You are on page 1of 16

ALESHA DURFEE Arizona State University

MARCIA K. MEYERS University of Washington*

Who Gets What From Government? Distributional


Consequences of Child-Care Assistance Policies

Given the fragmented structure of child-care spends on child-care assistance, but it has been
assistance in the United States, it has been diffi- difficult to track the amount and distribution of
cult to obtain accurate estimates of which fami- assistance through these separate and overlap-
lies are assisted, through which mechanisms, ping programs. In the absence of knowledge
and at what level. Making use of survey data about which families are assisted, and the amount
from New York City, we analyze the distribution of assistance they receive through various pro-
of several forms of public child-care assistance. grams, it has been nearly impossible to evaluate
Results suggest that about 40% of all families the adequacy of public child-care assistance.
with young children receive some form of In this article, we use survey data from a ran-
child-care assistance. Considering all forms of dom sample of families in New York City, along
assistance, the distribution of child-care help is with administrative data on public child-care
targeted in both expected and some unexpected expenditures, to describe the total package of child-
ways. Implications of these results are dis- care assistance received by families and to esti-
cussed in the context of U.S. child-care policies mate the level of assistance received by families
governing access and benefit levels. who vary in their child, parent, and family charac-
teristics. We find that a large share of all families
with young children receive some assistance
The use of nonparental child care has become
through tax credits, subsidies, or public preschool
increasingly common for young children in
programs. Although the level of child-care assis-
America. Federal and state governments help
tance is progressive by income, the distribution of
some families with the cost of these services
any assistance is not, resulting in a system that is
through several programs, each of which target
only moderately progressive overall and targets
different populations, use different mechanisms, help in both intended and unintended ways.
and provide different levels of assistance. We
know something about what the United States
BACKGROUND
About two thirds of all children younger than
Women and Gender Studies Program, Arizona State
5 years (Smith, 2002), and three quarters of chil-
University, P.O. Box 873404, Tempe, AZ 85287-3404 dren younger than 5 years with an employed
(alesha.durfee@asu.edu). mother (Capizzano & Adams, 2004), are in some
*School of Social Work and Evans School of Public Affairs, form of nonparental child care on a regular basis.
University of Washington, 4101 15th Avenue NE, Seattle, Whether they choose to use informal or formal
WA 98105. modes of care for their children, most families
Key Words: child care, family policy, low-income families, pay for the care provided. Nearly all parents
maternal employment. who use child-care centers (other than public
Journal of Marriage and Family 68 (August 2006): 733–748 733
734 Journal of Marriage and Family

programs), family child care, or in-home care- Center for Children in Poverty, 2004) and federal
givers pay for those services, and an estimated tax expenditures totaled nearly $3 billion in 2000
33% of those who use relative caregivers also (Campbell & Parisi, 2003). Because they are non-
pay for that care (Hofferth, Shauman, & Henke, refundable, however, the federal credit and about
1998). Considering all types of care, data from one half of state tax credits are completely
the National Survey of America’s Families indi- unavailable to those whose incomes are too low
cate that 48% of families with an employed to incur tax liabilities. For families who can claim
mother and at least one child younger than the benefits, tax credits are modest. The credit
14 years make cash payments for child care received through the federal credit averaged
(Giannarelli, Adelman, & Schmidt, 2003). $440 per claimant as of 2001 (National Center
These expenses are substantial for many fami- for Children in Poverty).
lies. Child care consumes nearly 8% of total fam- Means-tested subsidies for the purchase of pri-
ily income for families with an employed mother vate care provide a more substantial benefit to
and at least one child younger than 14 years a much more restricted population of low income
(National Research Council and Institute of Med- and employed families. The federal government
icine, 2003). The child-care burden is consider- currently funds means-tested subsidies through
ably higher for families with younger children, three block grants that are managed and supple-
averaging 10% of income among families with mented by states: the Child Care and Development
children younger than 3 years. Child-care ex- Fund Block Grant, the Temporary Assistance to
penses are also particularly steep relative to Needy Families Block Grant, and the Social
income for poor families. Families with incomes Services Block Grant. Federal spending for the
under $18,000, for example, spend nearly one three block grants approached $7 billion in
quarter of their income (23%) on child care. 2000, and states spent an additional $2.7 billion
(Gish, 2002). Unlike child-care tax credits,
appropriations for means-tested subsidies are
Child-Care Assistance Programs
capped and assistance is rationed by federal and
Federal and state governments provide selective state eligibility rules and application procedures.
and targeted assistance with child-care expenses In their study of child-care subsidies in 17 states,
through three main mechanisms: tax credits, Collins, Layzer, Kreader, Werner, and Glantz
means-tested subsidies, and public preschool (2000) estimate that as of 1999, 4 of the 17 states
programs. Government controls both the total served between 20% and 25% of children who
amount of assistance and its distribution at the were eligible for assistance under federal rules
family level through budgetary appropriations and 6 served 15% or fewer of these children.
and through policies that govern eligibility and For families who receive subsidies, benefits are
benefit levels. often substantial, covering the cost of care up to
Tax credits are available to the broadest popu- state-established maximum reimbursements
lation of child-care users, but they provide the minus family copayments. There are no reliable
most modest benefits. Employed parents with national estimates of average subsidy benefits.
taxable income and work-related child-care ex- A study of five states conducted by Meyers et al.
penses are eligible to claim the Federal Child (2006) estimates that median family benefits
Care and Development Tax Credit, which range between $190 and $363 per month. Sub-
deducts a portion of these expenses from their sidy assistance is also limited in duration by eligi-
taxable earnings. The Federal Child and Depen- bility rules, income thresholds, and progressive
dent Care Tax Credit is progressive, declining copayment schedules that limit or discourage
from 35% to 20% of allowable expenses for high- continuous receipt. Meyers and her colleagues
er income families. Twenty-six states have estimated that the median period of subsidy
adopted similar credits against state income receipt in the five states ranges from 3 to 7
taxes, and a small number of families use months.
employer-managed Dependent Care Assistance The third mechanism through which govern-
Plans to deduct child-care expenses from their ment assists with child-care expenses is the provi-
taxable income. The federal tax credit is claimed sion of direct services, most commonly through
by a large number of families: In 2001, an esti- early education programs. Public preschool
mated 6.2 million tax filers claimed expenses programs are typically free or of very low cost for
related to child or other dependent care (National parents but are limited in availability. Eligibility
Distribution of Child-Care Assistance 735

for most preschool programs is restricted to istics influence the receipt and amount of child-
4-year-old children and, outside of a small hand- care assistance. The size and characteristics of
ful of states, by policies that target services the populations eligible for various forms of
on disadvantaged families. The single largest assistance are determined by policies that limit
public program is the federal Head Start program availability (e.g., through capped appropriations)
for low-income 4-year-old (and some 3 year old) and that control access (e.g., through eligibility
children. Federal appropriations for Head Start and intake policies). No federal or state child-care
totaled nearly $5.3 billion in 2000 (Administration program serves all eligible families, however,
for Children, Youth and Families, 2002). Some and the distribution of assistance within eligible
states supplement Head Start funding, and 36 populations reflects individual and family factors
states also fund prekindergarten programs, usually that influence parents’ decision to use nonparen-
targeting children at economic or educational risk tal care, the type of care they use, their decision to
(Doherty, 2002). Children with disabilities and take up benefits, and their success in obtaining
other special needs may also receive public care them.
through early intervention (special education) pro- Nearly all studies of the family-level determi-
grams. Preschools provide a benefit that is compa- nants of the receipt of child-care assistance have
rable to the cost of private child care, averaging examined the receipt of means-tested assistance.
$3,455 per child enrolled in public prekindergar- As expected, studies find that the likelihood of
ten and $6,934 for children enrolled in Head Start receiving such assistance is greater among fami-
(Barnett, Robin, Hustedt, & Schulman, 2003). lies with income below the federal poverty line
Like subsidies, these benefits are rationed by and among those in which all parents are
eligibility rules and capped appropriations. As of employed (Burstein, Layzer, & Cahill, 2001;
2000 – 2001, the share of all 4-year-old chil- Giannarelli et al., 2003; Meyers et al., in press).
dren enrolled in some form of public early educa- Research suggests that the probability of subsidy
tion—including Head Start, prekindergarten receipt also varies with family characteristics that
programs, and special education programs— predict greater need for nonparental care because
was about 31%, with considerable variation across of employment and caregiving demands. Fami-
states (Barnett, Robin, Hustedt, & Schulman lies who have children younger than 6 years,
2003). those headed by single parents, and those with
mothers having high school degrees or some
college are more likely to receive child-care sub-
Who Gets Help?
sidies (Blau & Tekin, 2003; Georges & Wagmiller,
Given the fragmented structure of child-care assis- 2001; Giannarelli et al.; Meyers, Heintze, & Wolf,
tance in the United States, it has been difficult to 2002). Blau and Tekin (2003) and Georges and
know how these budgetary and program policies Wagmiller (2001) report higher probabilities of
interact to determine the level and distribution of subsidy receipt among families with greater
assistance at the family level. Separate federal nonwage income, which is consistent with other
and state programs collect enrollment and benefit findings that families with a prior history of wel-
data using different and noncomparable measures, fare receipt, and those in target populations for
and no national survey has collected comprehen- welfare-to-work programs, are more likely to
sive information on child-care assistance for over be assisted (Burstein et al., 2001; Giannarelli
a decade. The most recent national estimate of et al.; Meyers et al., 2002). Families who are
how many families receive assistance other than Black or Hispanic (Blau & Tekin) and those
tax credits is provided by Giannarelli et al. who are nonimmigrant (Burstein et al.; Georges
(2003), who use data from the National Survey & Wagmiller) are more likely to receive subsi-
of Families to estimate that about 12% of em- dies, which may reflect differences in parents’
ployed families with a child younger than 13 years preferences, their welfare experiences, or their
receive some type of nontax assistance from success in negotiating the subsidy system. The
government or other organizations. Unfortunately, effect of maternal age appears to be curvilinear,
this survey does not include data on tax bene- with the probability of assistance increasing with
fits or on the amount of assistance received by age but declining beyond the age of 36 (Blau &
families. Tekin; Georges & Wagmiller).
Given data limitations, research has also been Other, primarily qualitative evidence suggests
limited on the question of how family character- that the organization of public programs may
736 Journal of Marriage and Family

influence which eligible families receive assis- Estimating the relationship between family char-
tance. Adams and Snyder (2003) describe burden- acteristics and assistance is complicated by the
some application processes for means-tested targeted structure of child-care programs. The
subsidies in some parts of the country, which, cou- same family-level characteristics—such as chil-
pled with complex eligibility requirements, may dren’s ages and mothers’ employment status—are
prevent all but the most determined parents from likely to operate through ‘‘supply-side’’ program
getting and retaining subsidies for which they are dynamics (i.e., determining eligibility or priority
eligible. These findings are echoed in studies from for assistance) and ‘‘demand-side’’ behaviors
the parent perspective, which reveal that parents (i.e., influencing parents’ decision to pursue and
often have limited and inaccurate information success in taking up benefits) (Witte & Queralt,
about their eligibility for assistance and encounter 2002). Given the cross-sectional data from a sin-
substantial impediments in their interactions gle city used in this study, it is not possible to dis-
with subsidy offices (Chaudry, 2004; Meyers & tinguish between policy and behavior factors or
Heintze, 1999; Powell & Cahill, 2000). to separate the supply- and demand-side effects
of family characteristics and behavior. Our intent
Research Questions is to describe the current distribution of assistance
as it results from the interaction of policy, family
Our knowledge of who is getting assistance, and characteristics, and individual behavior.
how much they are receiving, remains limited in Conducting these analyses with data from
important respects. Recent national surveys pro- a single city had the advantage of holding the pol-
vide some information about child-care use but lit- icy environment constant across families. This al-
tle or no information about government assistance. lows us to examine variation across families who
Nearly all studies examining the family-level varia- differ by sociodemographic and other character-
tion in the receipt of child-care assistance have istics but who experience the same set of pro-
studied means-tested subsidies, providing little grams, eligibility, and benefit rules. The use of
or no information about the full package of assis- data from a single jurisdiction also allows us to
tance through subsidies, tax credits, and preschool use administrative data for detailed imputations
programs. Virtually, no research goes beyond esti- of the value of assistance.
mates of receipt to examine the value of assistance Conducting the analysis in a single city had the
received at the family level. Child-care assistance disadvantage of representing a single population
programs have grown in recent years, but we still and policy environment, potentially limiting gen-
know remarkably little about who is assisted eralizability to the United States as a whole. New
through these programs and what they receive. York City provides a particularly interesting case
In this article, we examine the distribution of study for examining child-care assistance in the
multiple forms of child-care assistance along United States. New York State is one of only
two dimensions: the probability of getting assis- six states that have authorized prekindergarten
tance and the level of assistance received. We services for all 4 year olds, although the delivery
use data from cross-sectional surveys of random of these services has lagged behind the promise
samples of New York City families with any chil- (Doherty, 2002). New York State is 1 of 26 states
dren younger than 6 years to answer the following that provides a child-care tax credit and 1 of 11 in
questions: which this credit is refundable (Children’s
1. How many and which families receive some Defense Fund, 2003). New York City provided
form of child-care assistance through tax an estimated $230 million in means-tested sub-
credits, means-tested subsidies, and public sidy assistance in fiscal year 1999 – 2000
preschools? (Fender, O’Brien, Thompson, Snyder, & Bess,
2. Among families receiving assistance, how 2002). Although this represents a substantial
much assistance do they receive? investment, the city has been unable to serve all
3. Considering both the probability of receipt eligible families. Priority is given to families cur-
and the value of assistance, how does total rently receiving public assistance, and the city
child-care assistance vary with child, parent, maintains a significant waiting list of low-income
and family characteristics? working families. By using data from New York
City, we can thus look across the full range of
This article provides descriptive, rather than child-care programs—tax credits, subsidies, and
causal, analyses of the distribution of assistance. early education programs—within a system that
Distribution of Child-Care Assistance 737

is similar to the United States as a whole in ration- methods, response rates, weighting procedures,
ing assistance in both early education and subsidy and detailed descriptions of imputation methods,
programs. see Garcia & Meyers, 2005.)
In terms of comparability to the United States as The New York Social Indicators Survey de-
a whole, analyses by the Urban Institute indicate fines family as a respondent, the respondent’s
families in New York State are about as likely as resident spouse or partner, and resident children
those nationwide to have a child younger than younger than 18 years. This differs from the
5 years in nonparental care and to use care arrange- definition of both family and household used by
ments that are similar on average (Sonenstein, the census as it includes unmarried domestic
Gates, Schmidt, & Bolshun, 2002). Relative to partners in the family unit but excludes other
other families, however, low-income families family members, as well as nonrelated adults
(under 200% of poverty) in New York are less and children, who may be living in the house-
likely to be using center-based care and more hold. For these analyses, we restrict our sample
likely to use relative care as a primary arrange- to families with at least one child between the
ment. The proportion of families with an em- ages of 0 and 5. Of 703 cases in this sample, we
ployed mother and a child younger than 13 excluded 58 (8%) because of missing data on
years who receive some form of nontax child-care mother’s employment (n ¼ 19, 3%), race (n ¼
assistance from government (or other organiza- 13, 2%), ethnicity (n ¼ 5, 1%), immigration sta-
tions) is higher in New York (16%) than nation- tus (n ¼ 8, 1%), or other demographic charac-
wide (13%), which is consistent with the teristics. The number of cases excluded because
availability of both subsidy and preschool assis- of item-missing data was small and was not ex-
tance (Giannarelli et al., 2003). It may also reflect pected to bias our estimates. The final analysis
higher levels of need in New York. Child-care sample included 646 families.
costs are higher in New York than they are nation-
ally; in 1999, families in New York State who
Measures
paid for child care spent an average of $348 per
month, more than the average of $303 spent by The New York Social Indicators Survey includes
families nationwide (Giannarelli et al.). The pro- unusually detailed questions about families’ use
portion of children living in poverty in New York and payment arrangements for child care in the
State is also higher than that in the rest of the coun- prior 12 months. Data on the use of child care,
try. Nearly one quarter (22%) of New York chil- and receipt of tax-based, subsidy, or public pre-
dren lived in poor families in 1998, compared to school assistance, are collected at the family level
18% of children nationwide (Fender et al., 2002). for all children younger than 13 years. This limits
our ability to analyze care arrangements at the
child level but provides quite comprehensive
METHOD
information about the total package of assistance
at the family level.
Data
All respondents with children younger than 13
To analyze the level and distribution of child-care years were asked about their use of several forms
assistance at the family level, we make use of of child-care and preschool programs in the prior
pooled data from Waves II (1999) and III year for all their children, and about direct pay-
(2002) of the New York Social Indicators Survey. ment and sources of assistance (if any) for each
The survey is designed to document individual type of care. Measures of the receipt of assistance
and family well-being across multiple domains; are constructed directly from survey responses.
it also measures the sources and extent of external Although survey respondents are expected to
supports from government, family and friends, provide accurate information on receipt, they
community and religious programs, and employ- are unlikely to know the value of government
ers. The survey is conducted every other year assistance that is often given directly to providers.
using computer-assisted telephone interview To estimate the value of assistance received, we
technology and random digit dialing to interview impute values (conditional on receipt) for public
a representative sample of approximately 1,500 preschool enrollment, child-care subsidies, and
families from the five boroughs of the city. (For child-care tax credits using both New York Social
additional information about the New York Indicators Survey and New York City adminis-
Social Indicators Survey, including sampling trative data.
738 Journal of Marriage and Family

For families reporting any use of public pre- ily characteristics with the probability of receiv-
schools, the value is imputed using government ing child-care assistance and the value of
cost methods; we divide total direct government assistance received. As many families receive
expenditures (for Head Start and public preschool zero assistance (because they do not use child care
programs) by the number of children in the family or they use care but receive no assistance), the dis-
who were enrolled, adjusting for both the number tribution of the dollar value of assistance is highly
of months and whether they attended full- or part- skewed and precludes the use of ordinary least
day programs. As we do not include public expen- squares (OLS) regression with the full sample of
ditures on early intervention programs for children recipient and nonrecipient families. Instead, we
with special needs, this imputation method may derive our estimates in two steps. We first use
underestimate the per child value of public pre- logistic regression to estimate the probability of
school services. For families reporting the receipt receiving any assistance and the probability of
of means-tested subsidies, the value of these subsi- receiving each of three forms of assistance (tax
dies is calculated using data from the New York credits, subsidies, and public preschool). We next
State Market Rate Survey, which contains infor- estimate the value of assistance conditional on
mation on the average private sector child-care pri- receipt, using OLS regression with the subsets
ces in each region of the state and is used to set of families reporting any assistance. In the second
child-care subsidy reimbursements. The total stage of the analysis, we use these coefficients to
value of subsidized care is imputed for each family, estimate the total assistance received by families
adjusting for factors that determine reimbursement who vary by selected sociodemographic charac-
rates to private providers: the number and ages of teristics. We simulate the value of the total
children in subsidized care, the type of care (e.g., child-care package by multiplying the estimated
center or relative care), and the duration of care probability of receiving any assistance by the esti-
(i.e., number of months and full day or part time). mated value of assistance, holding all other char-
For families who reported claiming child-care tax acteristics at the sample mean.
credits, the value of the federal and state Child There are two alternative estimation ap-
and Dependent Care Tax Credit and the New York proaches: a tobit or a two-stage (Heckman) model.
State Child Care Credit is simulated using federal We reject the first approach because it assumes that
and state tax rules in combination with income estimates will be biased if zero values represent
data from the New York Social Indicators Survey, censored observations with an unobserved non-
families’ reported child-care expenditures, and zero value; for our descriptive purposes, we
estimated tax liabilities calculated using the TAX- assume that zero values represent valid cases of
SIM model of the National Bureau of Economic no assistance rather than censored observations.
Research (Feenberg & Coutts, 1993). A two-stage approach would be expected to pro-
Additional measures of child and family char- duce more reliable estimates if unobserved charac-
acteristics are based on self-reported data from teristics of families that increase the probability of
the New York Social Indicators Survey, includ- getting assistance are correlated with the amount
ing presence of any infants (aged 0 – 2) or pre- of assistance received. This might be true, for
schoolers (aged 3 – 5), total number of children example, if families who expect to qualify for
younger than 6 years, disability/health status of higher subsidies are more likely to use child care
one randomly selected focal child, the number and to apply for and receive assistance. Because
of nonparental adults living in the household, the characteristics that determine families’ eligibil-
total market and nonmarket family income as ity are the same as those that determine benefit
a percentage of the federal poverty level, moth- amount, however, we are unable to find a satisfac-
ers’ current participation in paid employment, tory identification variable. In this regard, our esti-
education or training activities, and the respond- mates should be interpreted as describing the
ent’s (usually mother’s) marital status, race/eth- existing distribution of assistance within this pop-
nicity, nativity and citizenship status, education, ulation, given both policy and individual behav-
and disability/health status. iors, rather than as a causal model. For a more
complete discussion of a similar estimation with
Analysis medical expenses, see Duan, Manning, Morris,
and Newhouse (1983).
Our multivariate analysis proceeds in two stages. On the basis of prior research, we expect that
We first estimate the association of child and fam- families’ receipt of child-care assistance will vary
Distribution of Child-Care Assistance 739

with (a) characteristics affecting families’ needs in school, or in training activities. Approximately
and resources for nonparental care, including 41% of families with any children younger than
family structure and characteristics associated 6 years receive some form of governmental child-
with the probability of maternal employment; care assistance during a 1-year period. Child-care
(b) characteristics affecting eligibility for vari- tax benefits are by far the most common form of
ous forms of assistance, including child age, assistance, claimed by 25% of families. Approx-
maternal employment status, and family imately 18% of families have at least one child
income; and (c) characteristics affecting pa- enrolled in a public preschool program. Only
rents’ decision to seek assistance and success in 9% of families reported that they received help
obtaining it. Because these outcomes are highly (through subsidies, vouchers, or direct payment)
correlated, and sensitive to similar family chara- for the costs of private care in centers, family
cteristics, we choose to estimate reduced-form child care, or informal sitting arrangements.
models. We model the probability of receiving When we consider only families in which the
assistance and the conditional value of assis- mother is currently employed or in school or
tance as a function of child, parent, and family training, 50% of families report some form of
characteristics that are expected to influence assistance. Receipt of subsidies and preschool
child-care assistance directly, via eligibility and services are about as common among families
take-up of benefits, and indirectly through their with and without an employed mother, but the lat-
effect on maternal employment and use of non- ter are 10 percentage points more likely to have
parental care. In supplementary analyses, we claimed child-care tax benefits.
experimented with several interaction terms in As Table 1 suggests, families are likely to
the models but none improved either the fit of receive more than one form of assistance during
the model or the explanatory power of the the year. Among families receiving subsidies,
included terms. for example, 41% also have at least one child in
We include an indicator for maternal employ- public preschool and 42% claim child-care tax
ment, but, unlike many studies, we do not limit credits. Families receiving tax credits are less
our sample to families with an employed mother. likely to be receiving other types of help; 22%
This allows us to estimate the distribution of also have a child in public preschool, and 15%
assistance within the entire population, as well are receiving subsidies.
as to estimate the association of maternal employ- This rate of assistance is considerably higher
ment with this assistance. We also do not limit our than that reported by the Urban Institute for
sample to families using any form of nonparental New York State. When we estimate receipt using
care. The large majority of families in the sample a sample and measures that correspond more
report some use of nonparental care on a regular closely to those used by Giannarelli et al. (2003)—
basis (71%). We include all families in our esti- considering only families with an employed
mates because we are interested in the distribu- mother and measuring receipt of only means-
tion of child-care assistance across the entire tested subsidies or preschool services—we esti-
population of families with young children. For mate that 25% of families in New York City
our final simulations, we limit the sample to fam- were assisted in comparison to the 16% esti-
ilies with an employed mother as these are the mated by Giannarelli et al. This difference likely
families assumed to have the highest child-care results from more detailed questions in the New
needs and are the targets of most child-care poli- York Social Indicators Survey and to the collec-
cies. All point estimates are weighted using data tion of data about assistance received in the prior
from the 2000 census; regressions are estimated 12 months, in contrast to the 1-month measure
using unweighted data. used by Urban Institute analysts.
Table 2 reports the weighted demographic
characteristics for the full sample and for those
RESULTS
who do and do not receive any governmental
assistance. Results from the multivariate esti-
Probability of Getting Assistance
mates of receipt are reported in Table 3. Results
In Table 1, we report rates of child-care assistance are consistent with prior research for most family
receipt in the prior year (weighted) among all and child characteristics. Four results are particu-
families with children aged 0 – 5 and among fam- larly notable. First, families in which the mother
ilies in which the mother is currently employed, is currently employed (or in school or training)
740 Journal of Marriage and Family

Table 1. Rates of Assistance, New York City Families With Any Child Aged 0 – 5 Years (N ¼ 646)

Receive Any Have A Child Receive


Governmental Assistance Receive Governmental Enrolled at a Child-Care
With Child Care Child-Care Subsidies Public Preschool Tax Benefits

n M SE M SE M SE M SE

All families 646 .41 .02 .09 .01 .18 .02 .25 .02
Families receiving 290 — — .23 .03 .45 .03 .62 .03
governmental assistance
with child care
Families receiving 64 — — — — .41 .07 .42 .07
governmental child-care
subsidies
Families with a 124 — — .21 .04 — — .30 .05
child enrolled
in public preschool
Families receiving 185 — — .15 .04 .22 .04 — —
child-care tax benefits
Families with an 349 .50 .03 .09 .02 .20 .03 .35 .03
employed mother
Families with an employed
mother who
Receive some form of 186 — — .19 .03 .41 .04 .71 .04
governmental assistance
with child care
Receive governmental 36 — — — — .52 .09 .41 .09
child-care subsidies
Have a child enrolled 69 — — .25 .06 — — .42 .07
in public preschool
Receive child-care 136 — — .11 .03 .24 .05 — —
tax benefits
Note: All values are weighted.

are significantly more likely to receive assistance teristics have quite different associations with
with child-care expenses. Second, among fami- different forms of assistance. As would be ex-
lies with any children younger than 6 years, the pected, low-income families are significantly
probability of receiving any assistance is substan- more likely than higher income families to
tially greater if a family has any children aged 3 – receive means-tested subsidies. Current maternal
5 in contrast to families in which all the children employment status has a positive but nonsignifi-
are younger than 3 years. Third, probabilities are cant association with subsidy receipt. This is sur-
not significantly different between naturalized prising given that eligibility for many subsidies is
immigrants and native-born parents (the contingent on participation in employment or
excluded category) but are significantly and sub- employment preparation activities. This may
stantially lower for noncitizen immigrants. reflect limitations in the survey measures of
Finally, and most surprisingly, even with controls employment, such as failing to capture prior
for other family characteristics, lower income employment during the year. It may also reflect
families are not significantly more likely to get New York City policy that gives priority for sub-
assistance than those with income greater than sidies to families receiving public assistance who
300% of poverty (the excluded group); in fact, may not report their current status as employed or
the coefficients for the low-income indicators in school or training activities. As with any assis-
are negative. tance, families headed by noncitizen immigrants
When we examine particular forms of assis- are significantly less likely to be receiving subsi-
tance, we see that family and household charac- dies than are those headed by native-born parents.
Distribution of Child-Care Assistance 741

Table 2. Sample Characteristics, New York City Families With Any Child Aged 0 – 5 Years

Families Receiving Families Not


Full Sample Assistance Receiving Assistance
(N ¼ 646) (n ¼ 290) (n ¼ 356)

Variables M SE M SE M SE t test

Any infants (age 0 – 2) .55 .02 .45 .03 .61 .03 **


Any preschoolers (age 3 – 5) .62 .02 .75 .03 .53 .03 **
Number of children aged 0 – 5 1.10 .02 1.13 .03 1.09 .02
a
Child in poor health .07 .02 .08 .02 .06 .02
Respondent is single .33 .02 .40 .03 .27 .03 **
Number of nonparental adults residing in household 1.36 .07 1.13 .09 1.51 .09 **
b
Respondent is U.S. born .49 .02 .52 .03 .47 .03
Respondent is naturalized immigrant .21 .02 .26 .03 .19 .03
Respondent is noncitizen immigrant .30 .02 .22 .03 .35 .03 **
Respondent is Hispanic .38 .02 .37 .03 .40 .03
Respondent is Black .26 .02 .36 .03 .20 .02
Respondent’s education , high school degree .21 .02 .16 .03 .23 .03
Respondent’s education ¼ high school degree .35 .02 .36 .03 .35 .03
Respondent’s education  some college .43 .02 .48 .03 .43 .03
c
Respondent in poor health .10 .01 .05 .01 .14 .02 **
Family income  75% of federal poverty level .24 .02 .24 .03 .23 .03
Family income 76% – 200% of federal poverty level .32 .02 .29 .03 .34 .03
Family income 201% – 300% of federal poverty level .23 .02 .24 .03 .23 .03
Family income . 300% of federal poverty level .21 .02 .23 .03 .20 .02
d
Mother is employed .49 .02 .60 .03 .42 .03 **
Survey conducted in Wave II (1999) .51 .02 .37 .03 .60 .03 **
Note: All values are weighted.
a b c
Child is in poor health or has chronic health condition. Includes Puerto Rico. Respondent is in poor health or has chronic
d
health condition that limits employment. Mother is employed, in school, or in job training program.
*p , .05. **p , .01. ***p , .001.

The probability of having a child enrolled in income. Those families with income levels at or
public preschool has the expected, strong associ- below 200% of poverty are much less likely to
ation with having at least one child in the categor- claim child-care tax credits than more affluent
ically eligible age group. The coefficient for families. Holding family income constant, fami-
maternal employment is small and nonsignifi- lies with employed mothers are more likely to
cant, which is consistent with the availability of claim child-care tax credits than are families
preschool for both employed and nonemployed where the mother is not employed, in school, or
families. The probability of public preschool in job training. As with other forms of assistance,
enrollment is sharply higher for families in which noncitizen immigrants are less likely to be getting
the (focal) child has a chronic health or disabling help. In contrast to results for subsidies and pre-
condition, likely because of enrollment in early school, the probability of claiming tax credits is
intervention programs. Noncitizen immigrant also lower for respondents with limited education
families are not significantly less likely to have and disabilities or health problems.
children in public preschools than are native-born
families, though the coefficient is large and neg-
The Value of Assistance
ative. Indicators for low family income (in com-
parison to the excluded high-income group) are Table 4 reports the average annual value of assis-
positive but not significant. tance received (weighted), for all families and for
Like subsidies, the probability that families families receiving assistance. Among all fami-
receive tax credits is strongly associated with lies, child-care assistance averages $1,617;
742 Journal of Marriage and Family

Table 3. Logistic Regression Results, Receipt of Assistance for New York City Families With Any
Child Aged 0 – 5 Years (N ¼ 646)

Child
Receives Enrolled Receives
Receives Any Child-Care at Public Tax
Assistance Subsidies Preschool Benefits
B B B B
Predictor B SE B e B SE B e B SE B e B SE B e

Any infants (age 0 – 2) .09 .28 1.10 .93 .49 .39 .08 .30 .92 .13 .32 .88
Any preschoolers (age 3 – 5) .93** .29 2.54 .33 .53 .72 3.20** .57 24.64 .09 .32 1.09
Number of children aged 0 – 5 .13 .27 1.13 .33 .41 1.40 .08* .29 1.08 .27 .32 1.31
Child in poor health .66 .36 1.93 .30 .50 1.35 1.02** .39 2.79 .34 .46 .71
Respondent is single .31 .23 1.36 .06 .34 .94 .09 .28 1.10 .53* .25 1.69
Number of nonparental .29* .13 .74 .32 .21 .73 .04 .15 .96 .23 .15 .79
adults in household
Respondent is naturalized immigrant .09 .23 1.10 .17 .36 .84 .59 .31 .55 .28 .24 1.33
Respondent is noncitizen immigrant .56* .24 .57 .88* .39 .41 .32 .29 .72 .63 .28 .53
Respondent is Hispanic .73** .24 2.08 .27 .43 1.31 .65* .33 1.91 .74** .28 2.09
Respondent is Black 1.06** .26 2.90 .49 .44 1.63 .68* .35 1.97 .95** .28 2.59
Respondent’s .14 .30 .87 .04 .40 1.04 .30 .34 1.35 .94* .41 .39
education , high school degree
Respondent’s education ¼ some .09 .22 .91 .12 .34 .88 .04 .27 .96 .13 .24 1.14
college or more
Respondent is in poor health .97** .34 .38 .05 .45 1.05 .66 .45 .52 1.27** .47 .28
Family income , 75% .40 .33 .67 2.17** .59 8.73 .57 .42 1.76 1.06** .38 .35
of poverty level
Family income 75% – 200% .45 .27 .64 1.74** .54 5.69 .40 .36 1.50 .64* .29 .53
of poverty level
Family income 201% – 300% .11 .26 .90 .84 .57 2.32 .48 .35 1.62 .30 .27 .74
of poverty level
Mother is employed .44* .20 1.55 .31 .31 1.36 .08 .25 1.09 .65** .22 1.91
Survey is from Wave II .92** .18 .40 .86** .30 .42 .48** .23 .62 .94** .21 .39
Constant .72 2.76 4.64 1.07
2
v 131.29 55.12 139.78 135.19
df 18 18 18 18
% receiving 44.9 9.9 19.2 28.6
Note: Analyses are unweighted.
*p , .05. **p , .01. ***p , .001.

among families receiving any form of assistance, ceived, conditional on the receipt of any assis-
the average total benefit is a more substantial tance. We use the total value of the assistance
$3,979. The dollar value of benefits is greatest package, rather than estimating the values for
for preschool programs, followed by subsidies, specific programs, because the sample of recipi-
and it is much lower for tax credits. These esti- ent families is too small for reliable estimates. A
mates confound the value of assistance with the separate analysis (not shown) comparing the
probability of receipt, however. When examined two approaches indicates that they produce simi-
only for families receiving each type of benefit, lar substantive results.
2
the value of subsidies is the highest at $7,103, fol- The adjusted R suggests that the model ex-
lowed by preschools at $6,417 and tax benefits at plains 15% of the variation in the value of assis-
$851 per recipient family. tance. Relatively few individual coefficients
In Table 5, we report the results of OLS achieve significance, however. This is likely
regressions for the total amount of assistance re- because of the imputation of values using only
Distribution of Child-Care Assistance 743

Table 4. Average Value ($) of Assistance Received, All Assistance, and Conditional on Receipt

Receiving Any Child


Full Receiving Child-Care Enrolled in Receiving
Sample Any Assistance Subsidies Public Child-Care
(N ¼ 646) (n ¼ 290) (n ¼ 64) Preschool (n ¼ 124) Tax Credits (n ¼ 185)

Variable M Mdn SE M Mdn SE M Mdn SE M Mdn SE M Mdn SE

Total amount of 1617 0 149 3979 3079 283 9283 8122 674 5739 3753 441 2845 1152 360
government
assistance with
child care
Market value of 654 0 106 1610 0 243 7103 6874 574 1442 0 355 1149 0 316
child-care
subsidies
Market value of 1174 0 122 2889 0 245 2732 0 528 6417 6800 253 1414 0 247
public preschool
Amount of child-care 214 0 21 527 400 42 440 0 120 204 0 48 851 768 49
tax benefits
Note: All values are weighted.

a subset of family characteristics (described families and 48% for families with an employed
above) that determine different program costs mother, are similar to the observed means in the
or reimbursement levels. Among families re- sample.
porting any assistance, we adjust for the type, To examine differences in total assistance by
amount, and duration of care, but we cannot specific family characteristics, we set values for
adjust for other characteristics that might influ- selected characteristics, although holding the
ence the value of assistance received, for exam- value of other characteristics at the sample mean.
ple, differences in program quality and cost and Rather than estimating these results for ‘‘aver-
individual provider rates. age’’ maternal employment, however, we set this
Significant predictive factors for the total value value to 1 (mother is currently employed or in
of governmental assistance include the presence school or training). Results for families with non-
of an infant, race and ethnicity of the respondent, employed mothers (not shown) are substantively
and family income. The value of assistance to the same, although overall levels of assistance are
families with children younger than 2 years is lower.
significantly lower than assistance received by Considering all forms of assistance, total ben-
families who only have preschoolers aged 3 – 5. efits are progressive with income and child
The total value of assistance declines substan- disability. Controlling for all other family charac-
tially with family income; controlling for other teristics, employed families with income below
family characteristics, families with incomes at 75% of the poverty threshold receive about twice
or below 75% of poverty receive more in benefits the total benefit (about $2,700) per year in assis-
than do families with incomes at or above 300% tance than do those with incomes at or above
of poverty (the excluded category), consistent 300% of the threshold (about $1,300). Families
with differences in the value of assistance avail- with a disabled child receive about $1,400 more
able to lower and higher income families (i.e., than those without because of both a greater like-
subsidies and tax credits, respectively). lihood of getting help and a higher conditional
benefit amount.
The distribution of assistance is less progres-
Who Gets What From Government?
sive with respect to other indicators of disadvan-
In Table 6, we combine the two steps to simulate tage and potential need. Families who have
the total dollar value of assistance associated with children aged 3 – 5, but not younger, receive
selected child, parent, and family characteristics, almost $2,000 more in aid per year than do fam-
holding all other characteristics at the sample ilies who have only infants, and over $800 more
mean. Predicted rates of receipt, 43% for all per year more than those families who have both
744 Journal of Marriage and Family

Table 5. Ordinary Least-Squares Regression Results: Value of Assistance, Conditional on Receipt (N ¼ 646)

Value of All Child-Care


Assistance Among Those
Receiving Any Assistance

Variable B SE B

Any infants (age 0 – 2) .09 .28


Any preschoolers (age 3 – 5) .93** .29
Number of children aged 0 – 5 .13 .27
Child in poor health .66 .36
Respondent is single .31 .23
Number of nonparental .29* .13
adults residing in household
Respondent is naturalized immigrant .09 .23
Respondent is noncitizen immigrant .56* .24
Respondent is Hispanic .73** .24
Respondent is Black 1.06** .26
Respondent’s education , high .14 .30
school degree
Respondent’s education ¼ some .09 .22
college or more
Respondent is in poor health .97** .34
Family income , 75% of federal .40 .33
poverty level
Family income 75% – 200% of .45 .27
federal poverty level
Family income 201% – 300% of .11 .26
federal poverty level
Mother is employed .44* .20
Survey is from Wave II .92** .18
2
R .22
Note: Analyses are unweighted.
*p , .05. **p , .01. ***p , .001.

infants and preschoolers. Surprisingly, single- Social Indicators Survey, we are able to estimate
parent families actually receive slightly less in the receipt of the full package of child-care assis-
total aid than do two-parent families, even with tance available to families in the United States
controls for the number of young children in the and to use administrative data to estimate their
family; although their probability of getting help value at the family level.
is somewhat higher, the average benefit for those These analyses are also limited in some impor-
assisted is lower. Largely because of their lower tant respects. Our data on the receipt of assistance
receipt of assistance, noncitizen immigrant fami- is quite detailed at the family level but our impu-
lies receive about $1,000 less in total child-care tation of the value of assistance is necessarily
assistance than do native-born families and over limited. Our imputation of the value of nontax
$800 less than citizen immigrants. assistance is based on average program costs
and reimbursement rates, adjusting for the type
of care, type of assistance, number and ages of
DISCUSSION
children in subsidized care, and the amount and
These analyses are among the first to measure the duration of that care. This does not adjust for
distribution of various forms of child-care assis- variation in other factors that influence the cost
tance received by families with young children. of specific care arrangements, for example, varia-
Because of the level of detail in the New York tion in production costs across public preschool
Distribution of Child-Care Assistance 745

Table 6. Marginal Probability of Receiving Any Child-Care Assistance, Conditional Amount, and Estimated Total Amount

Probability of
All Child-Care Predicted Value of Predicted Value
Assistance Among Assistance Conditional Adjusted for
Those Receiving on Receipt of Any Predicted Probability
Any Assistance Assistance of Receipt

All families with child aged 0 – 5 .43 $3,823 $1,644


Family with an employed mother .48 $3,967 $1,904
Family with an employed mother and
Single-parent household .54 $3,552 $1,918
Two-parent household .46 $4,378 $2,014
All children younger than 6 years are infants .36 $2,539 $914
All children younger than 6 years are preschoolers .56 $5,029 $2,816
Family has both infants and preschoolers .59 $3,272 $1,930
Family income , 75% of poverty .44 $6,130 $2,697
Family income 76% – 200% of poverty .43 $4,578 $1,969
Family income 201% – 300% of poverty .52 $3,370 $1,752
Family income . 300% of poverty .54 $2,418 $1,306
Respondent is naturalized citizen .54 $3,841 $2,074
Respondent is noncitizen immigrant .38 $3,139 $1,193
a
Respondent is native-born U.S. citizen .52 $4,287 $2,229
Focal child does not have a disability .47 $3,873 $1,820
Focal child is disabled or in poor health .63 $5,069 $3,193
Note: All values are weighted. Marginal probabilities and predicted values are calculated holding all other independent vari-
ables at their mean.
a
Includes Puerto Rico.

programs or in fees across private providers. Our currently distributed. Our results suggest that
estimates of the value of tax credits is based on when we consider all forms of assistance, the
family income and general tax rules but may receipt of any child-care help is more common
not capture some adjustments to income that than analysts often assume. Although the United
affect the amount of credit for individual tax fil- States lacks any universal child-care programs,
ers. To the extent that these costs and tax credits about 40% of families with children younger than
vary systematically with family characteristics, 6 years report help with child-care expenses, and
our analyses will underestimate this association. about 50% of those with an employed mother
Our analyses also represent the experiences of receive some type of help. In a jurisdiction with
families in only one city, with relatively generous multiple forms of assistance—through pre-
child-care programs by U.S. standards. Earlier es- schools, state tax credits, and subsidy pro-
timates by Giannarelli et al. (2003) suggest that grams—families who receive any help are also
the receipt of nontax child-care assistance is more likely to receive help from more than one source.
common in New York than in the United States as Although a large share of families receive assis-
a whole. This suggests that our estimates of gov- tance, the amount of assistance is very modest
ernment help reflect the high end of what families for most. The average value of all assistance—
in the United States are likely to receive given about $1,620 annually among all families and
current federal and state policies. $3,980 among families receiving any assistance
With cross-sectional data from a single city, we —obscures substantial variation across families
cannot disentangle the causal mechanisms receiving various forms of help. The most com-
through which policy and individual behavior monly received form of assistance, child-care
interact to determine the distribution of child-care tax credits, are claimed by 25% of families
assistance. We can, however, describe at an but average only $851 annually per recipient
unusually detailed level how this assistance is family. In comparison, means-tested subsidies
746 Journal of Marriage and Family

are received by only about 10% of families but These policies interact with family characteris-
average more than $7,100 per family. The result- tics and individual behavior to produce the final
ing distribution is highly skewed, with about one distribution of assistance. This analysis does not
half of all families receiving nothing, many tell us how policy and family factors interact to
receiving modest levels of assistance, and a few distribute assistance. It does suggest that separate
receiving more substantial aid. child-care programs, when considered as a total
These analyses also reveal that the current dis- system, may have both intended and unintended
tribution of child-care assistance is targeted in distributional consequences. Providing higher
some expected and some unexpected ways. As benefits to lower income families, through means-
we would expect, families in which mothers are tested subsidies, are consistent with target
employed receive more assistance than those in efficiency. The failure of so many low-income
which they are not; families in which a child is families to get any help, however, results in
in poor health or disabled are more likely than a considerably less progressive final distribution
others to get help and to receive a larger package and leaves a substantial share of poor families
of assistance. Single-parent families are more without any help. The extension of preschool
likely to receive help than those with two parents, programs to many 4-year-old children is consis-
but they receive less assistance; the net result is tent with efforts to prepare children for the start
a similar package of assistance for single- and of public school. But families with children aged
two-parent families. More surprisingly, high- 3 – 5 receive nearly three times more assistance
and low-income families are about equally likely than do those with infants younger than 3 years,
to receive some form of child-care assistance. creating substantial inequities between families
Among recipients, the amount of assistance is with infants and those with older preschool-aged
higher for low-income families, who are most children.
likely to receive means-tested subsidies, than Beyond categorical exclusions and restric-
for higher income families who receive tax tions, the child-care system may produce unin-
credits. The resulting distribution is progressive tended and systematic biases. Whereas there is
by income, but less so than we expected: More no explicit policy vis-a-vis parents’ nativity, for
than one half of low-income families receive no example, families headed by immigrant parents
assistance, and, adjusting for both the probability get considerably less assistance than those with
and the value of assistance, families with incomes native-born parents. This may reflect differences
under 75% of poverty receive only about $1,400 in parents’ preferences about child care and gov-
per year more than those with incomes above ernment assistance. It may also reflect institu-
300% of poverty. Families headed by a citizen tional barriers that render child-care assistance
immigrant receive less assistance than those less accessible and/or acceptable to these parents.
headed by a native-born parent, and those that Additional research should focus on why some
are headed by a noncitizen immigrant receive less families are receiving help but others are not, and
than either. examine these distributional issues with nation-
What accounts for this distribution? No form ally representative data. Scholars are examining
of child-care assistance is universal for American the organizational and family dynamics that
families, and tax-based, subsidy, and preschool influence availability and take-up of means-
programs restrict assistance in a variety of ways. tested subsidies among low-income families.
Most fundamentally, child-care tax credits are the These programs are only one part of the total sys-
only form of assistance provided as an entitle- tem for child-care assistance in the United States,
ment for categorically eligible families, although however. To understand and evaluate this sys-
even these benefits exclude families who are too tem, it is important to broaden our inquiry to con-
poor to have a tax liability. Neither means-tested sider the full array of child-care assistance
subsidies nor public preschools are entitlements, programs and the total population of families
and neither are funded at levels sufficient to who benefit from them.
serve all eligible families. Within these funding These results, although limited to New York
constraints, federal and state child-care programs City, raise important questions for policy. The
restrict access, through eligibility criteria and United States differs from many other industrial-
application procedures, and benefit levels through ized countries in its lack of an integrated national
program design, reimbursement, and copayment program for early childhood care and educa-
policies. tion. Federal and state governments fund and
Distribution of Child-Care Assistance 747

administer multiple programs that target specific http://www.childrensdefense.org/earlychildhood/


populations and use a variety of policy tools. childcare/keyfacts2003_childcare_2.pdf
Taken together, these programs may provide Collins, A., Layzer, J., Kreader, J. L., Werner, A., &
the foundation for universal child-care assistance Glantz, F. (2000). National study of child care for
in the United States. Indeed, this study suggests low-income families: State and community sub-
that in a high-provision jurisdiction, assistance study interim report. Boston: Abt Associates.
is reaching close to half of families with young Doherty, K. M. (2002). Early learning. Education
children. As a system, however, these programs Week, 21(17), 54 – 67.
are poorly coordinated in both policy and deliv- Duan, N., Manning, W. G., Morris, C. N., & New-
ery, are failing to achieve a fully inclusive and house, J. P. (1983). A comparison of alternative
equitable system, and may be targeting assistance models for the demand for medical care. Journal
in ways that are unintended by policymakers. of Business & Economic Statistics, 1, 115 – 126.
Feenberg, D., & Coutts, E. (1993). An introduction
to the TAXSIM model. Journal of Policy Analysis
REFERENCES
and Management, 12, 189 – 194.
Adams, G., & Snyder, K. (2003). Essential but often Fender, L., O’Brien, C. T., Thompson, T., Snyder,
ignored: Child care providers in the subsidy K., & Bess, R. (2002). Recent changes in New
system. Washington, DC: Urban Institute. York welfare and work, child care, and child
Administration for Children, Youth and Families. welfare systems. Retrieved July 15, 2004, from
(2002). Temporary Assistance for Needy Families http://www.urban.org/url.cfm?ID ¼ 310564
program participants. Washington, DC: U.S. Garcia, S., & Meyers, M. K. (2005). Income packag-
Department of Health and Human Services, Admin- ing: Measuring the value of cash and in-kind trans-
istration for Children and Families. fers for New York City families. Working paper
Barnett, W. S., Robin, K. B., Hustedt, J. T., & Schul- from Columbia University School of Social Work.
man, K. L. (2003). The state of preschool: 2003 Georges, A., & Wagmiller, R. L., Jr. (2001, Novem-
state preschool yearbook. New Brunswick, NJ: ber). Child care subsidies and child care choices
National Institute for Early Education Research, of low-income parents. Paper presented at the
Rutgers, The State University of New Jersey. meeting of the Association for Public Policy Anal-
Blau, D. M., & Tekin, E. (2003). The determinants and ysis and Management, Washington, DC.
consequences of child care subsidies for single Giannarelli, L., Adelman, S., & Schmidt, S. (2003).
mothers (NBER Working Paper Series). Cambridge, Getting help with child care expenses. Occasional
MA: National Bureau of Economic Research. paper no. 62, Assessing the New Federalism
Burstein, N., Layzer, J. I., & Cahill, K. (2001). project. Washington, DC: Urban Institute.
National study of child care for low-income fami- Gish, M. (2002). Child care: Funding and spending
lies: Patterns of child care use among low-income under federal block grants. Washington, DC: Con-
families. Washington, DC: Abt Associates. gressional Research Service, Library of Congress.
Campbell, D., & Parisi, M. (2003). Statistics of Hofferth, S., Shauman, K., & Henke, R. (1998).
income bulletin: Individual income tax returns, Characteristics of children’s early care and educa-
2001. Retrieved July 15, 2004, from http:// tion programs: Data from the 1995 National
www.irs.gov/pub/irs-soi/01indtr.pdf Household Education Survey. Washington, DC:
Capizzano, J., & Adams, G. (2004). Children in low- U.S. Department of Education.
income families are less likely to be in center- Meyers, M. K., & Heintze, T. (1999). The perfor-
based child care. Retrieved July 15, 2004, from mance of the child care subsidy system. Social
http://www.urban.org/url.cfm?ID ¼ 310923 Service Review, 73, 37 – 64.
Chaudry, A. (2004). Putting children first: How low- Meyers, M. K., Heintze, T., & Wolf, D. (2002). Child
wage working mothers manage child care. New care subsidies and the employment of welfare
York: Russell Sage Foundation. recipients. Demography, 39, 165 – 179.
Children’s Defense Fund. (2003). Good child care Meyers, M. K., Peck, L. R., Davis, E. E., Collins, A.,
assistance policies help low-income working Kreader, J. L., Georges, A., Weber, R., Schexnayder,
families afford quality child care and help children D., Schroeder, D., & Olson, J. A. (2006). The
succeed. In Key facts: Essential information about dynamics of child care subsidy use: A collaborative
child care, early education and school-age care, study of five states. In R. Cabera, L. Hutchens, &
2003 edition. Retrieved July 1, 2004, from L. Peters (Eds.), From welfare to childcare: What
748 Journal of Marriage and Family

happens to young children when mothers exchange Smith, K. (2002). Who’s minding the kids? Child
welfare for work? Mahwah, NJ: Erlbaum. care arrangements: Spring 1997. In Current popu-
National Center for Children in Poverty. (2004). lation reports (pp. 70 – 86). Washington, DC: U.S.
National participant and spending data for the fed- Census Bureau.
eral Child and Dependent Care Tax Credit. Sonenstein, F. L., Gates, G., Schmidt, S. R., &
Retrieved July 8, 2004, from http://www.nccp. Bolshun, N. (2002). Primary child care arrange-
org/policy_detail_14.html ments of employed parents: Findings from the
National Research Council and Institute of Medicine. 1999 National Survey of America’s Families.
(2003). Working families and growing kids: Retrieved June 15, 2004, from http://www.urban.
Caring for children and adolescents. Washington, org/url.cfm?ID ¼ 310487
DC: National Academy of Sciences. Witte, A. D. & Queralt, M. (2002). Take up rates and
Powell, R., & Cahill, M. (2000). Nowhere to turn: trade offs after the age of entitlement: Some
New York City’s failure to inform parents on public thoughts and empirical evidence for child care
assistance about their child care rights. New York: subsidies (Working paper 8886). Cambridge, MA:
NOW Legal Defense and Education Fund. National Bureau of Economic Research.

You might also like