged by Early Maternal Employment?" Journal of Family Issues 16, no. 2 (1995): 149–69; Peter Barglow and others, “Developmental Follow-up of 6–7 Year Old Children of Mothers Employed during Their Infancies," Child Development and Human Development 29, no. 1 (1998): 3–20.
44. One study finds that the negative impact of first-year employment is temporary; see Elizabeth Harvey, “Short-Term and Long-Term Effects of Early Parental Employment on Children of the National Longitudinal Survey of Youth," Developmental Psychology 35, no. 2 (1999): 445–59. Others suggest that some effects persist; see Han, Waldfogel, and Brooks-Gunn, “The Effects of Early Maternal Employment" (see note 42); and Waldfogel, Han, and Brooks-Gunn, “The Effects of Early Maternal Employment" (see note 42).
45. This calculation assumes that having a mother who works in the first year of a child's life reduced cognitive test scores by about a quarter of a standard deviation (Waldfogel, Han, and Brooks-Gunn, “The Effects of Early Maternal Employment") (see note 42) and that a 1 standard deviation decline in test scores is associated with a 16 percent increase in wages (Winship and Korenman, “Economic Success") (see note 3).
46. Joseph Altonji and Thomas Dunn, “An Intergenerational Model of Wages, Hours and Earnings," Journal of Human Resources 35, no. 2 (2000): 221–58.
47. Rebecca Blank, “Evaluating Welfare Reform in the United States," Journal of Economic Literature 40 (2002): 1105–66; Bruce Meyer and Dan Rosenbaum, “Welfare, the Earned Income Tax Credit, and the Labor Supply of Single Mothers," Quarterly Journal of Economics 116, no. 3 (2001): 1063–114.
48. See for example the review in Robert Moffitt, Means-Tested Transfer Programs in the United States (University of Chicago Press, 2003).
49. Jeffery Grogger and Lynn Karoly, Welfare Reform: Effects of a Decade of Change (Harvard University Press, forthcoming).
50. For details, see Jens Ludwig and Susan Mayer, “Ending Poverty by Changing “Culture”: The Poverty- Prevention Paradox," Working Paper, Georgetown University (2006).
51. For example, Greg J. Duncan and Saul Hoffman, “Teenage Underclass Behavior and Subsequent Poverty: Have the Rules Changed?" in The Urban Underclass, edited by Christopher Jencks and Paul Peterson (Brookings, 1991), pp. 155–74. Note that encouraging women to complete high school and delay fertility until marriage can greatly reduce their own chances of living in poverty, particularly among African Americans.
52. www.childtrendsdatabank.org/indicators/75UnmarriedBirths.cfm (accessed February 22, 2006).
53. Isabel Sawhill and others, “Results to Date," in Welfare Reform and Beyond: The Future of the Safety Net, edited by Sawhill and others (Brookings, 2002), pp. 9–19.
54. www.religioustolerance.org/us_rel1.htm (accessed February 22, 2006).
55. Christopher Jencks and others, Inequality: A Reassessment of the Effect of Family and Schooling in America (New York: Basic Books, 1972).
This feature: Ludwig, J. and Mayer, S. (2006) “Culture" and the Intergenerational Transmission of Poverty: The Prevention Paradox. From The Future of Children, Vol.16 No.2, Fall 2006, a publication of The Woodrow Wilson School of Public and International Affairs at Princeton University and The Brookings Institution.
