Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
3 Trends in the Care of Children W orking parents in the United States use a variety of child care providers, including care by family members (parents, grand- parents, other relatives, and self-care), care by neighbors and friends (nonrelative care in the child's home or in the provider's home), family day care homes (care for a small group of children in a provider's home), child care centers, and other organized activities. In this chapter, we describe the types of care used by parents, the supply of care, family expen- ditures on child care, and quality. Given the evidence presented in Chapter 2 showing a rise in the labor force participation among mothers, we focus in this chapter on child care usage by families in which the mother is employed. PATTERNS OF CHILD CARE USE The evidence presented is drawn from the Survey of Income and Pro- gram Participation (SIPP) from spring 1999. We first discuss children of preschool age, defined here as ages 0 to 5, and then children of school age, ages 6 to 14. Preschool-Age Children According to data from the 1999 SIPP, there were an estimated 22.0 million children ages 0 to 5 in the United States in spring 1999 (March- June), of whom 12.2 million (55 percent) had an employed mother. More 42
TRENDS IN THE CARE OF CHILDREN 43 than a third of the preschool children (39 percent) of working parents had more than one child care arrangement during the week, and the average number of arrangements per child was 1.6. The primary child care arrange- ment was defined as the arrangement used for the most hours per week.1 Average hours per week in the primary arrangement were 32.2, and total hours in child care per week, including primary and other arrangements, were 39.1. Working parents traditionally often relied on relatives to care for their children. Evidence suggests that the use of relatives for child care declined during the last half of the 20th century as parents began to rely more on family day care homes and child care centers (Hofferth et al., 1991; Uttal, 1999). However, in 1999, almost half the preschool children of working mothers in the United States were cared for by relatives (48 percent) for their primary arrangement (see Table 3-1). This included care by parents (20 percent), grandparents (21 percent), and other relatives (7 percent). In 1999, a little more than half of the preschool children of employed mothers (52 percent) were in some type of nonrelative care, including children in day care centers, nursery schools, or kindergarten (31 percent), family day care homes (11 percent), and nannies or babysitters in the child's home or other nonrelative arrangements (10 percent). Table 3-1 shows the distribution of primary child care arrangements of preschool age children for families classified by various characteristics. The proportion of children whose primary care arrangement is nonrelative care and the distribution of children across types of care vary substantially by household structure, education, race, ethnicity, nativity, income, location of the household, age of the child, and job characteristics of the mother. Unmarried mothers who were previously married use more nonrelative care than do married mothers, particularly center care. In total, 56 percent of their primary child care is in nonrelative arrangements compared with 53 percent for married mothers. This difference is probably due in part to lack of availability of the spouse to provide child care while the mother is at work. Black mothers are about 7 percentage points more likely to use center care than are whites, but blacks are less likely to use other forms of nonrelative care than whites. Hispanic mothers are less likely to use center care and family day care homes than whites. These patterns have been noted in other data sources, and they persist when other household charac- 1When there were two (or more) arrangements with the same number of hours per week, the primary arrangement was assigned in the following order of priority: center, family day care home, school, organized before- or after-school activity, other nonrelative, grandparent, other adult relative, sibling, mother while at work, child's other parent, self-care.
44 WORKING FAMILIES AND GROWING KIDS TABLE 3-1 Percentage Distribution of Employed Mothers' Primary Child Care Arrangements for Children from Birth to Age 5, Spring 1999 Family Demographic Day Other Grand- Other Characteristic Center Care Nonrelative Parent parent Relative All 31 11 10 20 21 7 Married 31 12 10 24 18 6 Widowed, divorced, 36 12 8 12 24 8 separated, married spouse absent Never married 28 8 10 11 32 11 White 32 14 9 22 18 5 Black 39 5 8 13 26 9 Hispanic 20 7 13 21 24 15 Other 24 3 7 20 40 7 Noncitizen 19 4 14 25 23 16 Native 32 12 9 20 20 6 Native citizen 28 7 9 15 33 8 < High school 24 5 10 15 23 20 High school 23 8 8 24 25 7 Some college 30 12 12 18 22 7 College graduate 36 16 9 20 15 3 Full time 33 13 10 15 21 7 Part time 25 8 8 31 21 8 Day shift 35 13 9 15 20 7 Nonday 19 5 10 33 24 9 < poverty 30 7 11 17 24 12 1-2 poverty 24 7 8 24 25 11 2+ poverty 34 13 10 20 19 5 Public assistance 33 8 9 10 30 11 No public 31 11 10 21 20 7 assisstance Age 0 17 11 9 29 26 8 Age 1 19 14 14 22 25 7 Age 2 24 13 11 23 23 7 Age 3 27 12 8 20 23 11 Age 4 42 10 9 15 18 7 Age 5 51 7 7 15 14 5 South 37 10 8 15 23 7 Non-South 28 12 10 23 20 8 (continued)
TRENDS IN THE CARE OF CHILDREN 45 TABLE 3-1 Continued Family Demographic Day Other Grand- Other Characteristic Center Care Nonrelative Parent parent Relative Metro 31 10 10 21 21 7 Nonmetro 30 17 10 16 20 8 NOTES: Data are weighted. Unweighted sample size is 3,066. Other nonrelative excludes school and includes babysitters, nannies, and other unspecified nonrelatives. Hispanics are excluded from the white, black, and other race categories. Public assistance includes Tempo- rary Aid for Needy Families, food stamps, and other programs such as General Assistance and SSI. Full time = 35+ hours each week in the month prior to the survey. Part time = 1-34 hours each week. SOURCE: Tabulations from SIPP 1996 panel, wave 10 (Spring 1999). teristics are held constant (see Blau and Hagy, 1998; Chaplin et al., 1999). Nonnaturalized immigrants use less center care and family day care homes than natives, but noncitizens use more care by other nonrelatives. Use of center and family day care arrangements increases with the mother's education. Only 24 percent of primary care arrangements for children of employed high school dropouts were in centers, compared with 30 percent for college attendees and 36 percent for college graduates. Fam- ily day care use rises from 5 percent for high school dropouts to 16 percent for college graduates. More educated mothers may be more aware of the potential benefits for child development of market child care and may be better able to afford such care. Nonrelative care is much more heavily used by mothers who work full time (56 percent) rather than part time (40 percent), and by mothers who work a day shift (58 percent) rather than some other work schedule (34 percent). The day/nonday difference is entirely due to higher use of centers and family day care homes by day shift workers, and it is no doubt ex- plained by the fact that many centers and family day care homes do not offer care at nonday shift hours.2 The full-time/part-time difference is also mainly in use of centers and family day care homes, and it is probably due 2Kisker et al. (1991:45) report that 10 percent of centers and 6 percent of regulated family day care homes in the Profile of Child Care Settings (explained further in the section on child care supply) sample offered weekend care, and 3 percent of centers and 13 percent of regu- lated family day care homes offered evening care.
46 WORKING FAMILIES AND GROWING KIDS to the fact that relative care arrangements may be easier to make for a few hours per day than for eight to nine hours per day. Families with incomes below the poverty line use more center care than families with incomes between one and two times the poverty line. This is probably due to greater access to child care subsidies through the welfare system. Families with incomes more than twice the poverty line also use more center care than the poor and the near-poor. This is probably due to the greater ability of higher-income families to pay for center care. Overall use of nonrelative care increases sharply with the age of the child after age 1, from 47 percent at age 1 to 65 percent at age 5. Most of this increase is accounted for by center care, rising from 17 percent at age zero to 51 percent at age 5. The developmental benefits of center care may be perceived by parents to be strongest beginning at age 3, as children approach school entry age. Moreover, families have fewer options for nonrelative care when their children are infants and toddlers, or they may prefer to have their very young children cared for by a family member given the convenience, proximity, and reliability that they may feel is more easily attainable using family members or friends (Folk and Yi, 1994). Patterns of nonrelative care use across regions of the United States are generally quite similar. The main exception is that center care is more heavily used in the South than in other regions. This pattern has been widely noted and has persisted over time (see Hofferth et al., 1991:Table 2.15; see also Capizzano et al., 2000a, for evidence on variation in child care use across states). It is also the case that the rate of full-time employ- ment by mothers of young children is higher in the South than in other regions (Blau, 2001:Chapter 2, note 4), but the connection between these two facts and the direction of causality are not clear. Patterns of nonrelative care use across metropolitan and nonmetropolitan areas are fairly similar.3 In summary, relative care is more often used by Hispanic families, families with less education, families with a mother who works part time, families with younger children, and families with lower incomes. Further- more, working nonstandard hours is associated with a greater reliance on relatives for child care (O'Connell, 1993; Casper, 1997; Casper and O'Connell, 1998). Among married, dual-earner households, split-shift work increases father participation in child care responsibilities, thus decreasing child care costs and increasing parent involvement (Presser, 1994; Brayfield, 3SIPP has been collecting data on child care since 1985, so in principle it is possible to examine trends in child care use over the period 1985-1999. However, there have been several important changes in survey design that make comparisons over time difficult. Smith (2002) shows trends in child care use in the SIPP data for children ages 0-4 (not 0-5, as in Table 3-1) with employed mothers. The results show substantial fluctuation over time in the use of nonrelative care (as defined here), with some suggestion of a decline in recent years.
TRENDS IN THE CARE OF CHILDREN 47 1995; Garey, 1999; Hoffman et al., 1999). Grandparents are particularly involved in child care for single mothers who work nonstandard hours (Presser, in press; see Hofferth et al., 1991, for similar findings on use of relative care by Hispanics). Finally, as noted earlier, 39 percent of preschoolers with employed mothers had more than one child care arrangement: 28 percent had two arrangements, and 12 percent had three or more arrangements. The per- centage of children in multiple arrangements is especially high when moth- ers work nonstandard hours; for example, 51 percent when mothers work a rotating schedule (Presser, in press). In 1999, the majority (60 percent) of secondary arrangements were in relative care, compared with 49 percent of primary arrangements. Use of multiple arrangements is most common when the primary arrangement is a parent (45 percent have a secondary arrangement) or a center (48 percent). Average hours per week in second- ary arrangements are 15. School-Age Children There were an estimated 35 million children in the United States ages 6 to 14 in spring 1999, of whom 22 million (63 percent) had an employed mother. For the great majority of the children of employed mothers (80 percent), the primary child care arrangement was school. School serves as a child care arrangement in most cases for six to seven hours per day during the school year. As noted later, unsupervised care increases during the summer months. Table 3-2 classifies children ages 6 to 14 according to their primary arrangement other than school. Arrangements for school-age children are classified as center, nonrelative (including family day care homes), orga- nized activities, parent, grandparent, other relative, and self-care. Orga- nized activities include before- and after-school programs, lessons, sports, and clubs. The distribution of primary nonschool arrangements for chil- dren ages 6 to 14 was 5 percent in center care, 9 percent in nonrelative care, 12 percent in organized activities, 37 percent by parents, 14 percent by grandparents, 12 percent by other relatives, and 12 percent by self-care. Average hours per week in the primary care arrangement were 14.4, with longer hours in centers and nonrelative care (21.4 and 18.8, respectively) and shorter hours in organized activities (10.4). Average hours per week in all arrangements other than school totaled 20.7. Thus, the typical school- age child is in some kind of child care for a significant amount of time. Use of the different types of child care for school-age children did not vary much by marital status, but married mothers were more likely than other mothers to use organized activities. This pattern appears throughout the table: use of organized activities is higher for more advantaged families,
48 WORKING FAMILIES AND GROWING KIDS TABLE 3-2 Percentage Distribution of Employed Mothers' Primary Child Care Arrangements Other Than School for Children Ages 6-14, Spring 1999 Demographic Other Organized Grand- Other Characteristic Center Nonrelative Activity Parent parent Relative Self All 5 9 12 37 14 13 12 Married 5 8 12 42 10 11 12 Widowed, divorced 6 10 11 24 20 16 14 separated, married spouse absent Never married 5 11 10 22 26 19 8 White 5 8 13 39 11 11 14 Black 5 8 12 31 19 17 9 Hispanic 4 13 8 33 17 19 7 Other 4 9 8 37 19 13 10 Noncitizen 3 10 5 43 12 20 6 Native 5 9 13 36 13 12 13 Native citizen 8 13 8 38 14 12 7 < High school 3 8 6 37 14 24 8 High school 5 8 9 37 17 13 12 Some college 5 10 11 36 14 12 13 College graduate 6 9 18 37 9 9 13 Full time 6 10 11 32 15 14 13 Part time 2 7 12 49 11 11 9 Day shift 5 9 13 35 13 13 13 Nonday 3 9 8 44 16 13 7 <poverty 5 8 8 37 19 16 8 1-2 poverty 3 9 10 37 15 18 9 2+ poverty 5 9 13 37 12 11 14 Public assistance 4 10 7 25 24 22 8 No public assistance 5 9 12 37 13 12 12 Age 6 11 17 10 35 18 9 1 Age 7 8 12 12 35 21 11 1 Age 8 6 12 12 38 16 13 2 Age 9 5 14 10 37 17 13 5 Age 10 5 9 13 40 13 14 6 Age 11 5 8 13 36 13 15 11 Age 12 2 5 11 37 11 15 19 Age 13 0 2 12 37 7 15 27 Age 14 1 2 10 34 6 11 33 South 7 8 12 32 15 15 10 Non-South 9 10 10 39 13 12 13 (continued)
TRENDS IN THE CARE OF CHILDREN 49 TABLE 3-2 Continued Demographic Organized Grand- Other Characteristic Center Nonrelative Activity Parent parent Relative Self Metro 5 9 12 37 13 13 12 Nonmetro 3 9 12 36 14 14 13 NOTES: Data are weighted. Unweighted sample size is 6,489. Nonrelative includes family day care homes, babysitters, nannies, and other unspecified nonrelatives. Hispanics are excluded from the white, black, and other race categories. Public assistance includes Temporary Aid for Needy Families, food stamps, and other programs such as General Assistance and SSI. Full time = 35+ hours each week in the month prior to the survey. Part time = 1-34 hours each week. SOURCE: Tabulations from SIPP 1996 panel, wave 10 (Spring 1999). regardless of whether advantage is measured by marital status, race/ ethnicity, education, income, receipt of public assistance, or nativity. Nonrelative care use for school-age children increases with the mother's education, and the gap in the use of organized activities is especially large by maternal education, with only 6 percent of high school dropouts com- pared with 18 percent of college graduates using organized activities. Nonrelative care is used more often by mothers who work full time (27 percent compared with 21 percent for part time) and by day shift workers (27 percent compared with 19.7 percent for nonday shift workers). The biggest differences that appear in the table are by the child's age, with 38 percent of 6-year-old children in nonrelative care, 27 percent of 10-year-old children in nonrelative care, and 13 percent of 14-year-old children in nonrelative care. These differences undoubtedly reflect parental percep- tions of differences in children's needs and capabilities by age, as well as the changing interests of children and the availability of appropriate programs for those age groups. Finally, self-care, or care without adult supervision, has been a major policy concern for school-age children. According to the 1999 SIPP data, the extent to which children spend at least some time caring for themselves varies somewhat by family characteristics, although these differences are small. For example, a higher proportion of children with married mothers spend time in self-care (12 percent) than children of never-married women (8 percent); white children spend more time in self-care (14 percent) than black (9 percent) or Hispanic (7 percent) children; children of full-time working mothers spend more time in self-care (13 percent) than children of part-time working mothers (9 percent); and children with mothers working the day shift spend more time in self-care (13 percent) than children with mothers working nonday shifts (7 percent). The use of self-care varies more by age of child than it does for family characteristics. For example, a third
50 WORKING FAMILIES AND GROWING KIDS of 14-year-old children spent time in care without adult supervision com- pared with less than 1 percent of 6-year-olds. Overall, the data on self-care should be interpreted with caution. The estimates of the number of children without adult supervision have varied across reports and depends somewhat on the month the question is asked (whether school is in session or not), the parents' interpretation of the term "self-care," and the parents' willingness to admit that their children are left home alone. For example, a study released by the Urban Institute based on data from the 1999 National Survey of America's Families (Capizzano et al., 2002) found that while 1 in 10 children under age 13 spent time alone, these children spent twice as much time unsupervised in the summer than during the school year. CHILD CARE EXPENDITURES In 1999, there were an estimated 21.8 million families with an em- ployed mother and at least one child ages 0 to 14. Table 3-3 (first row) shows that 43 percent of these families reported making a cash payment for child care for at least one child. The average weekly payment for all child care arrangements for all children among families that made any payment was $76.4 The average percentage of family income spent on child care was 7.5 percent, almost identical to the figures for 1997 and 1995 and up slightly from 7.3 percent in 1993, 7.1 percent in 1991, 6.8 percent in 1988, and 6.3 percent in 1986 (Smith, 2000).5 The fact that only 43 percent of employed mothers paid for child care may seem surprising, but the rate of use of nonrelative care by employed mothers of children, averaged over all children ages 0 to 14, is 42.1 percent. The rate of use of paid care was 56.8 percent for families with children ages 0 to 5 only, and 31.1 percent for families with children ages 6 to 14 only. The former group averaged $89 per week in child care expenditures (9 percent of family income), and the latter group averaged $55 per week (5.3 percent).6 4 This is sharply lower than the average weekly payment of $85 reported in the 1995 SIPP (Smith, 2000). This may be a result of the change from a fall interview in 1995 to a spring (March-June) interview in 1999. Smith (2002) suggests that the spring interview period includes months in which children were off from school and may not have been in their regular arrangement. 5 The SIPP questions on child care expenditure are intended to measure out-of-pocket cash expenditure by families. Since most child care subsidies do not take the form of cash, we can be reasonably confident that subsidies are not included in reports of cash expenditure. 6 These payments include a variety of hours of care, ranging from less than five hours a week to more than 40 hours a week. See Smith (2002) for a breakdown of payments by different characteristics.
TRENDS IN THE CARE OF CHILDREN 51 of % Income 5.3 3.7 9.3 9.8 4.5 7.5 8.5 5.8 7.0 4.1 2.9 (continued) 1999 Children Spring with Weekly Expenditure 55 55 58 53 52 56 65 83 66 56 31 Only Families 6-14 % Pay 31.1 31.9 22.5 28.1 33.0 24.2 31.3 24.6 43.7 31.0 16.9 Mother, of % Income 9.0 6.8 7.5 5.4 7.1 16.2 12.5 11.8 16.4 10.2 Employed an Children with 89 95 90 69 92 73 89 83 98 74 with Weekly Expenditure Only Families Families 0-5 % Pay 56.1 58.0 63.5 48.4 61.4 45.4 48.0 38.4 54.7 58.3 in of % Income 7.5 5.6 6.3 7.3 7.7 7.0 4.1 2.9 12.4 13.1 10.6 11.5 10.1 Expenditure Care Weekly Expenditure 76 78 71 66 75 71 80 88 96 82 66 52 31 Child Families All % Pay 43.0 44.7 37.5 40.9 45.9 35.9 41.3 30.6 56.3 58.5 43.7 31.0 16.9 Family child Total married 3-3 divorced absent married youngest of separated, spouse 0-2 3-5 6-8 9-11 12-14 TABLE Demographic Characteristic All Married Widowed, Never White Black Hispanic Other Age
52 WORKING FAMILIES AND GROWING KIDS ies. of % Income 6.4 5.6 3.0 5.4 4.6 5.3 5.3 8.3 4.6 3.3 2.3 8.9 4.2 2.5 11.5 19.5 categor race Children other with Weekly Expenditure 62 57 54 54 59 42 56 50 50 52 61 74 52 58 48 57 and Only Families 6-14 % Pay 18.1 24.6 32.9 43.0 32.6 28.2 32.0 27.3 21.6 35.7 45.1 46.9 20.8 26.5 29.5 36.7 black, white, week. of the % Income 9.9 8.2 6.6 8.6 5.7 8.4 7.6 6.5 5.0 7.2 5.0 19.2 11.8 11.5 22.2 11.9 each from hours Children 1-34 excluded = 75 73 81 93 77 93 71 66 93 58 81 81 with Weekly Expenditure 111 126 147 106 are time Only Families 0-5 Part % Pay 49.4 50.2 54.8 65.5 63.9 41.1 62.5 38.1 46.5 65.1 69.9 71.3 51.0 47.9 57.4 62.2 Hispanics 1999). survey. of 5,864. the (Spring % Income 8.5 7.4 5.1 7.2 6.2 7.3 8.5 6.1 5.0 3.6 5.9 3.9 15.2 10.5 22.8 10.6 is to 10 size prior wave sample month panel, Weekly Expenditure 73 67 71 88 79 63 79 63 64 73 97 65 70 67 84 108 the in 1996 Families Unweighted SIPP All % Pay 30.0 36.7 43.6 55.7 45.8 37.5 45.3 35.2 34.7 48.3 56.0 57.4 35.1 36.3 43.2 49.0 week each from weighted. Continued hours are 3-3 35+ earnings Tabulations school graduate 's Data = income school college time time shift $18,000 $18,000 time High < $18-35,999 $36-53,999 $54,000+ < $18-35,999 $36-53,999 $54,000+ TABLE Demographic Characteristic < High Some College Full Part Day Nonday Mother Family NOTES: Full SOURCE:
TRENDS IN THE CARE OF CHILDREN 53 The other rows of the table illustrate how family child care expendi- tures differ by child and household characteristics. In most cases, married mothers are more likely to pay for child care than are nonmarried mothers and spend more if they do pay, but the percentage of family income spent by married mothers on child care is less than half the percentage for nonmarried mothers. White families are 5 percentage points more likely to pay for child care than Hispanic families, who are in turn 5 points more likely to pay for care than black families. Black and Hispanic families who do pay for care pay a greater share of income than others (10.6 percent to 11.5 percent compared with 6.3 for whites). The rate at which families use paid child care peaks when the youngest child is ages 3 to 5, but the amount and the percentage of family income paid are highest when the youngest child is ages 0 to 2, at $96 per week and 10.1 percent of family income. The percentage paying for care increases with education, from 30 per- cent for high school dropouts to 55.7 percent for college graduates. The amount paid is higher for college graduates at $88 than for the other education groups ($67-73), while the percentage of income spent on child care falls monotonically with education, from 15.2 percent for high school dropouts to 5.1 percent for college graduates. These patterns also appear for families with only children ages 0 to 5 and for families with only children ages 6 to 14. Similar patterns are evident when families are classi- fied by income or by the mother's earnings, with the single difference being the much sharper gradient in the percentage of income spent on child care. For example, 22.8 percent of income is spent on child care by families with annual incomes less than $18,000, while only 3.9 percent of income is spent on child care by families with income over $54,000 per year. It is interest- ing to note that the amount paid rises more sharply with increases in the mother's earnings than with increases in family income. For example, among families with only children ages 0 to 5, the average payment rises from $58 to $106 from the lowest to the highest income group, and from $66 to $147 from the lowest to highest mother's earnings group. This pattern suggests a higher propensity to spend on child care out of the mother's earnings than out of other income sources. This is consistent with findings that income controlled by mothers is spent disproportionately on child-related goods and services compared with other sources of family income (for some evidence on this from studies done in the United King- dom, see Lundberg et al., 1997, and in Brazil, see Thomas, 1990). Of mothers who are employed full time, 45.8 percent pay for care; of mothers employed part time, only 37.5 percent pay for care. The differen- tial is even larger in families with only young children (63.9 compared with 41.1 percent). Part-time workers spend less on child care, but not in proportion to their hours worked. This suggests that part-time care is significantly more costly per hour than full-time care, a finding confirmed by data from other sources (Hofferth et al., 1991). If child care is priced by
54 WORKING FAMILIES AND GROWING KIDS the week or the month, then families using part-time care may be forced to pay for a full-time slot. It is also possible that families seeking full-time care choose less expensive care. The day/nonday shift differences are similar to the full-time/part-time differences. CHILD CARE SUPPLY The simplest way to measure the supply of child care is to consider the licensed capacity of existing day care centers and family day care homes. However, this approach is problematic for our purposes for several rea- sons. First, many family day care homes and some centers and preschools are legally exempt from licensing and registration requirements, and there- fore they are not found in state licensing lists. Second, the availability of babysitters, nannies, and relatives cannot be estimated in this manner. Third, licensed capacity does not distinguish between care provided during working and nonworking hours of employed mothers. Fourth, and most important, licensed capacity is a short-run measure that reflects the existing level of demand as well as supply. Should demand for nonrelative care rise in the future, one would expect licensed capacity to increase as well, through expansion of existing facilities and entry of new establishments. Bearing in mind the shortcomings of licensed capacity as a measure of supply, the following data provide an overview of the supply of nonrelative care. The Profile of Child Care Settings (PCS) survey of day care centers and regulated family day care homes estimated that in 1990 there were 80,072 day care centers with a licensed capacity of 5,333,067 children; and 117,995 registered family day care homes with a licensed capacity of 859,506 children (Kisker et al., 1991). The National Child Care Survey (NCCS) of 1990 estimated that 6,104,000 children under 13 were enrolled in a day care center in 1990 for their primary child care arrangement (Hofferth et al., 1991).7 The excess of enrollment over licensed capacity 7The PCS was a telephone survey of a sample of centers and family day care homes drawn from state lists of licensed facilities. A stratified random sample of counties was selected (with probability proportional to the number of children under age 5), and a stratified ran- dom sample of early education and care providers was drawn from the licensing list for each county. Counties were stratified by region, metropolitan status, and poverty level. Providers were stratified by whether they were home-based, Head Start, public school-based, or other center-based. The PCS data are over ten years old, but there have not been any nationally representative surveys of providers since the PCS. The data from the PCS should be inter- preted with the caveat that we do not know whether the patterns documented in 1990 have changed significantly. Failure to regularly collect data on providers is a significant shortcom- ing of the U.S. child care data collection system, as we discuss in Chapter 9 of this report. The NCCS was a telephone survey of households with children under age 13 located in the same counties drawn for the PCS survey. The sample was selected randomly from telephone numbers in the selected counties.
TRENDS IN THE CARE OF CHILDREN 55 may be accounted for by day care centers and preschools exempt from licensing. The NCCS also estimated that in 1990 3,193,000 children were enrolled in a family day care home as the primary child care arrangement. The large difference in this case between licensed capacity and enrollment undoubtedly reflects the fact that a large majority of family day care homes are believed to be unlicensed (National Research Council, 1990). A total of 37 percent of centers were for profit, of which 28 percent were independent and 9 percent were part of a national or local chain. Of the 63 percent of nonprofit centers, 6 percent were Head Start programs, 7 percent were based in public schools, 18 percent were religious-sponsored, 7 percent were sponsored by other groups (government agencies, community agen- cies, employers, etc.), and the remaining 25 percent were independent. These data are more than 12 years old, and there has not been a more recent national data collection that would reflect current licensed capacity. More recent data collected by the Children's Foundation (CF) indicate that there were 113,506 licensed day care centers in 2001, and 306,246 regulated family day care homes (Children's Foundation, 2002). However, the Children's Foundation does not provide data on licensed capacity. It has gathered data for over a decade on the number of centers listed with licensing agencies, and its figures for 1990-1991 can be compared with data from the PCS. The Children's Foundation estimate of the number of cen- ters in 1991 was 86,212, compared with an estimate of 80,072 by the PCS in 1990. This is a reasonably close correspondence. However, the Children's Foundation figure for family day care homes in 1990 was 223,351, much larger than the PCS estimate of 117,995. The reason for this large discrepancy is unclear. Another source of information on availability of child care is the Cen- sus of Services, conducted every five years by the U.S. Census Bureau. The 1997 Census of Services found a total of 62,054 day care centers (establish- ments with payroll), which is only about two-thirds the 97,046 figure for 1997 from the Children's Foundation (U.S. Census Bureau, 2002b). The 1992 Census of Services estimated 489,054 family day care homes (estab- lishments without payroll), compared with 265,347 in the Children's Foun- dation 1992 survey. The former figure applies to all family day care homes, while the latter refers only to regulated homes. The large differences across sources in estimates of the number of centers and family day care homes as well as the absence of recent information on licensed capacity illustrate the problems in measuring availability of private child care, thus making it difficult to draw any firm conclusions about the availability of private child care services. Yet another source of information related to the supply of child care is the recent estimate of the size and components of the U.S. Child Care Workforce and Caregiving Population conducted by the Center for the
56 WORKING FAMILIES AND GROWING KIDS Child Care Workforce and the Human Services Policy Center at the Univer- sity of Washington (May 2002), which was funded by the U.S. Department of Health and Human Services. Focusing on paid caregivers for children ages 0 to 5, the study estimates that there are 2.3 million such caregivers in the United States at a single point in time, including 550,000 (24 percent) working in child care centers, 650,000 (28 percent) providing family day care, 804,000 (35 percent) functioning as paid relatives other than family day care providers, and 298,000 (13 percent) as paid nonrelatives other than those working in centers or family day care programs (i.e., nannies). In addition to the paid child care workforce, the study estimated that approximately 2.4 million individuals provide unpaid child care during a given week. Most (93 percent) of these individuals are unpaid relatives, and the remainder are volunteers in centers or unpaid, nonrelative caregivers. No current national data track how much time and effort working families spend trying to find child care at a price they are able to pay. Several recent reports (U.S. General Accounting Office, 1997; Mezey et al., 2002) suggest, as would be expected and is implied by data reported earlier in this chapter, that the lack of adequate supply is especially trying for parents of infants and toddlers, for children with special needs, for older schoolchildren, and for children of families working nonstandard hours. It is important that the requisite data be collected for the nation on a regular basis. CHILD CARE QUALITY Two main approaches have been used to characterize the quality of child care. One is based on structural features of the child care setting that are thought to affect the developmental appropriateness of the care for children. These features include the size of the group in which care is provided, the ratio of adult caregivers to children, the overall education level and specialized early childhood education and training of the provid- ers, and the stability of the setting as measured by the turnover rate of the care providers. The other approach to measuring quality uses direct obser- vation of the developmental appropriateness of the care received, as re- corded by trained observers using standardized instruments. The ratings made by the observers are subjective in the sense that the observer makes a judgment about where on a given ordinal scale a child care setting lies (for example, a 1-7 ordinal scale or a "not at all," "somewhat," or "highly" characteristic scale). However, raters can be trained to produce ratings that are highly correlated with ratings of the same settings by other observers. And these process measures are believed to be more directly related than structural measures to child outcomes of interest.
TRENDS IN THE CARE OF CHILDREN 57 The most recent nationally representative data on the structural mea- sures are from 1990, and no nationally representative data are available on the process measures. Once again, failure to collect such data is a signifi- cant shortcoming of the child care data collection system in the United States. Here, we summarize the available information on quality. The Profile of Child Care Settings (Kisker et al., 1991) collected infor- mation on structural classroom characteristics from a nationally represen- tative sample of child care centers and regulated family day care homes through a telephone survey in 1990. There are limitations to telephone surveys for the purpose of measuring the structural characteristics, com- pared with recording them by direct on-site observation. Child care class- rooms are in flux during the day, with groups often changing composition and merging with other groups for some activities for part of the day and teachers moving among classrooms (Helburn, 1995). The PCS telephone survey provided a static snapshot of a typical period during the day, but it did not capture the dynamics of classroom composition throughout the day. Nevertheless, these data provide a useful overview of the distribution of quality characteristics. It is also important to bear in mind that regulated family day care homes are unlikely to be representative of unregulated day care homes, and that the latter are thought to be far more numerous than the former. Table 3-4 summarizes characteristics of centers and family day care homes (see Kisker et al., 1991, for more details). Average group size was 16 in centers and 7 in regulated homes. Group size increased with the age of children in centers, but not in family day care homes. The average group size in centers was 7 for infants, 10 for 1-year-olds, 12 for 2-year-olds, and 17 for 3- to 5-year-olds, which were within the range of maximum group size recommended by the National Association for the Education of Young Children (NAEYC) (see National Research Council, 1990) at that time. The average child-staff ratio was 9:1 in centers and 6:1 in regulated homes. The ratio of 4:1 for infants was at the high end of the range recommended by NAEYC at that time, and the ratios of 6.2:1 for 1-year-olds and 7.3:1 for 2-year-olds exceeded the NAEYC recommended ranges for these age groups. The average of 9.9 for 3- to 5-year-old children was at the high end of the NAEYC recommended level. The great majority of children in centers are 3 to 5 years old, so the majority of classrooms were (barely) within the range recommended by NAEYC. The annual turnover rate of teachers reported by centers was 25 per- cent (half the centers in the sample reported no turnover, and the other half reported turnover averaging 50 percent annually). With regard to educa- tion of the teachers, almost half (47 percent) had a four-year degree, 39 percent had some college education, 13 percent had a high school diploma or graduate equivalency degree (GED), and there were virtually no high
58 WORKING FAMILIES AND GROWING KIDS TABLE 3-4 Characteristics of Day Care Centers and Regulated Family Day Care Homes, 1990 Regulated Family Characteristic Day Care Centers Day Care Homes Average group sizea 16 7 Infants only 7 7 1-year-olds only 10 7 2-year-olds only 12 7 3-5-year-olds only 17 8 Average child-staff ratioa 9 6 Infants only 4 6 1-year-olds only 6 6 2-year-olds only 7 6 3-5-year-olds only 10 7 Annual rate of teacher turnover 25% Percentage of centers with any turnover 50% Turnover rate in centers with turnover 50% Average percentage of teachers with: At least a BA/BS degree 47% 11% Some college 39% 44% High school degree or GED 13% 34% No degree or GED 1% 11% Percentage of teachers who have had:b CDA training 25% 6% Teacher training 35% Other education training 40% Child care workshops or courses 54% 43% Child development or psychology courses 36% 28% Nurse or health training 26% Training by a R&R or government agency 5% 5% Social service training 4% 2% Other training 6% aExcludes programs that serve primarily handicapped children. bThe training information for center refers only to private, nonreligious-sponsored centers. SOURCE: Kisker et al. (1991). school dropouts (1 percent). Operators of regulated family day care homes had much less education, with only 11 percent having graduated from college, 40 percent with some college education, 34 percent with a high school diploma or GED, and 16 percent who were high school dropouts. Specialized training in early education, child development, or child care is also more common among center staff than in family day care homes. One- quarter of day care center teachers had earned a child development associ-
TRENDS IN THE CARE OF CHILDREN 59 ate (CDA) credential, compared with only 6 percent of family day care home operators. Many studies (summarized in National Research Council and Institute of Medicine, 2000) have found that more formal education and specialized training for caregivers, a higher ratio of caregivers to children in their care, and lower caregiver turnover rates are, other things being equal, associated with a higher quality experience for children. Whether these correlations represent causal effects is uncertain, but it is likely that better working conditions, including higher wages and smaller group size, attract caregivers with more formal education and specialized training, and the associated better outcomes for children are the result. As indicated earlier, there are no nationally representative samples of child care centers with measures of process quality. Two studies with reasonable sample sizes have measured process quality in site-specific samples of child care centers that may be representative of centers in the selected sites.8 The Cost, Quality, and Outcomes (CQO) study and the National Child Care Staffing Study used the same instruments to rate qual- ity--the Early Childhood Environment Rating Scale (ECERS) and its in- fant-toddler counterpart (ITERS). These instruments take about three hours to complete and rate each observed classroom on 30 to 35 items using a scale of 1 to 7 for each item. As a guide to the intended interpretation of the scores, ratings of 1, 3, 5, and 7 are designated by the instrument designers as representing inadequate, minimal, good, and excellent care, respectively (see Harms and Clifford et al., 1980, and Harms et al., 1990). Summary scores are obtained by averaging over the items. Table 3-5 presents descrip- tive statistics on quality ratings in day care centers from these two studies by site, age of children in the classroom, and the auspices of the center (for profit or nonprofit). It is important to bear in mind that the samples in these two studies are drawn from a small number of states, and while they are likely to be reasonably well representative of centers in those states, we have no way to determine whether they are nationally representative. The overall average rating in both studies is just under 4, or about halfway between minimal and good. The authors of the CQO report refer to this level of quality as "mediocre" (Helburn, 1995). Quality varies substantially across locations, with the highest quality sites (Boston, California, Connecticut) rated almost a full point above the lowest quality sites (Atlanta, North Carolina, Seattle). Classrooms with preschool-age children are almost always rated to be of higher quality than infant-toddler rooms, by a fairly wide margin in the 8A third study, discussed below, is based on a sample of children and consists of the centers and all other child care arrangements in which the sample children were enrolled.
60 the the of scale and the for ITERS Staffing (1.07) (1.07) (1.04) (1.13) (1.02) (1.07) (1.05) (0.72) (1.45) (0.97) (0.96) by and underlying same CQO Care Infant-Toddler 3.57 3.60 3.66 3.85 3.29 4.09 3.89 4.51 3.69 4.48 3.63 centers the the the description of Standard 227 ECERS has Child for in This between Measured (and analysis as (0.96) (0.97) (0.89) (0.99) (0.95) (0.97) (0.87) (0.61) (1.40) (0.89) (0.73) (1989) individual National al. factor Mean Nonprofit Preschool 4.41 4.66 4.25 4.33 4.31 4.39 4.30 4.72 4.14 4.79 3.99 et classrooms the measures. the on comparisons Centers 665 from and so Scales: (1993) (1989) and scores summary Care Whitebook the derived (CQO) study two Study (1.02) (0.70) (0.89) (1.07) (0.60) Study (0.98) (0.86) (0.57) (1.37) (0.83) (1.06) and Day differently, Rating the Study in Infant-Toddler 3.33 3.86 3.40 4.00 2.54 3.43 3.04 3.16 3.86 3.84 3.37 CQO include Study, measures of Staffing the not derived Outcomes CQO for does average was Outcomes Quality and Care the summary (0.99) (0.88) (0.85) (1.02) (0.83) (0.90) (0.84) (0.86) (1.04) (0.75) (0.86) but and Environment Profit of centers two Care Child NCCSS For Preschool 4.07 4.27 4.09 4.46 3.28 3.59 3.32 3.66 4.23 3.74 3.30 Quality, 401 CQO the unweighted in Quality, the Child includes the Toddler description Cost, National from it of are for set from Cost, (1.07) (0.96) (0.95) (1.05) (1.08) (0.99) (0.96) (0.72) (1.24) (0.90) (0.84) caution. the Infant Centers classrooms here data Rather, scores (1995) All 3.99 4.36 3.94 4.24 3.44 3.92 3.57 4.44 3.96 4.09 3.62 with from 731 and al. is et release score. presented ITERS Distribution made size Cryer and be public average 3-5 figures Tabulations See Childhood Sample The the Carolina should or The ECERS (NCCSS). Sites Sites the TABLE Early Deviation) All California Colorado Connecticut North All Atlanta Boston Detroit Phoenix Seattle NOTES: NCCSS. NCCSS. items items. as NCCSS SOURCES: Study
TRENDS IN THE CARE OF CHILDREN 61 CQO data.9 With only a few exceptions, nonprofit centers received higher average quality ratings than for-profit ones.10 The National Institute of Child Health and Human Development has been leading a longitudinal Study of Early Child Care (SECC) since the early 1990s. This study selected a sample of births in hospitals in 10 sites around the United States and has been following the children and their families ever since. Among other features of the study, the child care arrangements in which the study children were enrolled were visited and assessed for quality. Because this study was not limited to centers, a new quality assessment instrument was developed that could be used in a variety of different settings. This instrument, known as the Observational Record of the Caregiving Environment (ORCE), focuses on the caregiver's interac- tion with the study child, rather than on the overall quality of the arrange- ment. Like the ECERS and the ITERS, the ORCE takes about three hours to administer, and the final score is an average over the various observa- tions during the recording period. The scale for this instrument is 1 to 4, with 1 indicating that a particular dimension of positive caregiving was "not at all characteristic," 2 indicating "somewhat uncharacteristic," 3, "somewhat characteristic," and 4, "highly characteristic." Table 3-6 summarizes the distribution of ORCE scores at three differ- ent observation ages--15, 24, and 36 months--separately by type of child care arrangement (NICHD Early Child Care Research Network, 2000a). The most striking feature of the data is the relatively low quality of centers and family day care homes compared with care by the father, grandparent, or in-home babysitter. At age 15 months, positive caregiving is somewhat or highly characteristic of 28 percent of centers and 50 percent of family day care homes compared with 60 to 71 percent of fathers, grandparents, and in-home babysitters. The quality of the more informal types of care 9The ECERS and ITERS instruments are similar but not identical. It is not clear whether quality differences by age of children in the classroom are real or reflect different scales of the instruments. 10There is little systematic information on process quality in family day care homes. Kontos et al. (1995) studied about 200 family day care homes and relatives providing child care. They concluded that the majority of providers were providing care of adequate quality, about one third were providing inadequate quality care, and only 9 percent were providing good quality care. There is no published breakdown of quality by family characteristics in the CQO study. The NCCSS reported some quality breakdowns by a measure of socioeconomic status (SES) of the families, as reported by the center director. This is not a very useful measure, because directors were not given any instructions on how to define SES. Quality tends to be highest in centers reporting the highest family SES, lowest in the middle SES, and in-between for centers reporting the lowest SES (many of which are Head Start or other heavily subsidized pro- grams).
62 WORKING FAMILIES AND GROWING KIDS TABLE 3-6 Percentage Distribution of Observed Positive Caregiving in the NICHD Study of Early Child Care Positive Caregiving Rating Not at all Somewhat Somewhat Highly Child Age Characteristic Uncharacteristic Characteristic Characteristic 15 months Father 5 36 35 25 Grandparent 5 24 46 25 In-home 3 36 37 24 Family day care 7 44 34 16 Center 10 62 23 5 24 months Father 5 45 30 20 Grandparent 5 31 49 15 In-home 8 34 34 24 Family day care 9 50 28 14 Center 11 66 19 4 36 months Father 5 48 37 10 Grandparent 2 53 37 8 In-home 5 39 45 12 Family day care 2 63 29 5 Center 4 62 30 3 NOTES: The ORCE instrument has a rating scale of 1-4. The ratings here are based on averages over several time periods, with the following assignments: not at all characteristic = mean rating < 2; somewhat uncharacteristic = mean rating 2 to < 3; somewhat characteristic = mean rating 3 to < 3.5; highly characteristic = mean rating = 3.5. SOURCE: NICHD Early Child Care Research Network (2000a:Table 5). declines with the age of the children, as does the quality of family day care, while the quality of centers is lower at 24 than at 15 months but higher at 36 months than at either 15 or 24 months. By age 36 months, positive caregiving is somewhat or highly characteristic of 33 percent of centers and 34 percent of family day care homes compared with 47 percent of fathers, 45 percent of grandparents, and 56 percent of in-home babysitters. This differential may be due to the fact that the ORCE instrument emphasizes responsiveness to the individual child, and thus it might tend to produce higher quality ratings in small groups or individual care settings. It is important to emphasize that the figures presented in Table 3-6 are averages, and that there is substantial variation in quality within each type of care. This is illustrated for centers in Table 3-5 by the fact that the standard
TRENDS IN THE CARE OF CHILDREN 63 deviations within states and within type of child care center are on average about one point on a seven point scale. SUMMARY The evidence reviewed in this chapter highlights several important find- ings. The type of child care used by parents in the United States is very diverse in terms of setting, quality, and cost. Child care settings range from institutional arrangements, such as child care centers and preschools, to smaller family day care homes and in-home arrangements, with one adult hired to care for one child, to relative care. Child care is very diverse in quality, both within and across types of arrangements. About 56 percent of employed-mother families with preschool-age children only and 31 percent of families with older children only pay for child care. Among families that pay for child care, in 1999, the average percentage of family income spent on child care was 7.5 percent, for families with preschool age children only it was 9 percent, and it was 5.3 percent for families with school-age chil- dren. However, while low-income families are less likely to pay for care, child care is a major expenditure for those families who do pay, consuming about 23 percent of family income for families with annual incomes under $18,000. The average quality of child care in centers has been character- ized as "mediocre," and the quality of child care by relatives, in-home babysitters, and other informal providers has not been highly rated either (Kontos et al., 1995). Three implications of this portrait of child care are important for child development and public policy. First, the child care market has grown rapidly in the last 20 to 25 years, and many children and adolescents are spending many hours in the care of someone other than their parent. Nonrelative care is now the dominant source of child care in the United States, although relatives still play an important role in the care of children, particularly while they are very young. With regard to child care centers, the for-profit sector has grown the most rapidly. Thus the context in which public policy about child care will be made in the United States for the foreseeable future will in all likelihood include continued heavy reliance on a range of providers, rather than on a single delivery system. Most propos- als for reform of public policy about child care are based on the presump- tion that the bulk of child care in the United States will continue to be provided though this diverse delivery system, perhaps with greater subsidies from the public sector (e.g., Barnett, 1993; Helburn and Bergmann, 2002; Blau, 2001; Gormley, 1995; Walker, 1996; Lombardi, 2003). New combi- nations of financing child care are emerging, which build public supports into the child care that is delivered in a variety of child care settings (Vast, 1998; Stoney, 1998; Lombardi, 2003). While these new financing strate-
64 WORKING FAMILIES AND GROWING KIDS gies are promising, they are currently too small in scale to address the large amount of additional financing that may be needed to address problems in the child care market. Second, the major federal welfare reform of 1996 required or induced many low-income mothers to enter the labor market. Some of the reforms discussed during the debate over reauthorization of federal welfare legisla- tion in 2002 would impose even stricter work requirements for participa- tion in welfare. Child care during the mother's work hours is crucial for a successful transition from welfare to work. While many low-income moth- ers are able to find child care at no monetary cost from relatives and other sources, many other low-income mothers who do pay for child care spend a large fraction of their income on such care. The welfare reform of 1996 increased funding for child care subsidies, but the amount of child care subsidy funds available remains much lower than the amount that would be needed to serve the population eligible under the federal child care subsidy program. Thus an important implication is that funding for public child care subsidies may have to increase substantially if low-income families are to participate in the private market for child care in the new welfare envi- ronment. Chapter 9 discusses further the importance of expanding public subsidies for child care. Finally, the quality of child care is likely to have important conse- quences for the development of young children. A substantial amount of research on the impact of child care quality on development indicates that there is an effect, although the size of the effect is variable. The fact that the average quality of child care is mediocre in day care centers thus warrants concern. More generally, the lack of data on child care quality from a nationally representative sample of child care arrangements makes it diffi- cult to draw firm conclusions about the adequacy of child care quality in the United States, although indicators of structural aspects of quality raise serious concerns. Given the mediocre average for centers in large-scale but local studies, we are reasonably confident that a substantial portion of center care is in fact of minimal or inadequate quality. The child develop- ment literature does not provide clear guidance on the threshold, if any, below which child care quality becomes a serious risk to the development of children. But it seems safe to conclude that low quality is potentially an important concern in the child care market in the United States. Moreover, given the amount of time that children are spending in care, including a growing number of very young children, child care provides an important opportunity to promote their healthy development and overall well-being (Lombardi, 2003).