Abstract
This article discusses the consequences of family composition for poverty and income and its implications for policy. Marriage rates are declining, rates of nonmarital births are increasing (both poverty-increasing), while families are smaller, and there are more working mothers (both poverty-decreasing). Marriage remains less likely and nonmarital births more common for blacks than for whites and Hispanics, though even among whites, 36 percent of births were to unmarried mothers by 2011. On the other hand, divergent patterns across education groups are more common: marriage rates have continued to fall, but not for women with college degrees. Men’s earnings have fallen, and, after an increase, women’s have also declined—though less so for those with bachelor’s degrees. The article also discusses policy responses designed to reduce nonmarital childbearing (potentially reducing the number of children and families at high risk of poverty) and to help single-mother families (reducing the risk of poverty faced by such families).
Keywords
We live in a time of both dramatic changes in family composition and childrearing and growing research on and understanding of these family changes and their consequences. As often happens when researchers, public intellectuals, and policy-makers try to draw out the policy implications of research findings, the nation now seems to be divided into two camps—are the ongoing changes in family composition and their impacts on children a natural evolution of families to meet modern economic and social conditions or are they the fifth horseman of the apocalypse? As Daniel Patrick Moynihan famously said, “Everyone is entitled to their own opinions, but not their own facts.” So we intend to put a bevy of facts on the table before demonstrating that the authors of this article embody the split in the country over the policy implications of those facts. In considering the consequences of family composition for poverty and income, we begin with marriage, proceed to fertility, and then examine nonmarital births. We then discuss implications for policies, especially those designed to reduce nonmarital births and support single-mother families.
Marriage Rates
The most important fact about marriage rates can be summarized tersely: decline. With one exception examined below, every ethnic and education group in the nation has steadily become less likely to be married, since 1970. Figure 1 shows women’s marriage rates by age (1a), race/ethnicity (1b), and education (1c). Marriage rates by age show a consistent pattern of decline for every age group. The steeper decline for lower ages shows the related delay in marriage. More than 60 percent of women age 20 to 24 were married in 1970, but by 2010 this figure had fallen to 16 percent. The steepness of the decline in marriage rates is reduced as women get older, showing that many women simply delay marriage. Still, a 23 percent decline in the percentage of women who are married at ages 40 to 44 is not negligible. The line graphs showing the declines in marriage rates by women’s race 1 and education level reflect differences in levels (e.g., lower levels for blacks) but similar trends. The decline in marriage rates for women with a bachelor’s degree, however, stopped around 1990 and held steady for the next two decades, thereby differing sharply from the continuing declines for the other education groups. That women with a bachelor’s degree maintained about a 70 percent marriage rate for two decades, while women with less education continued to experience declines, suggests that women who would have been more successful at supporting themselves without a husband were nonetheless precisely the ones most likely to marry. Perhaps these women realize, and base their decisions about family composition on their knowledge, that marriage is good for their own and their family’s economic well-being and for their children’s social and intellectual development. Or perhaps the lower marriage rates observed for women who, all else equal, have more to gain from marriage are simply a sign that these women have fewer opportunities to marry.
Women’s Marriage Rates by Age
Women’s Marriage Rates by Race/Ethnicity
Women’s Marriage Rates by Education
Fertility and Family Structure
In addition to major declines in marriage rates for nearly every demographic group, there have been major changes in family composition. As shown in Figure 2, the percentage of women who live in a household with their children and are married at age 35 has declined from 78.1 to 51.1. All the other family forms have increased in percentage terms: single with children from 9.3 to 20.5, for example. A major development in fertility over the past decade is the decline in the number of women who have children. The percentage of women who were living with children at age 35 declined from 87.3 percent in 1970 to 73.4 percent in 2010, a decline of 15.9 percent. Similarly, there has been a 30 percent decline in the average number of coresident children women have at age 35, from 3.1 children in 1970 to 2.2 children in 2010. Nearly all the decline took place between 1970 and 1990.
Changes in Women’s Family Structure at Age 35, 1970–2010
The declines in the number of children occurred among women of all education levels and racial/ethnic groups between 1970 and 1990. As with the overall decline, the decline slowed down considerably for all education and racial groups between 1990 and 2000. For example, Hispanic women had the highest number of children (an average of 3.0 in 1970, falling to 2.0 in 1990, and then 1.8 by 2010), while whites had the fewest (2.7 in 1970, falling to 1.6 in 1990, and then 1.5 in 2010). Black women experienced the largest declines: from 3.0 children in 1970 to 1.6 by 2010. The patterns of decline and stabilization were similar across the four education groups, with consistent and substantial declines between 1970 and 1990, followed by small declines for the least educated women, and stability or small increases for more educated women between 1990 and 2010. Those with less than a high school diploma always had the highest average number of children, and those with a bachelor’s degree or more always had the fewest. While more educated mothers have, on average, more resources to sustain larger families, they also have more opportunities in the formal labor market that may serve as an incentive to have fewer children so they can work more and more consistently.
An important feature of changes in fertility over the period 1970 to 2010 is the dramatic increase in the share of births that occurred outside marriage. After the mid-1950s, the percentage of births to unmarried mothers increased almost every year until 2011. There was a slowdown in the early to mid-1990s, followed by another substantial increase until about 2008. Over the entire period between 1940 and 2011, the percentage of babies born outside marriage increased from 3.8 percent to nearly 41 percent. This increase reflected the combined effect of declines in marriage, declines in married women’s fertility, and increases in unmarried women’s fertility.
Figures 3a and 3b demonstrate that although nonmarital births have increased for all racial and education groups since 1970, the patterns differ substantially among the groups. In 1970, 37.6 percent of black births, and only 5.7 percent of white births, were to unmarried mothers. By 1994, the percentage of births to unmarried mothers had risen by about 20 percentage points for whites and 35 percentage points for blacks—but this represented a much larger relative change for whites than for blacks. The gap between the proportion of white and black births to unmarried mothers narrowed further after 1994, as the proportion of black births to unmarried mothers rose modestly (by about 2 percentage points) relative to the increase for whites (11 percentage points). Nonetheless, a significant gap remains, with 36 percent of white babies and 72 percent of black babies born to unmarried mothers in 2010. The review of birth certificates produced by Vital Statistics, which are the origin of our data on births, did not record whether people were Hispanic until 1990, but between 1990 and 2010 the percentage of Hispanic births that were outside marriage increased from 36.7 to 53.4 percent.
Percentage of Births to Unmarried Women by Race/Ethnicity, 1970–2010
Percentage of Women Age 35 Who Are Never-Married Mothers by Education Level, 1970–2010
Figure 3b, which traces the rise in the percentage of women with various levels of education who are never-married mothers at age 35, starkly portrays the large differences across education groups. The rate for all groups rose substantially, but by 2010, 17 percent of mothers with less than a high school degree were never-married mothers, compared to about 3 percent of women with a bachelor’s degree.
Children born to unmarried parents are substantially less likely to live in stable two-parent households. Many unmarried parents are cohabiting or in a romantic relationship at the time of the birth, and many parents have some plans to marry (Carlson, McLanahan, and England 2004; McLanahan 2009). However, nearly half of cohabiting relationships dissolve within three years (Kamp Dush 2011), and even five years later, only about one in five couples in these “fragile families” has married (McLanahan 2009). This relationship instability contributes to high levels of family complexity. Mothers and fathers who have children outside marriage are more likely than their married counterparts to go on to have children with other partners. Some estimates suggest that more than two in three children born to unmarried parents will be part of complex families, sharing one or both parents with half siblings when their parents have other children with other partners (Cancian, Meyer, and Cook 2011). Complex families potentially add new adults with additional resources to the mix—for example, a mother may be more likely to receive support from at least one father if she has child support orders from multiple fathers (Meyer, Cancian, and Cook 2005). But a father may also be less likely to maintain contact with his children and contribute to a household that includes his children and children of other fathers (Berger, Cancian, and Meyer 2012; Meyer and Cancian 2012).
Employment and Earnings
The implications of changes in family structure for income and poverty depend in large part on men’s and women’s employment and earnings patterns. Since 1970, the median earnings of prime-age men (ages 25 to 55), including those with zero earnings, have declined by more than 30 percent. 2 This decline reflects reduced employment rates as well as lower median earnings for working men. In contrast, prime-age women’s median earnings have risen dramatically, reflecting both large increases in employment and higher earnings for working women. In 1970, median prime-age male earnings were 22 times those of prime-age women; in 2010, they were just 1.7 times higher. Even among those who actually worked, the earnings gap has declined; working men earned 2.3 times as much as working women in 1970, but only 1.3 times as much in 2010. Changes in men’s and women’s work and earnings may themselves reflect changes in family structure; fewer women can depend on a husband’s earnings, and fewer men have primary financial responsibility for a wife and children. The shrinking earnings gap has also dramatically changed the economic consequences of married and unmarried parenting: single mothers need not be poor, and access to a father’s earnings, while important, is a less certain income source than in the past.
While the earnings gap between men and women declined across all education and racial groups, the underlying changes vary substantially. Figure 4 shows the median earnings of prime-age men (4a) and women (4b) by education (including those who did not have earnings). The most educated men had median earnings of about $60,000 throughout most of the period, while the most educated women’s earnings more than doubled, from about $18,000 to more than $38,000. The earnings gap for men and women with a bachelor’s degree narrowed both because more women now work and because their average earnings have increased. In contrast, among those with less than a high school degree, just over half of women had no earnings in both 1970 and 2010, for a median of zero, while the median earnings of men with less than a high school degree fell by more than 60 percent, from about $36,000 to $14,500; thus, the gap between men’s and women’s earnings narrowed for the less educated because of men’s precipitous decline. For those with a high school degree or some college, both a significant decline in men’s median earnings and a substantial increase in women’s earnings contributed to reducing the earnings gap between 1970 and 1990.
Median Earnings of Males, by Education
Median Earnings of Females, by Education
While there is a substantial decline in the earnings gap between men and women overall, patterns differ by race and ethnicity. For example, in 2010, men’s median earnings were still about 75 percent higher than women’s for whites ($39,000 relative to $22,000 for men and women, respectively), and about 90 percent higher than women’s for Hispanics ($22,000 and $12,000, respectively). The earnings gap between men and women is far smaller for blacks. In fact, by 2010, the median earnings gap was just over 10 percent ($21,000 for men and $19,000 for women), reflecting higher employment rates for black women that largely offset a relatively modest earnings advantage for employed black men. 3 Thus, for black families, especially, the negative economic consequences of single motherhood appear to have declined substantially over time. Of course, as we discuss further below, marriage, employment, and fertility decisions are intertwined, making it more difficult to sort out the causal story and the implications for policy.
Poverty
The decline in marriage and increase in nonmarital births, all else equal, will increase poverty. On the other hand, the reductions in family size will reduce (needs-adjusted) poverty, and the declining gender earnings gap will reduce poverty rates for single-mother families relative to married-couple families. But the patterns of family formation and employment are unevenly distributed across racial groups, contributing to different levels of poverty across education and racial groups.
Poverty rates vary greatly with family composition, education level, and race and ethnicity. We examine the differences in trends in the official poverty rates by each of these variables within married-couple and female-headed families. Families headed by a single mother in 1970 faced poverty rates ranging from 57 percent for those with less than a high school degree to about 30 percent for those with a high school education or some college. Very few single mothers were college graduates in 1970 (6 percent in our sample), but their poverty rate was just 10 percent. Families headed by married couples in 1970 had poverty rates below 4 percent, except those headed by a high school dropout (about 12 percent). The differential poverty rates by marital status largely persisted within education groups, even as poverty increased for those without a high school degree (to 72 percent for single-mother families and 32 percent for married parents), or only high school (to 47 percent for single-mother families and 13 percent for married parents), and less so for those with a bachelor’s degree (13 percent for single mothers and 2 percent for married couples).
Comparing poverty rates by family structure and race/ethnicity again highlights the key role of marriage: married-couple families have substantially lower rates of poverty within all three racial/ethnic groups (Figure 5). Families headed by white single mothers—the least likely to be poor—were nonetheless almost twice as likely to be poor in 1970 as those of black and Hispanic married couples (and five times as likely to be poor as white married-couple families). Poverty rates fell substantially among black and Hispanic single-mother families in the 1990s, and then increased in most years in the 2000s.
Poverty Rates among Prime-Age-Headed Families with Children by Family Structure and Race, 1970–2011
Government Programs
Income from earnings is generally key to avoiding poverty: poverty rates for children from families headed by a single mother who does not work have been above 70 percent every year since the early 1970s and between 80 and 90 percent during most of those years. By contrast, the poverty rate for families headed by a lone mother who works full time has hovered around 20 percent since 1970 with only modest ups and downs. But both the federal and state governments support substantial transfer programs, including those programs for which families must have income below a given cutoff to qualify for the benefit. Federal means-tested payments of this type were about $762 billion in 2011 (Congressional Research Service 2012). If state and local means-tested spending is added, the total for means-tested spending is around $1 trillion per year (Haskins 2012). 4
Much of this means-tested spending goes to poor and low-income working families either to help them escape poverty or raise them further above the poverty line. In fact, arguably the most successful poverty reduction strategy the nation has developed is to encourage low-wage work by poor mothers and then to subsidize their earnings with the Earned Income Tax Credit (EITC) and the Additional Child Tax Credit, food stamps, child care, health insurance, and other benefits. Analysis by Richard Bavier shows that in 1990, lone mothers in the second quintile of income among female-headed families (between about $15,000 and $29,000 per year) had average earnings of about $13,000 (all figures in 2010 dollars), but many did not work. 5 They received about $4,300 in government benefits, bringing their average income to $17,300, which left most of them and their children well below the poverty level. By contrast, in part due to the work requirements of the welfare reform legislation, 10 years later, many more low-income mothers worked. As a result, their earnings increased substantially to nearly $17,700. But even at this level, they still receive government benefits of more than $4,300, on average, bringing them to $21,300 in income, above the poverty level on average.
What to Do
The facts are clear: marriage rates are declining, rates of nonmarital births are increasing (both poverty-increasing), families are smaller, and there are more working moms (both poverty-decreasing). Marriage remains less likely and nonmarital births more common for blacks than for whites and Hispanics, though even among whites, 36 percent of births were to unmarried mothers by 2011. On the other hand, divergent patterns across education groups are more common: marriage rates have continued to fall, but not for women with college degrees. Men’s earnings have fallen, and, after an increase, women’s have also declined—though less so for those with bachelor’s degrees.
The causal story behind these facts is substantially less straightforward. Are women marrying less because they are working and earning more, making marriage less attractive? Or has a decline in the availability of potential husbands with earnings that could support a family made steady employment more of a necessity? Are more children born to unmarried mothers because marriage is deemed less important, or less attainable? In this section, we highlight selected points of debate and their implications for policy. We focus in particular on policies to reduce nonmarital childbearing (potentially reducing the children and families at high risk of poverty) and to help single-mother families (reducing the risk of poverty faced by such families). In each case we aim to clarify when we agree on the problem, the solution, both or neither.
Reduce the number of single mothers
For many analysts and politicians, the solution to the demise of the traditional family and rise of complex families is to reduce nonmarital births and to increase marriage rates. In a simulation analysis based on Current Population Survey (CPS) data, Sawhill and Haskins (2003) found that the effect of increasing the marriage rate until it reached the 1970 level would be to reduce poverty by a little less than 30 percent. One might object that changes in marriage, fertility, and employment are interrelated (Cancian and Reed 2009), making a return to the marriage rates of 1970, holding women’s employment and fertility at contemporary levels, unrealistic. But putting that argument aside, what are the prospects for reducing the four-decade-long trend of growth in female-headed families?
Reducing nonmarital births
While the percentage of births to unmarried women has increased dramatically, the rate of teen births has declined from 61.8 per 1,000 women aged 15 to 19 in 1991 to 34.4 per 1,000 in 2010 (Hamilton and Ventura 2012). Although it is difficult to identify the specific causes of this decline, the Department of Health and Human Services has identified thirty-one model programs that have been shown by rigorous research to reduce teen sexual activity or teen pregnancy (Administration for Children and Families 2013). Sawhill and her colleagues have identified two additional cost-beneficial ways to reduce unintended pregnancy: mass media campaigns that encourage men to use condoms and expanded access for women to contraception provided by Medicaid. But their estimates suggest that the combined impact of these policies would reduce nonmarital births by only about 3 percent (Sawhill, Thomas, and Monea 2010). Reducing nonmarital births by 3 percent would still leave more than 1.6 million nonmarital births a year. It seems reasonable to conclude that nonmarital births will continue to swell the ranks of female-headed families for the foreseeable future.
Increasing marriage rates
The Bush administration mounted a comprehensive strategy to test the efficacy of marriage education and associated services through rigorous research demonstration projects involving multisite, random assignment demonstrations. The demonstration programs were of two types. The Building Strong Families (BSF) study consisted of programs in eight sites that offered relationship skill training and other services to unmarried parents. Averaging across all eight sites, the evaluation found no significant impacts on either the couples’ relationship quality or the number of couples who remained together or got married after 15 months (Wood et al. 2012). The Oklahoma site, however, had a pattern of positive effects on the couples that included relationship skills, romantic involvement, coparenting, and father involvement. The Baltimore site had a few negative effects, including an increase in family violence. At the 36-month follow-up, both the positive effects at 15 months from the Oklahoma program and the negative effects from the Baltimore program had mostly faded. However, the Oklahoma program was again a partial exception: at the 36-month follow-up, 49 percent of the children in the Oklahoma program group still lived with both their parents as compared with 41 percent of control children.
The second Bush demonstration program, Supporting Healthy Marriage (SHM), was similar in many respects to BSF. Only results from the 12-month survey are publicly available (Hsueh et al. 2012), and while SHM produced a fairly consistent pattern of positive impacts on marital behaviors and emotions, there were no impacts on the couples staying married. Taken together, these two marriage programs provide very modest support for any claim that good programs will have an impact on marriage rates or even keeping parents together. Perhaps the most positive outcome was the finding that the Oklahoma program increased by about 20 percent the share of children living with both parents 36 months after the program began.
We both support the expansion of programs designed to reduce unintended nonmarital births and are confident that they would lead to a modest reduction in the incidence of female-headed families. Although we disagree about how much money should continue to be invested in the relatively expensive (about $10,000 per couple) healthy marriage programs, neither of us thinks we are on the cusp of learning how to increase the share of married-couple families. Haskins would recommend spending a few tens of millions of dollars trying to improve the healthy marriage programs, especially by trying to replicate the activities and protocols being followed by the Oklahoma program.
Helping families headed by single mothers
Neither of us expects to see a substantial decline in the proportion of children born to unmarried parents going forward. We must therefore analyze what can be done to improve outcomes for children in single-parent families.
Federal and state means-tested programs disproportionately provide assistance to single mothers and their children. While total expenditures on these programs have grown, there has been a shift away from income support targeting those with the most limited resources toward work supports that raise the total income of low-income families with working parents (Congressional Research Service 2012; Moffitt and Scholz 2010). This shift has simultaneously increased the returns to working (for example, by supplementing earnings with the EITC) and reduced the options and supports available to mothers who do not work and their children (for example, by reducing cash welfare payments and imposing time limits and other restrictions). Because work is the primary way for single-mother families to escape poverty, means-tested programs that support work should be sustained and, ideally, enhanced. However, the shift to support working families has left the most vulnerable, including a growing number of children in deep poverty, worse off (Ben-Shalom, Moffitt, and Scholz 2011; Shaefer and Edin 2012).
Despite some gains, women generally earn less than men, and most less-educated single mothers will continue to need work supports if they are to avoid poverty. The EITC and the refundable part of the Additional Child Tax Credit are important work supports and should be sustained. More can and should be done to improve access to high-quality childcare that simultaneously allows mothers to work and children to receive early education. These supports can play an important role in allowing single mothers who work to sustain their families, but they generally do little to help mothers who cannot, or do not, work for pay. We know less about how to help these families, often seen as “hard to serve” and facing multiple barriers, including their own health and mental health limitations, as well as the need to care for children with health issues (Corcoran et al. 2001; Lein and Schexnayder 2007; Smith et al. 2002). A different set of policy tools will be needed to help these families make, and sustain, progress.
With about half of children spending at least some time living apart from one of their parents, an effective child support enforcement system is important, especially when the resident parent (usually the mother) has limited earnings. Unfortunately, low-income mothers often have children with men who have similarly limited prospects. And while they may have had children with multiple partners and thus owed support from more than one father, those fathers often have obligations to multiple families and may face major barriers in the labor market. The child support system has become increasingly effective in cases in which the father has regular formal employment, but many fathers make irregular payments due to unstable employment, incarceration, and other barriers (Ha, Cancian, and Meyer 2011). If child support is to become a reliable source of support, effective public policies that encourage and enable fathers to work and earn more will be needed (Cancian, Meyer, and Han 2011). An EITC for noncustodial parents who are providing support to their children is a good place to start (Carasso et al. 2008; Wheaton and Sorensen 2010). And there is growing evidence of the advantages of policies directing child support to families rather than diverting it to offset public costs (Wheaton and Sorensen 2007; Cancian, Meyer, and Caspar 2008). We are both encouraged by the recent expansion of federal efforts to test programs that include work supports, relationship or parenting programs, and in some cases child support services (i.e., the National Child Support Noncustodial Parent Employment Demonstration [CSPED] and the Parents and Children Together [PACT] evaluation of selected Responsible Fatherhood and Healthy Marriage programs).
Conclusion
There appears now to be enough research to support several conclusions that most social scientists, policy analysts, and politicians should at least tentatively accept. The first is that the changes in family composition and many of the various cultural and economic influences shaping these changes are now at least in their fourth decade. The upshot is fewer married-couple families, more nonmarital births, and more children being reared in female-headed families. These changes are in all likelihood here to stay, especially given our second conclusion: although there have been some modestly successful policies that have at least partially offset some of these worrisome trends in family composition, especially in reducing nonmarital births among teenagers, taken together none of our policies are very potent at this stage in their development. Third, there is agreement that children living in female-headed families are at elevated risk of living in poverty and facing a set of challenges to their development. In an age in which returns to human capital are growing, the nation is allowing many children to be raised in circumstances that undercut the chances for success in reducing poverty and increasing opportunity in the long run.
These facts do not mean that further research and program development will not yield more promising results. The authors of this article differ on the level of public resources we would devote to developing better policies and programs to reduce nonmarital births and increase marriage rates. But neither of us claims to know enough to expand policies, programs, and spending that would increase the share of children being reared in married-couple families. The most intuitive prediction from historical trends in family composition, and the indifferent success of the policies that we have devised and tested so far to offset these trends, is that the nation will face continuing high and perhaps even increased rates of children living apart from at least one of their biological parents. The authors also differ on the extent to which we believe single-parent childrearing necessarily undermines human potential, as well as the extent to which public investments can mitigate single-parent childrearing. But we both agree that policies that support working mothers’ efforts to provide for their families should be sustained and that more should be done to require and enable more fathers to support their children financially. If simple, stable, two-parent families are in decline, we owe it to the next generation to develop and test programs and policies that reduce the risks associated with the family complexity that inevitably results from the lack of these types of families.
Footnotes
Note:
This article was prepared for the IRP Family Complexity, Poverty and Public Policy Conference. The authors thank Marcia Carlson, Daniel R. Meyer, Isabel Sawhill, and an anonymous reviewer for comments on earlier drafts; Deborah Reed for advice and consultation on related analysis; Kimberly G. Howard of the Brookings Institution for skilled data analysis; and Emily J. Warren of the Institute for Research on Poverty for additional research assistance.
Notes
Maria Cancian is a professor of public affairs and social work, and an affiliate and former director of the Institute for Research on Poverty at the University of Wisconsin–Madison.
Ron Haskins codirects the Brookings Center on Children and Families and Budgeting for National Priorities Project.