Economists and scholars concerned with welfare policy have been concerned primarily with how economic conditions and policies influence adults' work, income, and life choices and, secondarily, with how these aspects of parents' lives affect their children (e.g., Hauser & Sweeney, 1997; Haveman & Wolfe, 1994; see Foster 2003 for economists' approaches to child development). One question they often pose is, how important is family income per se, above and beyond the other factors associated with poverty (e.g., depression, education level)? To what extent are the correlates of poverty due to factors other than income (e.g., single mother families, low parent education)? Some argue that how children develop is based less on monetary resources and more on parenting practices and skills, role modeling, moral character, and other familial characteristics. Some believe the solution to poverty is to provide income to the poor rather than try to change their family structures, education levels, and so forth; others believe that addressing these nonmonetary issues (e.g., provide children with role models of working parents) are key.
This was one of the primary debates surrounding the passage of the 1996 PRWORA, which ended 61 years of federally guaranteed cash assistance to poor families (Aid to Families with Dependent Children or AFDC) and in its place created a time-limited program designed to increase employment (Temporary Assistance for Needy Families or TANF; see Greenberg et al., 2002 for provisions of this law). This law emphasizes rapid movement from welfare to work, and thereby greatly reduces opportunities for welfare recipients to engage in such human capital development activities as education and job training—activities that many argue are key to successfully reducing poverty. Proponents of the "work first" philosophy argued that getting parents into the workforce quickly would benefit families by providing important work experience for adults, positive role models for children, and (possibly) increased family income. Opponents worried that strong employment requirements coupled with few supports would exacerbate hardship for families already at-risk.
Although welfare reform was intended to affect parents' behavior, the majority of public assistance recipients are children (Children's Defense Fund, 2001). By age 10, over one-third of all U.S. children will have lived in a household receiving welfare. The rate is even higher for black, non-Hispanic children (Levine & Zimmerman, 2000).
During the same period that welfare reform took place, work supports for low-income families were increased. The largest cash program for the poor is the federal Earned Income Tax Credit or EITC, first established in 1975, which provides a refundable tax credit for earnings, but not for other income. During the 1990s, as a means of supporting for working poor families, the maximum federal benefit more than doubled and some states included similar credits on state income taxes as well. Medicaid eligibility was expanded to include all low-income children, not just those in the welfare system, and funds for child-care assistance to low-income families increased dramatically between 1991 and 2000 (Fuller, Kagan, Caspary, & Gauthier, 2002). Earnings supplements and other work supports can increase family income with public funds that reward employment and are relatively free of the stigma associated with "welfare."
Faced with new participation mandates, sanctions, and time limits, and with the incentive of the EITC and the context of a strong economy, significant numbers of welfare recipients entered the workforce during the first 5 years after PRWORA was passed. A dramatic 51% reduction in caseloads between 1994 and 2000 led many to describe welfare reform as an unmitigated "success." The percentage of children living in poverty declined by about 17% between 1996 and 1999 (Primus, Rawl-ings, Larin, & Porter, 1999), and the decline was particularly apparent for African American single mother families (Haskins & Primus, 2002). With the economic recession following 2000, however, the number of children in poverty increased from 11.7 million in 2001 to 12.1 million in 2002 (U.S. Census, 2003). More importantly, there has been an overall increase in the number of children living in poverty whose parents are working, including those meeting the work requirements set by welfare reform. The percentage of poor children living in working families increased from 34% in 1995 to 43% in 2000, before falling to 40% in 2001 (Wertheimer, 2001; Dalaker, 1999).
Although the changes brought about by the 1996 legislation were partly responsible for national caseload reductions and higher employment rates among former welfare recipients (Blank & Schmidt, 2002), it is not clear what these changes have meant for the daily lives of families and children. Examining time-series data pre-and post-welfare reform sheds light on how children are faring since the enactment of PRWORA. Some national indicators of child well-being have improved in the post-1996 years, and some have worsened. Although these changes cannot be attributed to welfare reform per se, these indicators are associated with economic security, and provide examples of trends in children's well being in the post-welfare reform era. For example, teen fertility rates have decreased and high school completion rates have increased (see Duncan & Chase-Lansdale, 2001); however, the percentages of low-birth weight babies and families headed by a single parent have increased post-welfare reform (Annie E. Casey Foundation, 2003).
HOW DOES POVERTY AFFECT CHILDREN?
Examining changes in policy and economic conditions in relation to children's well-being provides some information about the role of poverty in children's lives. For investigators interested primarily in understanding children's development, the primary goal is to understand and improve developmental outcomes for children (e.g., cognitive development, school achievement, social competence, and physical health). The theories that guide this work rest on the assumption that the physical and social environments of poverty are responsible for many of the developmental outcomes associated with it (see McLoyd, 1998). Many economists have a more focused goal: to understand how public policies can influence outcomes for children and families. That leads them to ask about the effects of changing conditions that are amenable to public policy (e.g., family income, medical care, and child care) as opposed to factors that are not easily shaped by public policy (e.g., parents' intellectual capabilities, mental health, and parenting practices).
How Important Is Income?
We already noted that "poverty" includes a complex set of individual, family, community, and cultural factors that are difficult to disentangle. Poor families are often headed by single mothers, so family structure may contribute to children's developmental patterns (e.g., McLanahan, 1997). Because African Americans, Hispanic Americans, and Native Americans are especially likely to be poor, some of the effects of poverty may be a result of racial discrimination and disadvantage (see Duncan & Rodgers,
1998). Many parents in poor families lack education and perform poorly on tests of intellectual ability. They are likely to be unemployed or to work in unstable jobs with irregular hours that pay low wages and lack benefits. Some groups of poor families live in violent or disorganized neighborhoods and in physical conditions that pose risks of environmental toxins and injury. Poor parents experience more life stress, depression, mental health problems, and lack of social support than do affluent parents (Duncan, 1984; Duncan & Brooks-Gunn, 1997a; Duncan & Chase-Lansdale, 2001; Edin & Lein, 1997; Mayer, 1997a; McLoyd, 1998; Scarbrough, 1993). All of these family characteristics could contribute to low educational attainment, health problems, and behavior problems for children. How much does income per se contribute to the outcomes associated with poverty, independent of these other conditions? Are the developmental risks of poverty a result of income per se, or are they are result of other factors that are associated with poverty? By implication, could policies that increase family income produce improvements in children's well-being?
The most frequently used method for determining "effects" of poverty or family income is multivari-ate analyses of surveys and longitudinal databases, controlling statistically for a large of number of variables. The National Longitudinal Survey of Youth (NLSY), the Infant Health and Development Project (IHDP), and the Panel Study of Income Dynamics (PSID) have been analyzed in several studies investigating family income "effects." In an edited book, Duncan and Brooks-Gunn (1997a) assembled 13 longitudinal investigations of individuals ranging from early childhood through late adulthood in which family incomes during childhood and developmental outcomes were measured. The investigators conducted parallel analyses to determine the unique contribution of family income, with controls for related demographic and family structure variables. A summary of the findings indicated that family income during children's early years was related to cognitive and academic development in early and middle childhood and to adult earnings and work. Family income during adolescence was less likely to predict grades, educational attainment, or adult job characteristics. These authors then performed further analyses on the PSID data comparing the relations of family income at three points in development (age 0 to 5, 6 to 10, and 11 to 15) to the number of years of completed schooling in adulthood. Income during the child's first 5 years of life was the major predictor of completed schooling (Duncan & Brooks-Gunn, 1997b). In a more recent analysis of the NICHD Study of Early Child Care, poverty during the child's first 3 years and during ages 4 through 8 had cumulative effects, with children living in families with persistent poverty showing lower academic performance at third grade than did those who had experienced less enduring poverty (NICHD ECRN, in press).
Mayer (1997a) reached different conclusions using similar methods of analysis on many of the same datasets. She argued that the effects of income on developmental outcomes are modest at best when the many potential confounding variables are taken into account. By implication, raising the income of poor families would have negligible effects on cognitive skills, school achievement, or behavior problems. For example, according to her calculations, doubling the incomes of poor families would reduce the overall high school dropout rate in the United States from 17.3% to 16.1% and would reduce the overall teen childbearing rate from 20% to 18% (pp. 144-145). One reason, she suggested, is that government policies have ensured that many poor children receive such basic necessities as food, shelter, and health care. Because noncash government benefits (e.g., Medicaid, food stamps, child-care subsidies) are not counted as "income" in most investigations, the effects of these family resources are not reflected in studies of income. In another paper, Mayer (1997b) demonstrated that cash income was not synonymous with material hardship (lacking food, housing, or medical care). For several reasons, then, cash income is not a very good or complete index of the material resources available to a family.
A basic methodological conundrum exists in all of these studies. Although large numbers of related family variables are controlled in regressions, there is always the possibility that other unmeasured family characteristics could have contributed to the "effects" of income or poverty. The method of choice for determining causal direction is, of course, a random assignment experiment in which income is manipulated, but experimental designs to evaluate the effects of changing income are rare. Two sets of investigations meet this definition, the Income Maintenance Experiments of the 1960s and 1970s and the Next Generation experiments testing welfare and employment policies in the 1990s.
In the 1960s and 1970s, the federal government funded a set of experiments to evaluate the impact of a guaranteed minimum income on poor families with one or two parents in New Jersey; Gary, Indiana; Seattle; Denver; and rural areas in North Carolina and Iowa (Haveman, 1986). People were randomly assigned to receive a guaranteed minimum income or to a control group. The size of the guaranteed minimum and the tax rate associated with earnings was varied. There were methodological imperfections in these studies (Haveman, 1986; Rossi & Lyall, 1978), but they are of interest as we attempt to understand how poverty and income affect families and children.
The most serious drawback for our purposes is that these studies paid scant attention to measuring outcomes for children or parenting, and instead focused primarily on adult work effort and earnings, and marriage and divorce. Salkind and Haskins (1982) summarized available information concerning outcomes for children, and we have expanded their summary using a few other sources (Goenveld, Tuma, & Hannan, 1980; Institute for Research on Poverty, 1976; Kershaw & Fair, 1976; Mayer, 1997a; Office of Income Security, U.S. Department of Health and Human Services, 1983).
One consistent effect of the guaranteed income was increased expenditure on housing. Families in the experimental groups were more likely than controls to buy a house or to spend additional money on rent. They also purchased such durable goods as major appliances. One could reasonably argue that improvements in housing might improve the physical conditions in which children were living as well as offering them better schools and better neighborhoods. The studies contain almost no measures of parent-child relations or children's socio-emotional development; however no effects were found on the few superficial indexes that were reported.
The effects on children's school performance and attendance are somewhat mixed. For elementary-school-age children, there is some evidence of positive effects but, for high-school-age adolescents, there is little indication of a positive effect on achievement or aspirations. Nevertheless, in two studies, children in the experimental treatments were more likely to complete high school and they had higher levels of ultimate educational attainment than did those in control families. Adolescents in experimental families also were employed less than controls, perhaps because they were spending more time in educational pursuits (Salkind & Haskins, 1982). Other literature has indicated that extensive employment among high school students can interfere with optimal development (Steinberg & Dornbusch, 1991), so this finding may also indicate a "positive" outcome.
The Income Maintenance Experiments provided some support for the conclusions from longitudinal, correlational investigations. There is a modest positive effect of family income change on elementary school-age children's academic and school-related outcomes, but little evidence of positive effects on adolescents. Given the fact that the absolute increase in income in the experimental families was relatively small, this result is particularly notable. On the other hand, no social experiment has a "pure" manipulation; the experimental effects could have been influenced by participants' awareness that they were in a time-limited experimental program and, in some studies, by additional counseling services provided to some subgroups.
In the 1980s and 1990s, the federal Department of Health and Human Services granted several states and localities permission to experiment with rules for receiving welfare, but they required these efforts to be carried out with high quality evaluations. As a result, a large number of random assignment experiments were carried out testing variations in requirements for employment, time limits, earnings supplements, and child-care assistance policies for low-income families, almost all of which were headed by single mothers. Again, economic outcomes for parents—employment, welfare receipt, and income—were emphasized, but approximately a dozen of these studies also collected information about children's development, parent psychological well being, and family process.
The Next Generation Project was designed to synthesize the results of these studies, comparing the different policies tested. Several themes emerged. First, policies that led parents to increase employment without some form of earnings supplement (e.g., keeping part of their welfare grant, or direct wage supplements) did not increase overall family income because their increased earnings were accompanied by loss of welfare dollars (Morris, Huston, Duncan, Crosby, & Bos, 2001).
Second, policies that included earnings supplements, which increased overall family income, led to better school achievement and social behavior for elementary-school age children. Policies that increased employment without earnings supplements had no clear positive or negative effects on children's school achievement or behavior (Morris et al., 2001; Zaslow, Moore, Brooks, Morris, Tout, Redd, & Emig, 2002). For children who were adolescents when their parents entered a program, however, there were some negative effects on school achievement and involvement, regardless of variations in policies (Gennetian, Duncan, Knox, Vargas, & Clark-Kauffman, 2002; Zaslow et al., 2002). One reason appeared to be that teenagers had increased responsibility for caring for their younger siblings when their mothers increased employment time; teens also received less supervision.
This random assignment demonstration experiment was designed to test the effectiveness of an employment-based antipoverty program for families living in two inner city areas in Milwaukee, Wisconsin who were economically poor. Participants in the New Hope experiment were recruited beginning in July of 1994 and ending in December of 1995. The Child and Family Study (CFS) sample included all of the 745 sample members who had one or more children between the ages of 13 months and 10 years 11 months at the time of random assignment. Up to two children in each CFS family were identified as "focal children" for the study. The majority of families were headed by a single female parent, and most of these women were unemployed and receiving public assistance.
For adults assigned to the experimental group who worked full time (i.e., a minimum of 30 hours per week), New Hope provided job search assistance, wage supplements that raised their family income above the poverty threshold, and subsidies for health insurance and child care. Community service jobs were available for people who could not find other employment. Applicants in a control group did not receive these benefits. New Hope led to somewhat higher family income for program group families.
Surveys were administered to parents and focal children 2 and 5 years after random assignment; at the 5-year follow-up, children were between the ages of 6 and 16. A mail survey was sent to teachers of children whose parent gave permission. Teacher-reported outcomes are based on the reports of the 547 teachers who responded to our request. In-person interviews with parents and children were conducted in the family's home. Parents provided information about their children's achievement and social behavior, and children were given several standardized tests and questionnaires.
Children's academic progress was assessed by both parents and teachers. Parents rated their children's overall level of achievement on a 5-point scale ranging from poor to excellent. Based on their knowledge of recent report cards, parents evaluated their child's performance in reading, mathematics, and written work on 5-point scales (a = .87). Teachers rated children's academic performance using the Academic Subscale of the Social Skills Rating System (Gresham and Elliot, 1990).
Children's educational expectations were also assessed. Children (ages 9 and over) were asked to indicate how sure they were that they would finish high school, go to college, and finish college using 5-point scales (1 = not at all sure, 5 = very sure) (Cook, Church, Ajanaku, Shadish, Jeong-Ran, & Cohen, 1996).
Children's positive social behavior was assessed by both parents and teachers using the Positive Behavior Scale (Quint, Bos, & Polit, 1997). Its 25 items include items about compliance/self-control (for example, "thinks before he/she acts," "usually does what I tell him/her"), social competence and sensitivity (for example, "gets along well with other children," "shows concern for other people's feelings"), and autonomy (for example, "tries to do things for him/herself," "is self-reliant"). The parent or teacher responds on a 5-point scale concerning how often the child shows the behavior described ranging from "never" to "all of the time." Items for adolescents were adapted to be age-appropriate. The alpha was .91 for parents' ratings and .96 for teachers' ratings.
Children's problem behavior as assessed by parents and teachers using the Problem Behavior Scale from the Social Skills Rating System (Gresham & Elliot, 1990). The measure has two subscales. Externalizing problems include aggression and lack of behavior control (for example, "is aggressive toward people or objects," "has temper tantrums") (a = .81 for parents and .92 for teachers). Internalizing problems include social withdrawal and excessive fearfulness (for example, "appears lonely," "acts sad or depressed") (a = .61 for parents and .78 for teachers).
In surveys conducted 2 year and 5 years after random assignment, children in New Hope families were making better educational progress, had higher aspirations and motivation for school, and showed more positive social behavior than those in control group families (Huston et al., 2001; Huston et al., 2003). The reasons for these positive outcomes (i.e., the processes by which the New Hope program affected children's outcomes) are discussed later in this chapter. The New Hope study also collected ethnographic and qualitative information on some families. We are in the process of conducting yet another follow up interview with the children, families, and teachers approximately 8 years after random assignment.
Was this article helpful?