Journal Issue: The Next Generation of Antipoverty Policies Volume 17 Number 2 Fall 2007
Estimating the Benefits of Marriage for Reducing Child Poverty
Promoting healthy and stable marriages will not be easy, especially among poor couples and unmarried parents with children. Moreover, it is not clear how much of an economic boost marriage provides, given the uncertainty over the effects of selection noted earlier. Nevertheless, it is useful to consider the range of effects on child poverty that might result.
One straightforward method (referred to as a shift-share analysis) estimates how the child poverty rate would change if a certain share of children were “shifted” from single-parent to two-parent households. For example, in 2000, 15.6 percent of all children were poor.65 The poverty rate was substantially lower (8.2 percent) for children living with married parents than for all other children (31.2 percent). In that year, 67.8 percent of children lived with two married parents. If the share of children with married parents in 2000 were the same as it was in 1990 (72.5 percent), then the overall share of children in poverty in 2000 would decline to 14.5 percent (see table 2). The 1.1 percentage point decrease from 15.6 percent to 14.5 percent would represent a 7 percent decline in child poverty. If the share of children living with married parents in 2000 were the same as it was in 1980 (76.7 percent), the overall share of children in poverty would decline to 13.6 percent—a 2 percentage point decline that corresponds to a 13 percent decline in child poverty. Finally, if the share of children with married parents in 2000 were the same as it was in 1970 (85.2 percent), then the overall share of children in poverty would decline to 11.6 percent—a 4 percentage point decline that corresponds to a 26 percent decline in child poverty.66
A more sophisticated approach is to simulate marriages by hypothetically matching single women and men in the population on the basis of factors such as age, education, and race. This approach makes it possible to estimate how these “marriages” would affect family income and child poverty. This approach is particularly useful because it takes into account the earnings of men, thus decreasing the influence of selection. It also acknowledges that appropriate matches will not exist for everyone. When Adam Thomas and Isabel Sawhill applied this method to Current Population Survey data from 1970 and 1998, they found that if single mothers married eligible men at the same rate as they did in 1970, overall child poverty would fall from 16.9 percent to 13.5 percent.67 Their study is notable because it adjusted the findings for a wide range of benefits and taxes, such as food stamps, the earned income tax credit, and child care expenses.
We use the Thomas and Sawhill findings to estimate the effects of smaller increases in marriage (using interpolation based on the assumption of a linear relationship). If the share of children living with married parents were the same in 1998 as it was in 1990, then child poverty would decline 6 percent. If the share were the same as it was in 1980, child poverty would decline 13 percent. Finally, as noted, if the share were the same as it was in 1970, child poverty would decline 20 percent. Note that the findings from this method differ somewhat from the shift-share analysis, at least in part because Thomas and Sawhill adjusted for a variety of benefits and taxes and in part because the matching procedure adjusts for some sources of selection.
To provide a range of estimates, we used a third method: time-series regression based on the years 1970–2004. The shares of children living in poverty and living without two parents increased in tandem from 1970 through the early 1990s. During the second half of the 1990s, the share of children living in poverty declined, while the share of children living without both parents remained stable. This decline in poverty (without a corresponding change in single-parent households) may have been due to the passage of welfare reform legislation in 1996, as well as to general improvements in the economy during these years. After 2000, however, child poverty and the share of children living without two parents both began to increase again. The regression analysis suggested that increasing the share of children living with two parents to 15.6 percent—the same share as in 1970—would result in a 29 percent decline in the share of children living in poverty—a decline larger than that found by Thomas and Sawhill.68
To put these figures in perspective, it is useful to think in terms of the absolute number of children who might be affected. In 2000, 11 million children lived in poverty.69 It is estimated that increasing the share of children living with two married parents to the same share as in 1970 would lift between 2.2 million and 3.2 million children above the poverty threshold, depending on the estimation method used. (Given that child poverty rates have increased since 2000, these figures underestimate the total number of children who would benefit. Moreover, the figures ignore benefits to the mothers of these children, who also would be lifted out of poverty.) These estimates lead to a straightforward conclusion. Although marriage programs will not eliminate all child poverty, or even most of child poverty, increasing the number of stable marriages will improve the economic well-being of many children and their mothers. Even if we conservatively assume that half of the estimated effect of marriage is due to selection, the decline in child poverty would be substantial.70
Can the policies we recommend bring about increases in the share of children living with two parents comparable to the changes shown in table 2? We believe the answer is yes, but it will take time. Based on our estimates of reductions in nonmarital births and the number of children whose parents divorce every year, our policy recommendations, if fully implemented, would take about ten years to make two-parent families as common as they were in 1990 and about twenty years to make them as common as they were in 1980. The slow rate of change reflects a process that social scientists call cohort replacement. That is, the current high rates of nonmarital fertility and marital instability originated in an era when childrearing outside marriage was seen as socially acceptable and divorce, even when a couple had children, was widely viewed as a reasonable solution to a less-than-satisfying marriage. As new cohorts enter their childbearing years with different attitudes about nonmarital births, the ability and commitment to use contraception more effectively, better relationship skills, and a stronger commitment to the norm of lifelong marriage, they will gradually represent a larger share of the population. But because cohort replacement is a slow process, most social change occurs gradually. Policymakers will need to be patient, and the leadership of different political parties must agree that these are worthwhile long-term goals.
Of course, other developments may intervene to speed up the process of change. For example, pro-marriage policies, even when directed at specific populations (such as couples about to marry or in the early years of marriage), may gradually generalize and permeate the larger culture. Similarly, other antipoverty proposals described in this volume, if implemented and successful, would reinforce the marriage-focused policies that we recommend. Poverty and family disorganization mutually reinforce one another. Although our goal has been to suggest how poverty might be lowered through family interventions, policies that improve people’s economic resources and decrease economic hardship also help to strengthen marriages and families. In this sense, any policy that lowers the rate of poverty in the United States is a pro-marriage policy.