The effects of welfare and employment programs on children's participation in Head Start

ARTICLE IN PRESS

Economics of Education Review 26 (2007) 17–32 www.elsevier.com/locate/econedurev

The effects of welfare and employment programs on children’s participation in Head Start Young Eun Changa,, Aletha C. Hustonb, Danielle A. Crosbyc, Lisa A. Gennetiand a

Department of Child Development and Family Studies, Seoul National University, Seoul 151-742, Korea Department of Human Ecology, University of Texas at Austin, GEA115/A2700, Austin, Texas 78712-1097, USA c Department of Human Development and Family Studies, University of North Carolina at Greensboro, Greensboro, NC 27402-6170, USA d Senior Research Associate, MDRC, 16 East 34th Street, NY 10016, USA b

Received 28 September 2004; accepted 25 January 2005

Abstract We examine the effects of 10 welfare and employment programs on single mothers’ use of Head Start for their 3- to 4-year-old children, considering concurrent program effects on employment, income, and the use of other types of childcare settings. In general, these welfare and employment experiments increased parental employment and the use of center- and home-based childcare, but decreased families’ use of Head Start. The findings suggest that two types of policies affecting low-income families—welfare and employment on the one hand, and early childhood intervention on the other— are operating independently, and may actually conflict. Policy-induced increases in maternal employment generated an increased need for childcare; however, mothers did not appear to use Head Start to meet this need, instead increasing their use of other types of care arrangements. The part-day, part-year structure of Head Start at the time of these programs and issues of eligibility are discussed as potential barriers to the use of Head Start among low-income families under a welfare system that requires parents to work. r 2006 Elsevier Ltd. All rights reserved. JEL Classification: I20; I28; I30; I38 Keywords: Head Start; Childcare; Welfare; Policy

1. Introduction Since its beginning in 1965, Head Start has served 22 million children from low-income families with the goal of improving school readiness, social skills and health. In 2004, enrollment exceeded 905,000 Corresponding author. Tel.: +82 2 880 6821.

E-mail addresses: [email protected] (Y.E. Chang), [email protected] (A.C. Huston), [email protected] (L.A. Gennetian). 0272-7757/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.econedurev.2005.01.009

children (US Department of Health and Human Service (DHHS), 2005). By providing a full range of services, including nutritious meals and snacks, immunizations, and opportunities for parental involvement, Head Start serves not only children in poverty, but their families as well. Poverty is consistently linked to low educational performance in children, and evidence suggests that recovering from early educational deprivation later in life is difficult. Given the sizable literature showing its role in alleviating some of the detrimental effects of

ARTICLE IN PRESS 18

Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

growing up in poverty for young children (e.g., Barnett, 1995; Currie & Thomas, 2000; Lee, Brooks-Gunn, Schnur, & Liaw, 1990), Head Start is considered one of the most significant achievements of President Lyndon Johnson’s ‘‘War on Poverty’’. Families whose children are the intended recipients of Head Start, those with incomes below the poverty line, have also been the targets of major changes in welfare policy over the past several years. The new policies strongly encourage parents of young children to be employed, and eligibility for cash assistance has been sharply curtailed. One significant policy change from the late mid-1980s to through the 1990s is that several states implemented increasingly stronger work requirements. Indeed 12 states in the 1990s altered their work exemptions to require mothers of children as young as 6 months to participate in employment or employment-related activities. A recent study revealed that elimination of these types of exemptions was associated with more rapid exits from welfare through employment (Hofferth, Stanhope, & Harris, 2002). To the extent that these exemptions and similar types of employment-related policies lead to increased maternal employment, they are likely to increase parents’ need for childcare. A recent report examining the effects of 13 programs designed to move welfare recipients into employment indicates that parents in such programs used more childcare for their preschool-aged children than did control group parents (Crosby, Gennetian, & Huston, 2005). A subset of the programs offering expanded childcare assistance as part of the package of policies being tested led to increases in centerbased care; programs that did not include additional childcare assistance increased the use of home-based care arrangements. Children’s participation in Head Start was not examined separately from other types of center-based arrangements in this study. The emphasis in welfare-to-work programs on childcare as a support for maternal employment is quite different from Head Start’s stated goal of promoting optimal child development. Nevertheless, as a free, center-based educational program for young children, Head Start is one of the choices parents can consider as a childcare setting while they work. It is important to know whether children’s access to this source of developmental enrichment is affected by changes in welfare and employment policies. In the present study, we examine how policies designed to stimulate employment for low-income

parents affect their use of Head Start, using data collected from random assignment evaluations of a diverse set of welfare and employment programs conducted during the period from the late 1980s to the late 1990s. We examine experimental impacts on the use of Head Start in comparison to impacts on maternal employment and families’ use of other types of childcare arrangements. Our purpose is to determine whether two types of policies affecting low-income families—welfare and employment on the one hand, and early childhood intervention on the other—are congruent, in conflict, or simply independent of one another. 1.1. Head Start and children’s development The main goal of Head Start is to prepare preschool-aged children to enter school with the basic academic skills typical of their more advantaged peers. As shown in Table 1, its implicit (or, perceived) success is evidenced by the increasing number of Head Start enrollments and expanding federal allocations over the past 15 years. Evaluations of Head Start have shown significant immediate effects on cognitive development, particularly academic achievement and IQ (Barnett, 1995; Ramey et al., 1999; Zigler, Abelson, Trickett, & Seitz, 1982). Although some studies have shown these effects to fade out over time, others find persistent beneficial effects of Head Start on cognitive ability (Lee et al., 1990), school achievement, Table 1 Head Start program enrollment and appropriation history Year

Enrollment

Appropriation ($)

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

540,930 583,471 621,078 713,903 740,493 750,696 752,077 793,809 822,316 826,016 857,664 905,235 912,345 909,608 905,851

1,552,000,000 1,951,800,000 2,201,800,000 2,776,286,000 3,325,728,000 3,534,128,000 3,569,329,000 3,980,546,000 4,347,433,000 4,658,151,448 5,267,000,000 6,200,000,000 6,536,570,000 6,667,533,000 6,774,848,000

Source: US Department of Health and Human Service (2005).

ARTICLE IN PRESS Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

grade retention, placement in special education, social adjustment (Barnett, 1995, 1998), and educational attainment and college attendance (Currie & Thomas, 2000). Currie and Thomas (2000) argue that the duration of effects depends on the dynamics of school quality and child ethnicity. For example, the early benefits of Head Start might be negated if children are in low-quality elementary school settings. Because of its extended support services for families, Head Start also might be more helpful for children from families whose first language is not English. Indeed, Currie and Thomas (1999) found that Head Start closes the gap between Hispanic and non-Hispanic children in test scores and grade repetition. It appears to help children start school with a heightened probability of initial success, resulting in better school achievement. Using siblings as statistical controls in a nonexperimental, longitudinal study, Garces, Thomas and Currie (2002) find that participation in Head Start increased White children’s probability of graduating from high school, attending college, and higher earnings as young adults, and decreased African–American youth’s probability of being charged with a crime. Although many studies reveal ‘‘fade-out’’ in Head Start’s effects on test scores, these findings suggest that early academic gains may translate into non-cognitive skills that contribute to success in later life. Several limitations of the research on Head Start effects have been noted in the literature. Very few studies have used random assignment designs, with the most common approach being to compare outcomes for Head Start children with children from similar populations who are eligible for but do not participate in Head Start. Control groups created for comparison purposes might differ from Head Start families on a variety of familial and parental factors that affect both the decision to participate in Head Start and children’s outcomes, as well as the likelihood that the limited number Head Start slots are allocated to the most disadvantaged children (Lee et al., 1990). These differences could create positively or negatively biased estimates of the effects attributable to Head Start. These studies have also tended to use primarily African–American samples, which, despite high rates of participation, do not make up the majority of children attending Head Start (Garces et al., 2002). The more general literature on the link between early care experiences and children’s socio-emotional

19

and cognitive development highlights the importance of care quality (e.g., NICHD ECCRN, 1998, 2000). Studies evaluating the quality of Head Start classrooms across states reveal fairly high levels of quality overall, and improvements of quality over time (Rensnick & Zill, 2003; US DHHS, 2000). Although these studies suggest that Head Start provides higher-quality care than other preschool programs and childcare centers, quality tends to be uneven across different areas. Currie (2001) points out that quality tends to be higher in programs based in areas with higher family incomes, and in those with fewer minority families. Therefore, findings on the positive effects of Head Start need to be interpreted in the context of varying levels of program quality. Despite these caveats, large collection of studies generated over the last several decades provides consistent evidence that Head Start confers benefits to children from low-income families, particularly with respect to academic outcomes during the transition to school. However, the question of whether this beneficial program is consonant with parents’ efforts to increase employment and leave welfare remains unanswered. In other words, it is unclear whether children of parents who are encouraged or required to work under new state and federal policies have been able to reap the advantages of the Head Start intervention. Despite its noted benefits, Head Start may become a less favorable childcare option as parents work more, work unpredictable hours or for parents who have multiple children with childcare needs who are not all eligible for Head Start. 1.2. Head Start and parental employment In 2003, 72% of Head Start families had one or both parents working. Although most single-parent families were working, the share not working was 39% in 2003, an increase from 34% in 2002 (Hart & Schumacher, 2004). The random assignment welfare and employment studies examined in this paper tested a number of policies designed to increase employment among single mothers; in some cases, the programs were also designed to increase overall family income through earnings supplements. Many of these policy features (e.g., mandatory participation in employment services, time limits on welfare receipt, earnings disregards) have now become part of federal law or of some state welfare policies. Because most of the programs did increase maternal

ARTICLE IN PRESS 20

Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

employment, any effects on Head Start use might be a result of changes brought about by employment. Employment creates a need for childcare, which Head Start may or may not be well equipped to serve. Employment may also generate changes and constraints on parents’ time that make it more difficult for parents, especially single parents, to take advantage of or participate in Head Start. In particular, the employment of low-income single mothers is often characterized as intermittent and during non-traditional hours including weekends or rotating shifts during the evening when Head Start centers are not usually open. Moreover, Head Start requires parents to be involved, to the extent they are able, in all aspects of the program, including policy decisions regarding program management, planning, and implementation (US DHHS, 1975). One study indicated that conflicts between employment and program schedules represented one of the most significant barriers to parent involvement in Head Start (Lamb-Parker et al., 2001), implying that parents’ needs to comply with welfare-to-work polices have hindered their ability to participate in program activities. Even though parental involvement is not mandatory for Head Start participation, it is possible that conflicts between work and Head Start activities play a role in the parents’ childcare decisions. 1.3. Issues of accessibility and eligibility 1.3.1. Access A central issue in this paper is the viability of Head Start as a childcare arrangement for parents who are increasing their involvement in employment-related activities. Although Head Start provides comprehensive services, including early childhood education, it was first designed as a summer program for 4- and 5-year-olds prior to formal school entry and was later extended to a part-day program that was operated on the annual public school schedule. For this reason, Head Start has often been perceived as a compensatory preschool intervention, and not as childcare. Moreover, it is not likely to satisfy all of the childcare needs for parents who work full time or unconventional hours. The necessity for Head Start to mesh with parents’ employment demands, however, has grown as the number of working poor parents has increased in recent years. When Head Start was first launched, far fewer poor mothers were em-

ployed than is the case today, and public policy was much more tolerant of un- or under-employment among mothers with young children receiving public assistance (Phillips & Cabrera, 1996). In the early 1990s, the majority of children in Head Start were from families with non-working parents or parents participating in pre-employment education and training activities (Hofferth, 1994). Recognition of this problem led to a provision in the Head Start Act of 1998 (US DHHS, 1998) that requires states to allocate funds to increase programs that span the traditional work day and the calendar year. Childcare block grant funds can also be used for ‘‘wraparound’’ childcare to extend care to full day and full year. Nearly half of the families with children enrolled in Head Start programs in 2003 reported a need for such full-time care; 54 percent of these families received 6 or more hours of Head Start services per day year round. Approximately, 19 percent of Head Start programs currently provide services for 8 or more hours per day (Hart & Schumacher, 2004). In other words, parents of children in more than 80 percent of programs might have to coordinate two or more childcare arrangements in order to work full-time. Moreover, these statistics represent those who were already enrolled in the Head Start. There are presumably many children who are eligible, but not enrolled in Head Start whose parents would require full day services to accommodate their employment schedules. Partnerships between Head Start programs and childcare centers have been recently considered as a potentially important way to expand the availability of full day, full-year services (Jones, 2003). Through partnerships, Head Start programs might acquire additional resources to create full day programs. In 2003, however, only 8 percent of center-based Head Start programs were operated by a childcare center partner (Hart & Schumacher, 2004). More flexibility across separate funding streams might be helpful in establishing partnerships (Flynn & Hayes, 2003). Even with current efforts to extend service provision schedules, it remains doubtful that Head Start can be a viable childcare option for low-income parents who work irregular and unconventional hours. Concomitant with these efforts to expand Head Start services, substantial expansions have occurred in the allocation and availability of childcare subsidies throughout the 1990s to the present. Federal allocations to childcare subsidies increased from $3.5 billion in 1996 to more than $8.0 billion in 2000 (Greenberg et al., 2002). Preserving parental

ARTICLE IN PRESS Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

choice in the types of arrangements that are covered by subsidies was one key provision in these federal allocations. For low-income, working families, such subsidies could substantially reduce the cost of alternative non-Head Start-based care arrangements that might also be a better match for their childcare needs and employment schedules. 1.3.2. Age Because Head Start has narrow age guidelines (US DHHS, 1999), working parents with more than one child may find that they cannot use Head Start, or that using it for one child requires them to negotiate multiple childcare arrangements for their younger and older children. They may find it more convenient to use home-based care by relatives or family childcare homes where children of different ages can be cared for in the same setting. In 2003, nearly 87% of children enrolled in Head Start were 3 or 4 years old (US DHHS, 2004). Early Head Start was launched in 1995 to address the needs of infants and toddlers from low-income families, but it still serves only a fraction of children in poverty who are under the age of three. In 2002, about 708 programs served roughly 60,000 infants and toddlers from low income families, making up 7 percent of the total Head Start population (Irish, Schumacher, & Lombardi, 2003). 1.3.3. Eligibility and family income Some employment and welfare policies have an explicit goal of increasing family income; even those without such an explicit goal are designed to increase parents’ economic self-sufficiency on the grounds that earnings will provide better family support than does welfare. If family income increases, children may become ineligible for Head Start because most children in the program must be from a family with pretax income at the poverty threshold or less. If a child is from a family receiving public assistance or is in foster care the child is eligible regardless of the family income. At least 90% of the children who are enrolled in each Head Start program must be from low-income families. For example, in 2004, a family of four with pretax income of $18,850 or less was eligible for Head Start (US DHHS, 2004). In 2003, nearly three quarters of the families served by the program earned less than 100 percent of the federal poverty level (Hart & Schumacher, 2004). These eligibility guidelines may impose unintended negative consequences on parents who are

21

increasing their employment, reducing their dependence on welfare and essentially meeting the goals of welfare reform. Once parents move from welfare to work and earn more income, they are likely to lose their eligibility to enroll in the program. One might argue that raising the income eligibility guideline would result in serving more children whose families are near the poverty threshold, but the Head Start program currently serves only 40% of eligible children. Although there were large increases in the number of Head Start slots available during the 1990s, the proposed budget in 2005 includes only a $169 million increase for Head Start (as compared to the budget in 2000 which increased Head Start’s allocation by $1 billion). Raising the income threshold for eligibility could merely produce longer waiting lists. 1.4. Conceptual framework and research questions In this paper, we examine how welfare and employment policies affect Head Start participation for children of single parents using data collected from a diverse set of programs that had a random assignment design and that took place throughout the late 1980s to the late-1990s. Although all of these programs began prior to the 1996 welfare law, many of the policies tested eventually became states’ Temporary Assistance for Needy Families (TANF) policies, albeit in less generous form, and all of the programs include policy components currently being implemented or considered by states. These policies include those aimed at increasing employment and earnings (e.g., requirements to participate in employment related activities), and family resources (e.g., financial incentives or earnings supplements that make work pay), as well as those policies specifically targeting childcare (e.g. offering expanded childcare assistance encouraging the use of center-based care arrangements). By comparing the outcomes of individuals and families in a control group, under the then-current policy environment, with the outcomes of individuals and families in a program group, subject to a new set of policies, we investigate and can state with confidence whether or not the policies tested affected the use of Head Start. These welfare and employment programs may influence the type of care families use, and use of Head Start more specifically, due to a variety of factors related to the program effects on employment, the scheduling and extent of employment, income and, cost and availability of types of

ARTICLE IN PRESS 22

Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

childcare arrangements. First, as shown in prior studies, these programs generally increased maternal employment and increased the need for childcare (Bloom & Michalopoulos, 2001; Gennetian & Michalopoulos, 2003). Some of this increased employment is new entry into employment, among those mothers who were not employed at the beginning of these studies, and some of this increased employment represents more hours of employment (e.g. moving from part-time to fulltime employment to satisfy work requirements). Head Start might be one type of childcare arrangement considered in the context of this increased employment. Second, some of these increases in income produced by a program, such as those programs with an earnings supplement, may allow parents to buy more or higher quality childcare. Since Head Start is free for income-eligible families, and is higher quality on average as compared to other types of center-based care arrangements, the purchasing power associated with more income is not likely to affect families’ decision to use Head Start arrangements. However, increased income might make some families ineligible for Head Start. Third, these programs might increase employment in occupations or jobs that have rotating schedules with non-traditional hours that are a mismatch with the hours of Head Start centers. In this case, childcare needs are likely to be met by relying on alternative center or home-based care arrangements. Finally, some of the treatments in these programs included a range of economic and administrative means of assisting families in meeting their childcare needs. A subset of the studies tested various childcare assistance policies by providing program group participants enhanced assistance in the form of: (a) additional resources to secure paid care, (b) case management/support services for childcare, (c) efficient reimbursement for childcare costs; and/or (d) restriction of subsidies to regulated care. These policy features are assumed to play an important role in parents’ decisions about childcare arrangement. Access to childcare assistance might also increase in the programs that had more generous earnings disregards since this increased ties to the welfare system. To the extent that assistance policies and practices make more types of childcare accessible (available and affordable) for working poor families, they may directly affect the types of care arrangements used by parents and experienced by children. An earlier report on a similar set of studies as those examined here

suggests that the assistance enhancements described above led to increased use of center-based (as opposed to home-based) care arrangements (Crosby et al., 2005). In this study, we expect trade-offs in the program groups between Head Start and other types of childcare arrangements and also between Head Start and employment and/or income. As parents increase their employment hours and require more childcare, parents might also increase their use of Head Start as a type of childcare arrangement. On the other hand, the availability of additional resources for purchasing care (through subsidies or increased income) may make other childcare alternatives more feasible and/or families ineligible for Head Start. This could lead to no effect in the use of Head Start—since there will be naturallyoccuring reliance on Head Start that is equivalent in both the treatment and control groups—or, possibly, declines in the use of Head Start as families in the treatment group might be more likely to use other types of childcare arrangements (in lieu of their new treatment-induced employment and employment schedules) as compared to control group families. With this general framework in mind, we specifically address three research questions in this study: First, do welfare programs have any effects on the employment and income of low-income single mothers? Second, are program impacts on mothers’ employment and income related to impacts on the use of Head Start? Third, are the changes in Head Start use in the program groups similar to changes in the use of other types of childcare? That is, do programs that increase or decrease the use of Head Start have similar or different effects on use of other forms of centerbased childcare, such as day care centers and preschools, or home-based care arrangements? 2. Method 2.1. The experimental studies Data were drawn from four welfare and employment studies, in which a total of 10 programs were evaluated. Two of these studies—the New Hope project and New Chance—were two-group research designs where families were randomly assigned to one program or experimental group or to a control group. More complex multi-group research designs were implemented in the next two studies. In the Minnesota Family Investment Program (MFIP),

ARTICLE IN PRESS Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

23

families in urban counties were randomly assigned to either the full MFIP program, to MFIP Incentives Only (which offered MFIP’s financial incentives without its mandatory work requirement) or to a control group. In the National Evaluation of Welfare-to-Work Strategies (NEWWS), three sites implemented head-to-head tests of two treatments. In each site, families were randomly assigned to either a program group with an employment focused program, a program group with an education focused program or to a control group. This in turn derives information about 6 programs. Although these studies shared the common goal of moving welfare and low-income families into work, each program had a different approach in terms of their use of earning supplements, mandatory employment services and childcare services. Head Start eligibility guidelines did not differ substantially across the study sites.

disadvantaged group of families headed by young welfare mothers, aged 16–22, who had first given birth as teenagers and who had dropped out of high school. The program was aimed at helping participants become self-sufficient by increasing their academic and vocational skills so that, over time, they could find and keep jobs and, in turn, reduce their receipt of public assistance. Program activities included life skills, family planning, parenting skills and counseling. Some New Chance sites provided on-site full-day center-based childcare; others offered center or home-based care nearby while mothers participated in activities. No mandatory employment services and financial incentives were provided in New Chance. Data on 1875 children were collected 18 months after mothers were enrolled in the program or assigned to the control group (see Quint, Bos, & Polit, 1997 for study details).

2.1.1. New Hope New Hope was a work-based antipoverty program operated in Milwaukee, Wisconsin from 1994 to 1998 (Bos, Huston, Granger, Duncan, Brock, & McLoyd, 1999; Huston et al., 2001, 2003). Because New Hope enrolled any adult in the area who had an income at or below 150% of the federal poverty level and was willing to work full time, the study included both welfare recipients and other lowincome parents. The program provided intensive services to provide direct economic support to these working poor families. Parents who worked at least 30 h per week were eligible for an earnings supplement, health insurance, and subsidized childcare. New Hope also provided intensive case management services such as job search assistance, including the opportunity to apply for a wage-paying community service job in case parents did not find full-time work. Data on children were collected 2 years after random assignment to either the program or the control group. The total New Hope sample includes 1,356 participants, 825 of whom were selected for the Child and Family Study (CFS) in order to examine the program’s effects on families and children (see Bos et al., 1999 for study details).

2.1.3. The MFIP MFIP was implemented in 1994 in three urban and four rural counties in Minnesota until 1998. Single-parent families who were applying for or currently on welfare were included in the child study. The evaluation tested two pilot programs: Full MFIP and MFIP Incentives Only. Full MFIP provided an enhanced financial incentives package as a supplement to earned income, allowing working welfare recipients to keep more of their income when they went to work. Direct childcare payments were made to the provider. The package also simplified public assistance rules and procedures by combining Aid to Families with Dependent Children (AFDC), Minnesota’s Family General Assistance, and Food Stamps into a single cash benefit program. Long-term welfare recipients were required to participate in MFIP’s employment and training activities, unless they were working 30 h a week or had children under the age of one. MFIP Incentives Only included all the features of the Full MFIP program except mandatory employment services. Data collected 3 years after random assignment were used in this study and 1900 families were included in the original child study report sample (see Gennetian & Miller, 2000 for study details).

2.1.2. New Chance New Chance operated between 1989 and 1992 at 16 sites in 10 states across the country. The New Chance Demonstration was designed to test the value of comprehensive services in assisting a

2.1.4. The National Evaluation of Welfare-to-Work Strategies (NEWWS) National Evaluation of Welfare-to-Work Strategies (NEWWS) was a multi-site study testing

ARTICLE IN PRESS 24

Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

mandatory welfare-to-work programs, which operated in the early-to-mid 1990s under the federal Job Opportunities and Basic Skills Training (JOBS) program. Evaluations of six programs in three cities (Atlanta, Georgia; Grand Rapids, Michigan; Riverside, California) included intensive information about children’s experiences and development. Welfare recipients enrolled in JOBS were required to participate in basic education or employmentrelated activities as a condition of receiving welfare. For the purposes of the evaluation, each of the three sites examined here operated both a program emphasizing job search as a first activity and a program emphasizing basic education as a first activity. Job-search-first programs required most participants to look for work immediately, while education-first programs initially placed participants in education and training programs to increase ‘‘human capital’’ before participants moved into employment. Information on children was collected 2 years after mothers were randomly assigned to one of the two programs or to a control group in their site (see Hamilton et al., 2001 for study details). 2.2. Sample The current study focused on the children of single mothers who were 3–5 years old at study entry. We imposed this age restriction so that we can capture children during the period in which they are eligible for Head Start. These follow-up periods ranged from 18 to 36 months after their mothers’ random assignment. This sample differs from the samples examined in the original evaluation studies where the program impacts on children in other age groups have been also examined. Sample sizes and baseline demographic characteristics for each study are presented in Table 2. In each study, the baseline characteristics of respondents and non-respondent were compared at the time of the follow-up assessments. The two groups did not differ substantially, generating confidence that attrition is not biasing our results or the internal validity of the estimates. More information on non-response bias can be found in the study reports cited above. 2.3. Measures Each study involved a follow-up survey approximately 18–24 months after random assignment. Response rates were quite high for each study, averaging at about 80 percent of the fielded sample

(see cites to original reports above). As part of this survey, mothers responded to a set of childcare questions. If mothers indicated that they had ever used any type of childcare during the follow-up period, they were asked a series of questions about the types of care children had experienced. Preschool, nursery school and childcare centers were classified as center-based care, which refers to any licensed and regulated care that takes place in a group setting. Center-based care also includes Head Start arrangements. (In MFIP, the variable centerbased care was re-coded to include reports of Head Start use collected in a different part of the survey and thus slightly differs slightly from prior analyses [Gennetian & Miller, 2000].) Mothers were also separately asked whether their child had participated in a Head Start program since the random assignment. Note that because use of Head Start was asked under the general rubric of childcare, it is possible that rates of Head Start use are undercounted somewhat. Some mothers may not view their child’s participation in Head Start as time in childcare, and, other mothers may not realize that the center-based arrangement their child is enrolled in is a Head Start program or is affiliated with a Head Start program. In a recent study, Garces et al. (2002) examined the validity of survey reports of Head Start use by comparing survey reports in a large longitudinal data with administrative records data and provided evidence that survey reports of Head Start use is fairly valid and reliable (see Garces et al., 2002 for detail). Home-based care arrangements include care by family members/relatives or non-relatives either in the child’s home or in the caregivers’ home. Licensed or certified family childcare homes as well as more informal, unlicensed care arrangements were included in this category. Information was also collected on mothers’ employment and income over the follow-up period using administrative records. The variable average quarterly employment refers to the percentage of quarters over the followup that a mother was employed; this variable is derived from state unemployment insurance records and, therefore, excludes any self-employment or employment that is not reported to an unemployment insurance agency. Average quarterly family income reflects earnings and public assistance (including AFDC and food stamps) measured from administrative records data, and any earnings supplements from the experimental program received by mothers between random assignment and

Marital status (%)

95.50 96.19 94.62 38.64 40.49 36.65 41.75 40.49 43.00

215 27.79 102 27.74 113 27.84

201 27.76 102 27.74 99 27.79

945 29.53 472 29.40 473 29.66

844 29.51 472 29.40 372 29.65

548 29.51 308 29.40 240 29.64

415 29.55 308 29.40 107 29.96

MFIP incentives only Control group Program group

Atlanta HCD-NEWWS Control group Program group

Atlanta LFA–NEWWS Control group Program group

Grand Rapids HCD–NEWWS 396 27.10 Control group 205 26.76 Program group 191 27.46

412 27.10 205 26.76 207 27.43

MFIP full Control group Program group

Grand Rapids LFA–NEWWS Control group Program group

Riverside HCD–NEWWS Control group Program group

Riverside LFA–NEWWS Control group Program group

6.31 6.34 6.28

5.56 6.34 4.71

0.83 0.42 1.34

0.42 0.42 0.42

1.99 1.96 2.02

2.33 1.96 2.65

14.70 43.37 14.94 41.56 14.02 48.60

16.79 41.79 14.94 41.56 19.17 42.08

95.87 96.19 95.56

40.80 41.18 40.40

40.00 41.18 38.94

46.53 29.70 45.45 28.79 47.06 30.15

202 20.02 66 19.85 136 20.10

New Chance Control group Program group

59.92 27.48 58.39 29.20 61.60 25.60

262 27.51 137 27.58 125 27.43

39.28 40.91 34.58

38.32 40.91 35.00

49.27 50.73 47.83

54.04 50.73 57.59

3.20 2.97 3.49

3.39 2.97 3.81

47.26 45.10 49.49

47.44 45.10 49.56

22.28 22.73 22.06

10.69 10.22 11.20

45.30 45.13 45.79

47.63 45.13 50.83

59.95 59.02 60.87

59.85 59.02 60.73

72.39 69.70 75.81

71.01 69.70 72.30

71.64 75.49 67.68

75.81 75.49 76.11

84.16 84.85 83.82

76.34 77.37 75.20

54.70 54.87 54.21

52.37 54.87 49.17

40.05 40.98 39.13

40.15 40.98 39.27

27.61 30.30 24.19

28.99 30.30 27.70

28.36 24.51 32.32

24.19 24.51 23.89

15.84 15.15 16.18

23.66 22.63 24.80

Black Latino White Never married Separated

New Hope Control group Program group

N

Ave. Ethnicity (%) age of parent

Table 2 Descriptive statistics for the baseline characteristics of participating families

24.58 24.03 26.17

25.55 24.03 27.50

60.19 58.54 61.84

61.62 58.54 64.92

62.80 65.25 59.68

64.97 65.25 64.69

77.11 78.43 75.76

73.95 78.43 69.91

9.41 9.09 9.56

59.92 58.39 61.60

26.99 29.22 20.56

29.20 29.22 29.17

56.07 60.98 51.21

58.33 60.98 55.50

36.97 38.77 34.68

37.25 38.77 35.73

47.26 50.00 44.44

47.91 50.00 46.02

43.56 46.97 41.91

68.70 68.61 68.80

Number of children in household

3.98 3.92 4.04

3.72 3.92 3.54

0.72 25.30 0.32 25.00 1.87 26.17

0.18 25.91 0.32 25.00 0.00 27.08

0.00 25.49 0.00 25.37 0.00 25.60

0.00 28.54 0.00 25.37 0.00 31.94

0.00 25.59 0.00 27.75 0.00 22.85

0.11 24.76 0.00 27.75 0.21 21.78

1.00 0.00 2.02

1.40 0.00 2.65

4.95 95.05 6.06 93.94 4.41 95.59

9.16 18.32 12.41 16.79 5.60 20.00

73.98 74.68 71.96

73.91 74.68 72.92

74.51 74.63 74.40

71.46 74.63 68.06

74.41 72.25 77.15

75.13 72.25 78.01

95.02 96.08 93.94

94.88 96.08 93.81

0.00 0.00 0.00

72.52 70.80 74.40

64.82 67.21 57.94

65.33 67.21 62.92

75.73 77.07 74.40

76.52 77.07 75.92

63.86 65.04 62.37

63.49 65.04 61.95

71.64 76.47 66.67

71.63 76.47 67.26

80.20 87.88 76.47

49.62 45.99 53.60

35.18 32.79 42.06

34.67 32.79 37.08

24.27 22.93 25.60

23.48 22.93 24.08

36.14 34.96 37.63

36.51 34.96 38.05

28.36 23.53 33.33

28.37 23.53 32.74

19.80 12.12 23.53

50.38 54.01 46.40

None o2 yr o ¼ 2 yr 1 or 2 3 or more

High Prior Prior AFDC school employment receipt: % degree (%) (%)

47.23 46.43 49.53

47.45 46.43 48.75

49.27 52.20 46.38

51.26 52.20 50.26

51.30 50.00 52.96

47.51 50.00 45.03

46.27 43.14 49.49

47.44 43.14 51.33

45.05 48.48 43.38

50.76 49.64 52.00

3.73 3.74 3.68

3.72 3.74 3.69

3.83 3.80 3.86

3.83 3.80 3.86

3.86 3.81 3.92

3.84 3.81 3.88

3.99 4.08 3.89

4.01 4.08 3.96

3.49 3.47 3.49

3.92 3.98 3.85

Male Ave. age child (%) of child

Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

ARTICLE IN PRESS 25

ARTICLE IN PRESS 26

Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

the follow-up survey (divided by the number of quarters in the study’s follow-up period). The one exception to this is the income measure from the New Chance study, which is a respondent report from the survey. 2.4. Analysis strategy Each of the studies used a random assignment experimental design to measure the effects or impacts of the program on child and family well being. For each of these studies except for New Hope, at the time of applying or re-applying for public assistance families were randomly assigned, in a lottery like process, to either a program group or to a control group. In New Hope, random assignment occurred at the time families voluntarily walked into the program office. In the present report, we compute program impacts on the childcare and economic outcomes for the subsample of children who were age-eligible for Head Start; we present differences between program-control group differences in any Head Start participation (%), use of any center-based care (%), use of any homebased care (%), employment (%), and income ($). Because parents may have used more than one type of childcare, the total percentage of the sample using center care, home-based care, and Head Start can add up more than 100%. To determine if program impacts were statistically significant, OLS regressions were conducted, regressing each outcome measure on a dummy variable for treatment group, as well as a variety of background factors. For the dichotomous outcomes, results were similar using logit or probit techniques. For ease of interpretation, we discuss and present findings from the OLS regression models. The covariates controlled for in the original impact analysis for each study were used in these analyses as well, and the specific characteristics varied slightly across studies. Generally, they included, mother age, ethnicity, marital status, mother education, mother employment, welfare history, number of children in family, and child’s gender. In New Chance data, an expanded set of covariates, including site, mothers’ age when the first child was born, mothers’ depression, etc., was available and included in the analysis. Tests of differences between program and control groups were 2-tail. Because we have access to the microdata, we pooled the data across the studies to estimate an average effect. The resulting estimate is

equivalent to a meta-analytic average, with appropriately adjusted standard errors. Compared to the simple technique of generating weighted means by averaging each estimate by sample size per study, using a pooled data and its adjusted standard errors to generate a meta-analytic average derives a more accurate estimate. 3. Results 3.1. Main results To answer our first research question of whether experimental programs affect the employment and income of mothers of 3–5 year old children, we examined the program-control differences on employment and income. Overall, mothers in program groups were more likely than mothers in control groups to be employed in most study programs (t(1) ¼ 7.03, po.001), and earn more income in some study programs (t(1) ¼ 3.28, po.001) (See Table 3). The second research question was: Are program impacts on mothers’ employment, particularly fulltime employment, related to impacts on the use of Head Start? It is assumed that more time at work requires mothers to find more hours of childcare. It is, however, questionable whether Head Start would have remained helpful for mothers as their employment reached full-time because Head Start usually operates on a part-time schedule. Therefore, we examined the correspondence between the magnitude and direction of program-control group differences on Head Start and on employment. These programcontrol differences are shown in Fig. 1. We examine program impacts on the likelihood of attending Head Start at any point after random assignment. As shown in Fig. 1 (see also Tables 3 and 4), there are small, but consistent decreases in Head Start participation among program group families as compared to control group families. The bar on the far right of Fig. 1 illustrates the average impact across studies. Even though only one program, Atlanta’s Labor Force Attachment (LFA) program, produced a statistically significant impact on the use of Head Start, the weighted average of impacts across studies is significant suggesting an overall decline in Head Start use (t(1) ¼ 2.79, po.01). Overall, mothers in the program groups were more likely than mothers in the control groups to be employed and were less likely to use Head Start. However, it does not appear that the magnitude of program-control differences in employment was

ARTICLE IN PRESS Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

27

Table 3 Effects of welfare and work programs on employment and income Employment and income programs

New Hope New Chance MFIP incentives only MFIP full Atlanta HCD-NEWWS Atlanta LFA-NEWWS Grand Rapids HCD-NEWWS Grand Rapids LFA-NEWWS Riverside HCD-NEWWS Riverside LFA-NEWWS Overall

Average quarterly employment (%)

Average quarterly income ($)

Control group

Impact

Control group

Impact

71.79 20.00 43.68 44.07 36.21 36.08 39.55 38.53 16.18 15.76 34.27

9.62** 1.68 7.02 19.79*** 3.56y 5.66* .12 12.20*** 8.27** 15.38*** 7.79***

3675.28 3851.11 2885.06 2869.57 2404.48 2413.12 2640.80 2589.95 2738.65 2816.84 2670.85

428.31** 166.76 304.88y 781.33*** 91.56 66.09 33.90 65.25 95.40 95.58 128.58***

Note 1: ***po.001, **po.01, *po.05, ypo.10. Note 2: Average quarterly income was calculated by dividing the average annual income by 4 in New Chance because only average annual income is available in New Chance data.

(%) 25

20

Percentage point impact

15

10

5

0

-5

-10

-15

New Hope

New MFIP Chance incentives only

MFIP full

Atlanta HCD

Atlanta LFA

Average quarterly employment

Grand Rapids HCD

Grand Rapids LFA

Riverside Riverside Weighted HCD LFA mean

Use of Head Start

Fig. 1. Effects of welfare and work programs on employment and use of Head Start: parents of children aged at 3–5 at study entry.

related to the size of program-control differences in the use of Head Start. Additional analyses were conducted to examine the possibility that families in the program groups had

increased family income, and thereby possibly lost their eligibility for Head Start (Table 3). Similar to the findings on impacts on employment, even though some programs (those that provided an earnings

ARTICLE IN PRESS 28

Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

Table 4 Effects of welfare and work programs on use of Head Start and child care: Children aged 3–5 during follow-up Use of childcare programs

Use of Head Start (%)

Use of center-based care (%)

Use of home-based care (%)

Control group

Impact

Control group

Impact

Control group

Impact

New Hope New Chance MFIP incentives only MFIP full Atlanta HCD–NEWWS Atlanta LFA–NEWWS Grand Rapids HCD–NEWWS Grand Rapids LFA–NEWWS Riverside HCD–NEWWS Riverside LFA–NEWWS

30.84 31.98 12.13 12.56 38.07 37.77 43.14 43.27 33.99 33.14

1.44 3.02 4.43 6.20 .80 7.74* 3.58 2.55 3.86 5.12

29.14 30.99 48.51 47.80 32.08 32.18 26.85 26.06 11.51 11.08

15.73** 22.82* .49 9.04 4.40 6.28y .34 1.27 7.06* 12.18**

70.43 66.54 77.64 77.74 38.20 38.32 76.52 75.74 41.62 40.91

10.82y 2.02 .06 4.30 8.46** .16 .54 4.82 19.55*** 17.97**

Overall

35.76

4.60**

27.00

52.78

7.04***

8.71***

Note 1: ***po.001, **po.01, *po.05, ypo.10. Note 2: Use of Head Start, any home-based care, and center-based care (not including Head Start) can add up to more than 100%.

supplement or had an enhanced earned income disregard) increased family income, these program impacts on income do not line up with program impacts on Head Start use. For example, the Atlanta LFA program significantly decreased Head Start use, yet had no statistically significant effect on income. The third question was: Are the program impacts on Head Start use parallel to changes in the use of other types of childcare? That is, do programs that increase or decrease the use of Head Start have similar or different effects on use of childcare centers or home-based childcare? In Fig. 2, program impacts on the use of different types of childcare are compared with impacts on Head Start use. In contrast to the effects on Head Start, these programs generally increased use of both centerbased and home-based care (see Table 4). Five out of 10 programs showed increased use of center care. The Atlanta Human Capital Development (HCD), Riverside HCD and Riverside LFA increased use of home-based care. T-tests of the weighted averages also indicate that, on average, program group parents increased their use of both center-based (t(1) ¼ 5.39, po.001) and home-based (t(1) ¼ 4.22, po.001) childcare over levels reported by control group parents. Thus, it does not appear that parents were trading Head Start use with other types of care (e.g. center care). Instead, the pattern of these results suggests that program-induced increase in employment among mothers of preschoolers was being met by program-induced increases in nonHead Start childcare arrangements.

3.2. Follow-up analysis As an effort to disentangle the relations of change in employment and income with childcare use, we conducted further analyses separately examining program impacts among those who were employed at study entry from those who were not employed at study entry. The results of this analysis using the pooled data sample are shown in Table 5. Strikingly, Table 5 shows that these programs decreased Head Start use among those who were employed at study entry as well as those who were not employed at study entry, despite differences in program effects on use of home-based care and on income. Among those who were employed at study entry, programs decreased Head Start use, increased center-based care, and increased employment and income. Among those who were not employed at study entry, programs decreased Head Start use, increased home-based and center-based care, and increased employment, but had no effects on income. These findings provide further evidence that increased income is not a possible explanation for the decreased use of Head Start. While this might be an indication that program group families are not meeting the income eligibility thresholds (despite their increased income), it is more likely that these findings are emblematic of the overall pattern of findings suggesting a strong link between increased employment and increased reliance on alternative types of care because Head Start is not a real ‘‘childcare choice’’ due to scheduling issues, or

ARTICLE IN PRESS Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

29

(%) 25

20

Percentage point impact

15

10

5

0

-5

-10 New Hope -15

New MFIP MFIP full Chance incentives only Use of Head Start

Atlanta HCD

Atlanta LFA

Grand Rapids HCD

Use of center-based care

Grand Rapids LFA

Riverside Riverside Weighted HCD LFA mean

Use of home-based care

Fig. 2. Effects of welfare and work programs on use of Head Start, center-based care, home-based care: Children aged 3–5 at study entry.

Table 5 Average effects of welfare and work programs using pooled data, by employment status of parent at study entry Employed at study entry Control group Average quarterly employment Average quarterly income Use of Head Start Use of center-based care Use of home-based care

56.4 3032.60 36.10 31.41 70.90

due to the influences of feasible choices encouraged by caseworkers. 4. Discussion In this analysis, we compared the effects of 10 different welfare and employment programs for parents on the likelihood that their 3-and 4-year-old children would participate in Head Start. In general,

Not employed at study entry Impact ***

8.30 225.41*** 4.61y 9.75** 2.12

Control group

Impact

19.06 2390.48 35.73 23.71 40.95

4.02*** 9.41 4.99* 7.64*** 10.43***

these programs produced overall negative impact on Head Start use, despite the fact that they increased parents’ employment and the use of both centerbased and home-based childcare. It appears that many employed mothers do not consider Head Start a viable option for childcare, despite the fact that it is a free program for young children with potential educational benefit. Most mothers in these studies used paid childcare arrangements as they entered or

ARTICLE IN PRESS 30

Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

increased employment, and many of the received subsidies or other forms of childcare assistance that enabled them to do so (Gennetian, Crosby, Huston, & Lowe, 2004). Although there are no fees to attend Head Start, this option may have had other monetary and nonmonetary costs (e.g., transportation, multiple childcare arrangements) that affected mothers’ childcare decisions. In her study of teen mothers, Gassman-Pine (2003) found a pattern of increased use of center care as a counterpart of decreased use of Head Start in some of these same programs where mothers’ employment or participation in education/training increased. The finding implied that employed mothers who need more childcare, often full-day care, traded Head Start for other center care arrangements. This particular pattern was not replicated in the present study. In Atlanta-JOBS (LFA) and Riverside-JOBS (HCD), there were increases in center care use and decreases in Head Start use, but this pattern did not appear to correspond to the increases in parental employment. The increases of employment in New Hope, MFIP Full, Grand Rapid-JOBS (LFA) and RiversideJOBS (LFA) were moderate to large, but these programs did not significantly alter (increase or decrease) use of Head Start. The pattern of impacts on Head Start and employment suggest that mothers did not use Head Start to satisfy their childcare needs while they were working. Full-time employment in particular is likely to require reliable and extensive childcare. The half-day and school calendar schedule of many Head Start programs (especially at the time of these studies) makes it likely that employed mothers interested in using Head Start would need to find additional childcare arrangements as well. Mothers may prefer a single care arrangement that covers their work schedule to combining multiple childcare settings that require transportation and other logistical arrangements. Many women in low-wage jobs work irregular and nontraditional hours, making it more difficult to have their children in a program that runs on a school-like schedule. Some of them may have other free childcare options (e.g. family members) that are attractive if they are available for the times mothers work or if mothers lack transportation to get to a Head Start program. Expanded childcare assistance or on-site childcare services offered by the programs in this study might have altered mothers’ childcare choice set as well. For example, New Hope and New Chance promoted formal childcare options (i.e. childcare

centers) both with financial and non-financial aids. In these programs, mothers might choose full-time childcare centers over other types of care, including free, half-day Head Start programs. It is also possible that increased employment led to more family income and in turn, the loss of eligibility for Head Start, but our findings regarding program impacts on family income do not generally support this hypothesis. One goal of this analysis was to determine whether two types of policies affecting low-income families—welfare and employment, on the one hand, and early childhood intervention on the other—are congruent, in conflict, or simply independent of one another. If the two were congruent, we might expect that as mothers increase their employment, they also increase their use of Head Start as an enriching care environment for their children. In this situation, it is possible that parents’ employment efforts and children’s experiences in Head Start would have synergistic or cumulative effects on particular child outcomes such as school achievement. On the other hand, if the two types of policy are not congruent, we would expect that families take up one or the other (e.g., not increase their work efforts and use Head Start, or work more and not use Head Start). The current set of findings suggests that, at the very least, the two sets of policies are operating independently, and may actually be in conflict. Participation in welfare and employment programs does increase women’s hours of employment and generates a need for childcare, especially for mothers of preschoolage children. These programs do not, however, increase families’ use of Head Start along with other types of childcare. The structure and eligibility requirements for Head Start appear to present some barriers when mothers are also meeting the demands of welfare reform and reaping the benefits of employment. Recognizing the problems that the Head Start schedule creates, policy-makers have made some efforts to provide funds to provide ‘‘wrap-around’’ childcare in order to create a full-day program. Our findings suggest that additional efforts to make Head Start a full-day, full-year program might enable more low-income families to achieve the dual goals of economic self-sufficiency and educational enrichment for their children. In addition, reducing the barriers between separate funding streams, especially between Head Start and childcare, might help creating a more seamless system of services for

ARTICLE IN PRESS Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

low-income families. Finally, targeting Head Start primarily to families below the poverty threshold may lead to discontinuity of care when a family transitions off welfare. This discontinuity has implications for parental job stability if new care arrangements are difficult to find. Moreover, instability of care may have negative effects on the development of economically disadvantaged children. Some parents leaving welfare may increase their earnings sufficiently to rise above the official poverty threshold, but most still have very low incomes and have children who could benefit from early educational opportunities. Policy makers might consider a sliding fee scale for children in low-income families, especially for those in which parents are working. Acknowledgements This paper is part of Next Generation, a project by MDRC that examines the effects of welfare, antipoverty, and employment policies on children and families. The authors thank Sharon McGroder and Hiro Yoshikawa for comments and suggestions.

References Barnett, W. S. (1995). Long-term effects of early childhood programs on cognitive and school outcomes. Future of Children, 5, 25–50. Barnett, W. S. (1998). Long-term cognitive and academic effects of early childhood education on children in poverty. Preventive Medicine, 27, 204–207. Bloom, D., & Michalopoulos, C. (2001). How welfare and work policies affect employment and income: A synthesis of research. New York: Manpower Demonstration Research Corporation. Bos, J. M., Huston, A. C., Granger, R. C., Duncan, G. J., Brock, T., & McLoyd, V. C. (1999). New hope for people with low incomes: Two-year results of a program to reduce poverty and reform welfare. New York: Manpower Demonstration Research Corporation. Crosby, D. A., Gennetian, L. A., & Huston, A. C. (2005). Child care assistance policies can affect the use of center-based care for children in low-income families. Applied Developmental Science, 9(2), 86–106. Currie, J. (2001). A Fresh Start for Head Start? Children’s Roundtable no. 5. Washington, DC: The Brookings Institution. Currie, J., & Thomas, D. (1999). Does Head Start help Hispanic children? Journal of Public Economics, 74, 235–262. Currie, J., & Thomas, D. (2000). School quality and the longerterm effects of Head Start. The Journal of Human Resources, 35, 755–774. Flynn, M., & Hayes, C. D. (2003). Blending and Braiding Funds to Support Early Care and Education Initiatives. Financing strategy series. Washington, DC: The Finance Project.

31

Garces, E., Thomas, D., & Currie, J. (2002). Long-term effects of Head Start. American Economic Review, 92(4), 999–1012. Gassman-Pine, A. (2003). The effects of welfare and employment programs on the child care use of low-income young mothers. Next generation working paper no. 19. New York: Manpower Demonstration Research Corporation. Gennetian, L. A., Crosby, D. A., Huston, A. C., & Lowe, E. (2004). Can child care assistance in welfare and employment programs support the employment of low-income families. Journal of Policy Analysis and Management, 23(4), 723–743. Gennetian, L. A., & Michalopoulos, C. (2003). Child care and employment: Evidence from random assignment studies of Welfare and work program. Next generation working paper no. 17. New York: Manpower Demonstration Research Corporation. Gennetian, L. A., & Miller, C. (2000). Reforming welfare and rewarding work: Final report on the Minnesota Family Investment Program. New York: Manpower Demonstration Research Corporation. Greenberg, M. H., Levin-Epstein, J., Hutson, R. Q., Ooms, T. J., Schumacher, R., Turetsky, V., et al. (2002). The 1996 welfare law: Key elements and reauthorization issues affecting children. Children and Welfare Reform, 12, 27–57. Hamilton, G., Freedman, S., Gennetian, L. A., Michalopoulos, C., Walter, J., Adams-Ciardullo, D., et al. (2001). How effective are different welfare-to-work approaches? Five-year adult and child impacts for eleven programs. New York: Manpower Demonstration Research Corporation. Hart, K., & Schumacher, R. (2004). Moving Forward: Head Start children, families, and programs in 2003. CLASP policy brief Head Start series no. 5. Washington, DC: Center for Law and Social Policy. Hofferth, S. (1994). Who enrolls in Head Start? A demographic analysis of Head Start-eligible children. Early Childhood Research Quarterly, 9, 243–268. Hofferth, S., Stanhope, S., & Harris, K. (2002). Exiting welfare in the 1990s: Did public policy influence recipients’ behavior? Population Research and Policy Review, 21, 433–472. Huston, A. C., Duncan, G. J., Granger, R. C., Bos, J. M., McLoyd, V. C., Mistry, R., et al. (2001). Work-based antipoverty programs for parents can enhance the school performance and social behavior of children. Child Development, 72(1), 318–337. Huston, A. C., Miller, C., Richburg-Hayes, L., Duncan, G. J., Eldred, C. A., Weisner, T. S., et al. (2003). New Hope for families and children: Five-year results of a program to reduce poverty and reform welfare. New York: Manpower Demonstration Research Corporation. Irish, K., Schumacher, R., & Lombardi, J. (2003). Serving America’s Youngest: A snapshot of early Head Start children, families, teachers, and programs in 2002. Policy brief no. 3. Center for Law and Social Policy. Washington, DC. Jones, M. G. (2003). Coordinating with Head Start programs to support low-income families. Issue notes (vol. 7(5)). Washington, DC: Welfare Information Network. Lamb-Parker, F., Piotrkowski, C. S., Baker, A. J. L., KesslerSklar, S., Clark, B., & Peay, L. (2001). Understanding barriers to parent involvement in Head Start: A research community partnership. Early Childhood Research Quarterly, 16, 35–51. Lee, V. E., Brooks-Gunn, J., Schnur, E., & Liaw, F. (1990). Are Head Start effects sustained? A longitudinal follow-up comparison of disadvantaged children attending Head Start,

ARTICLE IN PRESS 32

Y.E. Chang et al. / Economics of Education Review 26 (2007) 17–32

no preschool, and other preschool programs. Child Development, 61, 495–507. NICHD Early Child Care Research Network. (1998). Early child care and self-control, compliance, and problem behavior at 24 and 36 months. Child Development, 69, 1145–1170. NICHD Early Child Care Research Network. (2000). The relation of child care to cognitive and language development. Child Development, 71, 960–980. Phillips, D. A., & Cabrera, N. J. (1996). Beyond the blueprint: Directions for research on Head Start’s families. Washington, DC: National Academy Press. Quint, J. C., Bos, J. M., & Polit, D. F. (1997). New chance: Final report on a comprehensive program for young mothers in poverty and their children. New York: Manpower Demonstration Research Corporation. Ramey, S. L., Ramey, C. T., Phillips, M. M., Lanzi, R. G., Brezausek, C., Katholi, C. R., et al. (1999). Head Start children’s entry into public school: A report on the National Head Start/ Public School Early Childhood Transition Demonstration Study. Poster paper presented at the biennial meeting of the Society for Research in Child Development, Albuquerque, NM. Rensnick, G., & Zill, N. (2003). Understanding quality in Head Start classrooms: The role of teacher and program-level factors. Paper presented at the Biennial Meeting of the Society for Research on Child Development, Tampa, FL.

US Department of Health and Human Service. (1975). Head Start program performance standards: Appendix B-the parents (DHHS Publication no. OHDS 84-31131). Washington, DC: US Government Printing Office. US Department of Health and Human Service. (1998). Head Start act (as amended October 27, 1998). Washington, DC: US Department of Health and Human Service. US Department of Health and Human Service. (1999). Eligibility, recruitment, selection, enrollment and attendance in Head Start, http://www.headstartinfo.org/publications/perf_stds/ part1305.htm. US Department of Health and Human Service. (2000). Faces Findings: New Research on Head Start Program Quality and Outcomes, http://www/acf.dhhs.gov/programs/core/ongoing_ research/faces/faces_intro.html. US Department of Health and Human Service. (2004). The 2004 family income guidelines for 2004, http://www.headstartinof.org/publications/im04/im04_03a.htm. US Department of Health and Human Service. (2005). Head Start program fact sheet, http://www.2.acf.dhhs.gov/programs/hsb/ research/2005.htm. Zigler, E., Abelson, W. D., Trickett, P. K., & Seitz, V. (1982). Is an intervention program necessary in order to improve economically disadvantaged children’s IQ scores? Child Development, 53, 340–348.