Skip over navigation

Journal Issue: Marriage and Child Wellbeing Volume 15 Number 2 Fall 2005

For Love and Money? The Impact of Family Structure on Family Income
Adam Thomas Isabel Sawhill

Does Marriage Reduce Child Poverty and Increase Family income? A Closer Look at the Evidence

Earlier, we described a host of different ways in which particular living arrangements might affect families' economic resources. We concluded that marriage, especially, and cohabitation, to a lesser extent, produce economic benefits for children. We then presented data showing that the incomes of the three groups tracked our expectations. But we have not shown that a particular living arrangement affects income. As we noted earlier, even if family composition itself had no real effect on income, the incomes and poverty rates of married-parent, cohabiting, and lone-parent families might differ widely because of “selection.” Perhaps those people with the most economic resources are the most likely to marry, those with relatively limited resources are the most likely to cohabit, and those with the fewest resources are the most likely to become lone parents.

In light of this problem, how can one be certain whether family structure is helping to drive the differences in the incomes of married-parent, cohabiting, and lone-parent families? The short answer is that we cannot be absolutely sure. But researchers have tried to account for the phenomenon of selection, and their findings generally suggest that living arrangements do have an impact on families' incomes. In the following sections, we review this evidence by summarizing studies that have estimated the economic consequences of divorce, of out-of-wedlock childbearing, and of marriage.

The Estimated Effects of Divorce
In their 1994 book, Growing Up with a Single Parent, Sara McLanahan and Gary Sandefur chart income changes in stable and unstable families during a child's adolescence (a family is “unstable” if parents divorce or separate during that time). McLanahan and Sandefur find that among whites, the incomes of stable families increased from an average of $61,559 when the child was twelve to an average of $66,696 when the child was seventeen, while the incomes of unstable families dropped from an average of $62,367 when the child was twelve to an average of $36,662 when the child was seventeen.20 Among blacks, the incomes of stable families increased from an average of $39,040 when the child was twelve to an average of $40,934 when the child was seventeen, while the incomes of unstable families fell from an average of $28,197 when the child was twelve to an average of $18,894 when the child was seventeen.

The large difference in the average initial incomes of the stable and unstable black families makes it difficult to interpret the income data on black families. Perhaps there were systematic differences between families that this analysis did not capture. Given the substantially lower average initial income level of the unstable black families, it is difficult to rule out the possibility that economic distress actually induced marital disruption for many of these families. But for white families, the similarity between the two groups' initial incomes and the magnitude of the difference in their later incomes are rather striking.21 Thus, we are more confident about the generalizability of these findings for whites than for blacks.

The Estimated Effects of Out-of-Wedlock Childbearing
During the 1960s and 1970s, most of the rise in single parenthood was related to divorce. But over the past quarter-century, by far the most important cause of the rise in single-parent families has been out-of-wedlock childbearing.22 A large body of evidence demonstrates that children born to unmarried mothers are more likely to be poor than are other children.23 Some studies have attempted to control for the possibility that these mothers would have been poor regardless of whether they had had a child outside marriage. For instance, one group of researchers finds that even after controlling for race, family background, age, education, and employment status, women who have had a child out of wedlock are between 2 and 2.7 times more likely to be poor than other women.24

Over the past decade the growth in the share of children born outside marriage has slowed dramatically, in part because of a sharp decline in teen pregnancy and birth rates. Births per 1,000 teens aged fifteen to nineteen fell from 61.8 in 1991 to 41.7 in 2003. This decline has substantially reduced the number of children living in poor singleparent families. One recent study finds that the number of poor children would have increased by almost half a million and the 2002 poverty rate for children under six would have been nearly a full percentage point higher had teen birth rates not declined.25

The Estimated Effects of Marriage
Other studies have assessed the extent to which the decline in marriage and the spread of single parenthood over time have contributed to the growth of child poverty. Some of these studies use “shift-share techniques” to address questions of the following sort: “if the share of children living in single-parent families had remained constant since the 1960s, how would this have affected child poverty?” With some exceptions, these studies generally find that most, and in some cases all, of the increase in child poverty over the past thirty to forty years can be explained by changes in family structure.26 Some of these studies, however, find that growing economic inequality and limited income growth can also explain an important portion of the increase in child poverty during this period. Indeed, John Iceland finds that the association between economic factors and child poverty has in fact been stronger than the association between family structure and child poverty over time. He concludes that this was particularly true during the 1990s, when he finds no significant association between family structure and child poverty.27

Moreover, to say that changes in living arrangements can explain poverty increases is not necessarily the same as saying that they cause these increases. Indeed, one could argue that it is unreasonable to assume, as most of these analyses implicitly do, that the poverty rates of two-parent families would remain the same if many single parents were to marry. In a study that addresses this issue by controlling for family attributes that might affect families' economic well-being, Robert Lerman finds that living in a married-parent family confers large economic benefits relative to living in a single-parent family and more modest but still significant benefits relative to living in a cohabiting family. In one analysis, he finds that living in a married-parent family raises needs-adjusted income by 65 percent relative to living in a lone-parent family and by 20 percent relative to living in a cohabiting household.28

Even if marriage has historically affected family income and child poverty, however, there is no guarantee that increasing the marriage rate today would reduce poverty and improve family incomes in the future. Indeed, William Julius Wilson has hypothesized that there are not enough suitable men to allow for large increases in marriage within low-income black communities.29 And even if the marriage rate could be increased, the newly married families could differ from current and past married families in important ways that could make them more vulnerable to poverty. Several studies have therefore taken up the question of what would happen to the incomes and poverty rates of families with children if parents who are now single were to get married.

One such paper was published by Robert Lerman in 1996; we published another in 2002; and Wendy Sigle-Rushton and Sara McLanahan published a third in 2003.30 All three simulate hypothetical marriages by “pairing up” single women and men in various large data sets and then estimate how these simulated marriages would affect family incomes and child poverty. One advantage of these studies is that they correct for much of the selection bias found in other studies by matching women with men who are deemed to be suitable partners and then counting only the actual income that these men have to bring into a combined household.31 A second advantage is that because these analyses simulate marriages only for women for whom a potential husband can be identified, they address the critique that there are not enough “marriageable males” to allow for substantially more marriages to take place. A third advantage is that they sometimes adjust for the loss of benefits and the higher taxes that result from marriage.32 Thus, they provide some of the most powerful evidence to date of what could happen to the existing population if many single men and women were to marry. The methodologies and findings of these analyses are summarized in the box opposite.

Lerman used 1989 data and “married off” enough single mothers to return the marriage rate to that prevailing in 1971. His simulated increase in the marriage rate reduced the black child poverty rate in 1989 from 43.3 percent to 37.1 percent and the white child poverty rate from 11.7 percent to 9.8 percent. Among all children, the poverty rate fell from 17.1 percent to 14.7 percent. Among families participating in a simulated marriage, the new marriages pulled 43 percent of poor black children and 18 percent of poor white children out of poverty. Lerman also found that the simulation reduced income inequality among children by 26 percent.33

Our analysis was similar to Lerman's, although we used more recent data (1998) and replicated marriage patterns prevailing in 1970. We also made postmarriage adjustments to a wider range of benefits and taxes.34 Like Lerman, we found that raising the marriage rate would reduce child poverty considerably. In our simulation, the 1998 child poverty rate fell from 16.9 percent to 13.5 percent. Among families participating in the simulation, marriage reduced the number of poor children by 65.4 percent while raising average per capita income by 43.2 percent and the average income-to-needs ratio by 57.9 percent.35

Neither Lerman's simulation nor our initial analysis dealt with the issue of cohabitation. But we conducted a sensitivity analysis that assumed that the incomes and poverty rates of all cohabiting families participating in the simulation would remain unchanged after marrying. In this second simulation, poverty fell almost as much as it did in the original analysis. Another implication of our simulation is that the “marriageable male” hypothesis holds some salience for blacks. The anti-poverty effects in our simulation were about a third lower for blacks than for whites because we were unable to identify well-matched mates for some single black mothers. On the whole, however, both black and white families experienced large reductions in child poverty and large gains in family income.

Sigle-Rushton and McLanahan simulated marriages between unwed parents participating in the Fragile Families and Child Wellbeing Study. After combining the earnings of the unmarried parents in their data, they found that the new marriages pulled about 47 percent of the poor unwed mothers above the federal poverty line, although a larger share (about 53 percent) remained in poverty.36 These findings, together with others in their study, led them to conclude that labor market policies that encourage and reward full-time work might be more cost-effective for alleviating poverty than policies that promote marriage. Overall, however, Sigle-Rushton and McLanahan's simulation, like the others reviewed above, reduced child poverty dramatically among affected families.