U.S. Department of Health and Human Services

Child Care Used by Working Women in the AFDC Population: An Analysis of the SIPP Data Base

Lorelei R. Brush, Ph.D.

Analysis, Research and Training

October 15, 1987

PDF Version

This report was prepared under contract between the U.S. Department of Health and Human Services (HHS), Office of Social Services Policy (now the Office of Disability, Aging and Long-Term Care Policy) and SysteMetrics/McGraw-Hill. For additional information about the study, you may visit the ASPE home page at http://aspe.hhs.gov. The ASPE Project Officer was Sharon McGroder.

ABSTRACT

This paper presents the Survey of Income and Program Participation (SIPP) data on child care for working guardians on AFDC. These guardians use care by relatives 58 percent of the time for their youngest children under 6. They are thus using informal care arrangements which are generally free or low cost instead of more formal arrangements, for which one generally must pay. As expected from this pattern of usage, the data show they pay less often for child care. In addition, when they do pay, they pay less money than non-AFDC recipients, averaging $22.50 per week. Because the sample of working guardians on AFDC is very similar to the SIPP's sample of non-working AFDC recipients, we may expect that these data can help work/welfare program planners in determining child care requirements for new initiatives encouraging AFDC recipients to work. Arguments are presented to suggest that about half of new program participants will choose relative care, and that the amount participants will need to cover their payments for care will be similar to current expenditures.

INTRODUCTION

The purposes of this paper are twofold. First, we present information from the Survey of Income and Program Participation (SIPP) data base about the child care arrangements used by working women in the AFDC population. Second, we discuss the implications of the information for planning initiatives to involve AFDC recipients with preschool aged children in programs combining work and the receipt of welfare payments. Women on AFDC with children under age 6 (in some States under 3) are currently exempt from requirements to work; new programs may modify this exemption and assist these women in their transition to employment. Work/welfare reforms are planned to serve women with young children who are enrolled in the AFDC program, help provide these women with work experience, further education and training for employment, and by necessity, will include provisions for child care.

The SIPP data can assist work/welfare program planners in predicting the forms of care to expect among their client population and the approximate amount of money that clients might expect to pay, if the child care market place continues to operate much like the present market. Because the SIPP supplies information only on current usage, it cannot answer questions about the kind of care that should be subsidized, nor will it predict the patterns of usage if client choice, licensing requirements, and rules and regulations about child care subsidies differ markedly from present State practices.

In determining the significance of SIPP information for planning purposes, it is important to think about the differences between the AFDC mothers in the sample and those who will be affected by the new work/welfare initiatives at some future time. First, working AFDC mothers represented in this sample were probably voluntarily choosing to work since they were not required to by law. This might not be the case under future work/welfare proposals. And we do not know how child care usage patterns differ for recipients working voluntarily and those required to work. Second, one must assume that the market place found in early 1985 and represented by consumer preference/choice, child care supply, agency regulations about type of care subsidized, and frequency and amount of government subsidy will be similar to the market place which will operate under a new work/welfare statute. Third, welfare reform initiatives may change the requirement for who must participate in school, training, or work and, therefore, who will need child care. New initiatives may or may not substantially change the process by which welfare agencies subsidize child care or the relative amount of money available to subsidize care for the increased number of children. These factors may cause future child care usage patterns to differ from present usage. Readers of this analysis will have to decide for themselves how well the present will predict the future.

THE SIPP DATA BASE

The SIPP is sponsored by the Federal Government in order to gather longitudinal data on social and demographic characteristics, labor force participation, earnings and income, and participation in Federal programs. Each sample or "panel" is selected to represent the noninstitutional population of the United States. Representative households participate in the study over a period of approximately 2½ years, being interviewed every four months.

This report concerns the fifth set of interviews of the 1984 Panel which began its interviews in the fall of 1983 and ended in the summer of 1986. This wave of interviews contained a special topical module on child care which asked working guardians about the care arrangements for their youngest three children under the age of 15. Its interviews were conducted in January through April of 1985; for each household, the child care questions concerned the month immediately preceding the interview. Thus, the information on child care reflects care in the winter months of December 1984 through March 1985.

The child care questions were asked only of guardians who were employed, either part-time or full-time during the month preceding the survey. Of the total of 606 AFDC recipients who are members of the data base as a whole, only 95 (15%) are working guardians. Of these, only 50 have Preschool-aged children, the group of interest to planners of the work/welfare programs. Hence, while these 50 AFDC recipients represent a weighted sample size of 250,104 people, the small actual sample size means that tests for significance are suspect and that it s difficult to subdivide the group for me analyses. Conclusions draw from the tables must be qualified. They may not reliably predict the behavior of the overall group of working AFDC recipients with preschool children. However, they are the only recent national data on child care utilization patterns and are worth reporting, even with these caveats.

People familiar with AFDC data may find this percentage of working AFDC recipients relatively large. Over 15 percent of the SIPP AFDC recipients report that they work outside the home. AFDC statistics suggest that only about 6 percent of recipients are employed. The larger percentage on SIPP may be due to a different purpose for its data gathering. SIPP counts anyone who earns money outside the home as "working"; it does not inquire about how much you declare for tax purposes or whether you declare an income when you talk with an AFDC counselor. Its answers are anonymous, so individuals may well be reporting work arrangements on SIPP that they do not report to the IRS or AFDC office.

KIND OF CARE USED BY WORKING AFDC RECIPIENTS

AFDC recipients are members of lower income groups and do not have much money available to pay for child care. Therefore, one would expect that they would use the kinds of care that are less often paid for, such as care by relatives, rather than care by nonrelatives in either a family day care home or a day care center. (Note that these hypotheses simply discuss usage patterns, they are not meant to predict unconstrained preferences of recipients nor to suggest that one kind of care is of a higher quality than another.)

The data seem to support these hypotheses. Table 1 displays the patterns of use of child care by three groups of SIPP families: those who are now receiving AFDC; those who were at one time on AFDC, but are no longer receiving benefits; and those who have never been on AFDC. Only data concerning the youngest child under the age of 6 are shown since this is the population of greatest concern to those planning welfare reform initiatives. It is already mandatory for families with older children to participate in such programs.

There are several sizeable differences in usage patterns among these groups. Specifically, current AFDC recipients are using much more in-home care by grandmothers and other relatives and less care by staff in centers or preschoools than non-recipients. They are also using less care by members of the child's own family (basically, the parents).

These results are intuitively reasonable. In terms of care by parents, we know that relatively fewer fathers are available as caregivers to AFDC mothers than to non-AFDC mothers since AFDC recipients are predominantly single female heads of households. The less frequent use of center care or preschools may be because AFDC mothers have more trouble paying the fees generally required for center care, unless it is subsidized by the welfare agency. On the other hand, many AFDC recipients do have parents or other relatives who can supply them with child care at no or modest cost. They seem to use the care that is available and affordable.

There are a number of factors that differentiate AFDC mothers from other mothers. These factors may contribute to differences in the kind of care used. We expect, for example, that AFDC mothers work part-time (under 35 hours) more than other mothers. Many work under 20 hours a week. Previous analyses of child care data show that usage patterns for part-time and full-time working guardians differ (Brush, 1987): part-time workers make more use of informal family care -- father care for all ages of preschool children and grandmother care for infants; full-time workers more often use nonrelative and center care. AFDC mothers' use of care by relatives may be at least partially related to a higher percentage of part-time employment among these mothers.

Table 2 substantiates this expectation. It shows the percentages of the sample who work part-time (under 20 hours a week), part-time from 20 to 34 hours, and full-time or over 34 hours. Working guardians in the AFDC population are significantly more likely to work part-time (under 35 hours) than full-time, and they are more frequently working part-time than members of the group once on AFDC or the group who have never been on AFDC (X²=28.33, df=4, p.001). So, one would expect their usage patterns to reflect those of part-time workers, specifically, that they would use more care by relatives and less care by family day care providers and by centers.

To see if the AFDC population differs from others in choice of care, controlling for the hours the guardian works, we have constructed the three parts of Table 3. Part A looks explicitly at guardians who work under 20 hours a week; Part B shows the results for guardians working 20 to 34 hours per week; Part C displays data for those working over 34 hours. The data are very similar across all three tables: AFDC recipients use more in-home care by grandmother or other relative than do non-AFDC recipients; they use less care by members of the child's nuclear family and by centers than the comparison groups.

Specifically, data on Table 3, Part A show that even among the group working under 20 hours a week, there are differences between those now on AFDC and those never on AFDC. AFDC recipients are much more likely to use care by relatives (not the child's parents) than non-AFDC recipients. That is, they use much more grandmother care (31% versus 9%) and care by other relatives (18% versus 5%). They are also different from those not receiving AFDC in their lower use of care by parents (7% versus 48%).

The second part of Table 3 (Part B) displays the data for guardians who work part-time, but a fairly substantial number of hours each week (20 to 34 hours). The differences between the AFDC population and the non-AFDC population are very similar to those for workers employed under 20 hours per week. Grandmother care and care by other relatives is more frequent among AFDC recipients; care by parents is somewhat less frequent. On this table center care also appears somewhat less frequently among AFDC recipients.

Part C suggests that the results for care by other relatives hold even among full-time workers, though here the differences in care by parents or grandmother are less extreme. Parent care is still somewhat less frequent among AFDC recipients (12% versus 24% for those no longer on AFDC or 14% for those never on AFDC). Grandmother care is somewhat more frequent (18% versus 6% versus 15%). Care in centers is also somewhat less frequent than for the comparison groups (29% versus 41% or 37%). Therefore, we conclude that the use of care by relatives is not explained simply by the shorter work schedules of AFDC compared to non-AFDC working guardians. Regardless of their hours of work, AFDC recipients use more care by grandmothers and other relatives than do non-AFDC guardians.

The population of AFDC working guardians differs from those not on AFDC on a number of other variables that suggest that this population would use more grandmother care and less care by fathers, centers, and nonrelatives (see Brush, 1987 for details). The AFDC group contains a higher percentage of guardians who are:

• single or not currently living with a spouse (X²=283.16, df=4, p.001) ;
• under the age of 25 (X²=38.20, df=4, p.001);
• Black (X²=76.89, df=4, p.001);
• working in service occupations X²=71.60, df=8, p.001); and
• who have a family income under $12,000 per year (X²=322.10, df=6, p.001).

All other things being equal, members of these groups (except those in service occupations) were found to use more grandmother care than nonmembers, so it is not surprising that AFDC recipients also use more grandmother care. Single mothers use less father care; so do AFDC recipients. However, the patterns of results for center and nonrelative care are not as consistent. Black guardians use center care somewhat more frequently; guardians in service occupations and those in lower income families use center care somewhat less. Single parents use nonrelative care more frequently; lower income families use it less. The lower income status of the AFDC population may be more important in influencing type of care used than race and occupational group, resulting in the less frequent use of care by nonrelatives and centers among AFDC families.

Thus, whether one looks at differences in choice of child care by a family's AFDC status or by the specific descriptors of marital status, age, race or occupation of guardian, and family income, the results are similar. The data show that care by relatives other than the child's parents is a very frequent phenomenon among the AFDC population. All relatives (father, mother sibling, grandmother, and other relatives) care for 58 percent of the youngest children under 6 compared to 44 percent for guardians never on AFDC (see Table 1).

For planners of work/welfare programs, this number may be meaningful in predicting the percentage of children needing more formal market care. However, data from the earlier Brush paper suggest that it may be an overestimate for the following reasons:

• Young mothers (those under age 25) are over-represented in the AFDC group. They may well have additional children in the future. As the number of children in a family increases, data from the population of all working guardians suggests that relatives are less likely to be used as caregivers. Children may then be cared for by nonrelatives in a day care home or center.

• The earlier paper also suggested that care by relatives is more frequently used with infants and toddlers, but when children reach the ages of 3 and 4, many parents opt for center care, if they have a choice.

• The SIPP data on child care only discuss usage, not unconstrained choice. It may be that some mothers who are using relatives would prefer, for example, a family day care home for their toddlers or a center for their older preschoolers. If a work/welfare program provides more choice, AFDC recipients may choose relatives less often than presently indicated.

• Furthermore, although 58 percent of AFDC recipients use care by relatives and, in general, relatives are paid for care less often than nonrelatives and centers, we cannot assume that all 58 percent are unpaid or that relatives do not require payment. Unfortunately, the SIPP data base does not allow the relating of kind of care and cost of care, so we cannot specify how many relatives are paid or how much they are paid.

• Children aged 6 and above may need to be provided with care if the AFDC mother works full-time or works a shift outside of school hours. Such care is not included in the tables shown above.

Thus, we can probably rely on a frequent usage of relatives as caregivers by working AFDC recipients, though the prediction of percentage of families using relatives would perhaps be more realistic if it were viewed as "about half". If families are offered greater choice in child care, we should expect an increased use of centers for older preschoolers. If larger families are included, we may find fewer relatives used as caregivers and more use of family day care homes. (Such homes may take two or three children from the same family, allowing the mother to simplify child care arrangements. Her school-aged children could come to the home after school hours and join the younger children.)

In the next section we discuss the SIPP data on who pays for care, trying to clarify the degree to which the AFDC group must spend their own resources on child care. Please note that the information on payment only concerns the money that the guardian has paid for care and does not include any subsidies.

PAYMENT FOR CHILD CARE

Table 4 summarizes information on who pays for care. It suggests that about half of the families in this country with children under 6 pay for their child care¹ and that there are significant differences in the percentage who pay across the variable of AFDC status (X²=7.14, df=2, p.05). Those currently on AFDC pay cash for care 40 percent of the time; those who were once on AFDC, but are not currently, pay 49 percent of the time; those who have never been on AFDC pay for care 53 percent of the time.

Data from the earlier Brush paper reinforce these findings. That is, several variables which were shown in that paper to predict who pays for care are also variables which differentiate AFDC recipients from other working guardians. For example, mothers who work part-time, those in service occupations and those with lower incomes pay for care less often than the comparison groups. AFDC recipients frequently appear in these categories. So, from these individual descriptors of working guardians as well as from the comparison of AFDC recipients versus non-recipients, we see that the population of interest is less likely to pay for child care.

In addition to differences in whether or not care is paid for, there are significant differences in the median amount of money paid for care. The median for those currently on AFDC is $22.50 per week, but it is $40 per week for both other groups. The highest amount paid by any family now receiving AFDC assistance is $45, while 25 percent of families not on AFDC pay $50 or more a week. So, a lower percentage of AFDC families pay for care and, when they do pay, they pay a smaller total amount than non-AFDC families.

The fact that they pay a smaller amount could well be because AFDC guardians work fewer hours than non-AFDC recipients and therefore pay for fewer hours of child care. Unfortunately, we cannot check on differences in the per hour cost of care because the SIPP does not connect data on care of an individual child with data on payment for care. Thus, if there is more than one child in a family, we do not know if care for all children is paid for or care for only one. If there are multiple kinds of care for an individual child, we do not know which care is paid for. So, we may assume that a part of the reason AFDC recipients pay less money for care is that they are working fewer hours, but we cannot analyze the exact difference in cost per hour of child care.

A second reason that AFDC recipients pay less money for care may be that their child care is subsidized. SIPP data only reflect the amount of money parents pay, not the total cost of their child care or the amount of any child care subsidy.

These caveats on costs per hour of care do not preclude making inferences helpful to work/welfare planners. These numbers may still assist in predicting families' requests for 18 money for care, if we make two assumptions: that newly employed AFDC guardians will reflect about the same percentage of part-time employees as the AFDC recipients represented currently in SIPP; and that about the same amount of subsidized care will be available to newly employed families. Assuming that about 40 percent of new program participants will pay for care at an average of $22.50 per week, an overall average of $9 per week per recipient seems to be a reasonable estimate.

USING THE WORKING AFDC POPULATION TO PREDICT CARE FOR THE NON-WORKING AFDC POPULATION

The conclusions presented above on the approximate percentage of work/welfare program recipients who would use care by relatives and who would pay for care are predicated on the assumption that the group of AFDC recipients currently employed is very similar to the group that would be working under new work/welfare programs, and, as stated earlier, that the rules of welfare and the market place will not change between the time this sample was taken and the time work/welfare initiatives take place. Such similarity would allow us to assume that the child care choices of the non-working AFDC group, when they are required to work, might have the same pattern as currently exists for AFDC recipients who are now working. To test that hypothesis, we have done two further analyses of the SIPP data. First, we have compared all working and non-working AFDC recipients on available descriptive variables to see how they differ on factors which might influence choice of child care or payment for care. Second, we have examined the members of the households of non-working AFDC recipients to see if there are other adults available to serve as child care providers.

Table 5 shows the comparison of working and non-working AFDC recipients in sipp. Using X² tests of unweighted data, we found significant differences on only three of the descriptors: level of education of guardian (X²=17.01, df=4, p.001), family income (X²=10.88, df=3, p.05), and living in a metropolitan area (X²=4.95, df=l, p.05). Working AFDC recipients are more frequently in the higher education categories, have higher family incomes, and are less often living in metropolitan areas than non-working recipients.

The earlier paper suggested that there were some differences in choice of care depending on the variables of education, income and city size. Mothers with advanced degrees tend to use less father care than mothers with high school diplomas; families in the higher income groups use less grandmother care and more care by nonrelatives and in centers than families in lower income groups; families in metropolitan areas used more center care than families outside of cities. However, these differences for working guardians in general may not be relevant to the comparison of working and non-working AFDC mothers. The difference among AFDC recipients in guardian education is largely due to the difference in those who ended their educations with grade school. The difference is not in the percentage with advanced degrees. So, the finding about differences in level of education among AFDC recipients does not imply that their choices of care will differ.

Similarly, the differences in the two groups on income are not sufficiently extreme to suggest differences in child care usage. The working AFDC group has fewer members in the lowest income category than expected, and a few more in each of the higher income categories. But, in general, they are not a "high income" group and cannot be said to use child care in the same way that the general population with high incomes might elect care.

On the other hand, the significant difference between working and non-working AFDC recipients with regard to residence in cities may well mean that non-working recipients will use a somewhat different pattern of child care when they are employed. The SIPP analysis showed that more city residents use center care; the analysis of working AFDC recipients showed that relatively few use center care. A work/welfare population concentrated in metropolitan areas might be expected to use more center care than is true of the present set of AFDC working guardians.

So the first analysis of the appropriateness of making conclusions for work/welfare programs from the working AFDC population suggests that the groups differ in only minor ways, and that observations from the working AFDC population may be useful in predicting the behavior of the nonworking population, when they are gainfully employed. The only difference in kind of child care elected by the AFDC group currently not working may be an increased choice of center care due to the higher concentration of this group in cities where center care may be available.

The addressing of one caveat is in order here. The variables which are measured in SIPP and considered in these analyses describe observable characteristics of working guardians and their families. They do not include measures of such potentially relevant variables as motivation to work. We do not know if there are critical differences in motivation which might be related to choice of care. Are women who choose to work more likely to recruit relatives as caregivers whereas women who are required to work by a work/welfare program are more likely to use family day care homes or centers? The data cannot answer these questions.

The second analysis of the working and non-working AFDC recipients sought to describe the family members who might be available to care for children, should the recipient become employed. Though we do not have records of relatives living outside the household, we do know about other adults in the household. So, we searched the records of all household members to locate any individual over 18 years of age who was not in school or training and not currently employed to identify adults who might provide child care.

Table 6 shows the results. There are a substantial number of recipients for whom another adult is available for child care. A total of 29 percent have some other adult in the household (a spouse, relative or nonrelative) who is not in school or employed. This 29 percent can be further specified by gender: 24 percent of non-working recipients have at least one other female adult in the household; 5 percent have at least one other adult and all are male. However, it would appear that about one of every four non-working AFDC recipients has at least one other adult potentially available for child care. This is approximately the same percentage as in those AFDC households where the guardian is working. Therefore, it would appear that those who are presently working are not doing so solely because they have greater availability of relatives in their households.

Two sets of conclusions seem to follow from these data. First, it would appear that the working AFDC population is not substantially different from the non-working population, at least on the measures reported here from SIPP. So there is support for concluding that the usage patterns of child care for those currently non-working would follow the patterns of those now working. Second, there is considerable support for the notion that relatives are available to the non-working AFDC group to serve as child care providers, should these AFDC recipients go to work. At least, relatives are available to the same extent as is true for working recipients. When non-working recipients do begin work, they may not use as high a percentage of relative care as those recipients who are currently working because those presently working may receive subsidized care somewhat less frequently. Recipients now voluntarily working must search out free care where ever it is available, and probably use more care by relatives than they would if they had more options. So, one might expect that non-working AFDC recipients, when employed through a work/welfare program, would use care by relatives a substantial percent of the time, but possibly less than the 58 percent quoted for recipients currently employed. For costing purposes, working estimates of the percent of newly employed recipients who will use care by relatives can vary, depending on the degree of conservatism preferred.

POLICY IMPLICATIONS

The SIPP data has provided some useful information for work/welfare planners, although it must be accepted with caution because of the small sample of working AFDC recipients using child care. Under the assumptions that child care provided for children under 6 will be the most crucial support service issue, the following comments may be made:

1. The kinds of child care elected by current non-working AFDC recipients could well follow the pattern set by working AFDC recipients. Care by relatives will predominate, used by about one-half of participants.

2. For those new program participants who will pay for care, the amount paid will be modest, probably averaging about $25 per week.

There are, of course, policy decisions which could modify the present pattern of child care usage. For example, if policy makers wished to set standards and/or promote certain types of child care, they might require that all subsidized care for AFDC children with working mothers be licensed. In this instance, it is likely that the percentage of care by relatives would decrease dramatically while center care and nonrelative care would increase.

The SIPP data concern child care currently used by families, not necessarily the care that would be chosen, if increased or more frequent subsidies for child care were provided. If families enter a work/welfare program, they may well choose more 26 center care for older pre-schoolers (and even for elementary school children after the end of the school day). In both of these cases (a requirement for licensed care and increased subsidy), the expectation would be that the unit cost of care would increase. If relatively more funds are made available to AFDC families under a work/welfare initiative than were available to this sample in 1985, then we might expect more use of center care.

NOTES

1. In comparison, only 30 percent of working guardians with children of all ages in the SIPP population pay for care (Brush, 1987), a number considerably less than the percentage cited here. The major reason is that the group of households shown on Table 4 contains only those families with children under 6. These families pay for care much more often than families with older children.

BIBLIOGRAPHY

Brush, Lorelei R. Usage Of Different Kinds of Child Care: An Analysis of the SIPP Data Base. Paper prepared for the Office of the Assistant Secretary for Planning and Evaluation, October 14, 1987. [Executive Summary]