April-June 2016, Vol. 2, No. 2, pp. 1 –12
© The Author(s) 2016. http://ero.sagepub.com
Creative Commons CC-BY: This article is distributed under the terms of the Creative Commons Attribution 3.0 License
(http://www.creativecommons.org/licenses/by/3.0/) which permits any use, reproduction and distribution of the work without further
permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage).
In recent years, scholars and the public have grown more
worried about rising income inequality in the United States,
particularly because inequality is now higher than at any
time since before the Great Depression (Piketty & Saez,
2003, 2013). For some, high inequality is worrying because
it may negatively affect health, crime, social cohesion, and
other outcomes in the present day (Fischer et al., 1996; Van
de Werfhorst & Salverda, 2012). Yet for others, high current
inequality is most problematic because it could lead to
reduced social mobility in future generations by creating dif-
ferent opportunities for children at the top and bottom of the
income distribution (cf. Neckerman & Torche, 2007).
Increasing inequality may harden the class structure if it
affords children different experiences that influence their
Some evidence suggests that the U.S. class structure is
already more likely to reproduce inequalities than in the
past. For example, Reardon (2011) finds that the income-
based gap in achievement scores has increased over time.
Scores for children at the 90th percentile of the income dis-
tribution increased between the mid-1970s and the mid-
1990s, while scores for children at the 10th percentile of the
distribution remained stagnant. Bailey and Dynarski (2011)
find that high-income children’s college entry and comple-
tion increased over a similar period. Increases in the
income-based achievement gap are not solely the result of
class differences in schooling, because the gap exists
already when children enter kindergarten (Reardon, 2011).
However, more recent evidence finds a narrowing of
income-based gaps in school readiness since the 1990s
(Reardon & Portilla, 2015). In general, strong associations
between class and educational outcomes suggest low
intergenerational mobility, since achievement test scores
help predict future earnings (Farkas, 1996; Farkas &
Vicknair, 1996; Jencks & Philips, 1999).
In this article, I investigate trends in one potential expla-
nation for the increasing size of the income-based achieve-
ment gap among children entering school: parental spending
during the early years of childhood. Growing inequality in
spending could link income inequality to a growing gap in
childhood achievement if spending boosts achievement.
Little evidence directly shows that spending on young chil-
dren boosts later achievement, although, as I discuss in detail
below, some evidence supports the assumption that spending
might boost achievement.
It is unclear whether spending on young children has
become more unequal. Existing evidence shows that the gap
in parents’ monetary investments in children has grown over
the last half century between those at the top and bottom of
the income distribution (Kornrich & Furstenberg, 2013).
However, this evidence comes from spending on all children
under the age of 25, and much of the increase in spending
over time is on higher education and thus offers less infor-
mation about trends for younger children. The growth of the
income-based achievement gap among children just entering
school suggests a need for attention to parental investments
in young children as well.
Understanding the sources of shifting investments is also
important. Income has increased at the top of the income
distribution and could have led to increased spending. Yet
changes in income may explain only some of the changes in
parental expenditures. Increases in educational attainment,
women’s labor force participation, and both men’s and
women’s work hours could lead to increased spending. For
Inequalities in Parental Spending on Young Children: 1972 to 2010
This article investigates inequality in parental spending on young children over the period from 1972 to 2010. I find increased
spending among parents at the top of the income distribution but little change among parents at the bottom of the income
distribution. The gap in spending is equally attributable to increased spending on center-based care for preschool-age chil-
dren and spending on enrichment goods and activities. The article examines potential causes of increased spending, including
income, parental education, and wife’s work status, using decomposition analysis. Results indicate that higher incomes are
the largest cause of the increased gap in spending but that increases in wife’s earnings, college completion, and wife’s work
hours are also important for growth in spending.
Keywords: investment, children, child care, inequality
644180EROXXX10.1177/2332858416644180KornrichInequalities in Parental Spending
example, high-income families may have less time to spend
with children because they are more likely now than in the
past to be dual-career families, in which both spouses work
long hours and require more nonfamilial care. Beyond com-
positional changes, parents today may spend more because
of changing norms regarding childhood education and
development and a general diffusion of ideologies of inten-
sive motherhood and concerted cultivation (Hertz, 1997;
This article investigates changes in the income-based gap
in monetary investments in children under the age of 6, when
most children typically have entered school in the United
States. I examine spending on day care, babysitting, and
enrichment-related goods and services for children. I use
data from the Consumer Expenditure Survey (CES) from
1972 to 2010 to track changes in expenditures. I find that
inequality in spending grows substantially and that nearly all
of the increase in the spending gap results from higher
spending among high-income households. I use decomposi-
tion analysis to examine changes in spending over time,
comparing the sources of shifts among those at the top and
bottom of the income distribution. I find that the largest
share of change is attributable to increases in income, with
other increases in spending resulting from increases in wom-
en’s work hours and earnings as well as parental education.
Spending as Investment in Young Children
A literature on spending on children has begun to docu-
ment how parents use resources for and transfer resources to
children of all ages (Folbre, 2008; Hao & Yeung, 2015;
Kaushal, Magnusson, & Waldfogel, 2011; Kornrich &
Furstenberg 2013; Schoeni & Ross, 2005). A human capital
perspective considers investments in children to be one of
the most important a society can make, as investments deter-
mine the future productivity, health, and well-being of a gen-
eration of workers (Becker, 1975; Folbre, 2008). The two
primary forms of parental investment are time and money
(spending on children).
This article examines spending over time to see whether
inequalities of spending on young children have increased
and, particularly, whether high-income parents have increased
spending, since increases in the income-based achievement
gap were attributable mostly to higher test scores among
high-income children (Reardon, 2011). I examine spending
because it offers a potential mechanism to explain how
increases in income have resulted in increases in achieve-
ment. Yet I am unable to directly test whether spending on
young children increases cognitive and achievement-related
outcomes over time, as these data do not contain measures of
spending and achievement over the long period in question.
An investigation of the changing gap in spending offers use-
ful information about whether spending could plausibly be
linked to changes in achievement over time.
Nonetheless, the assumption that spending could improve
outcomes deserves some discussion. In this article, I exam-
ine two areas of targeted spending: child care and spending
on achievement-related goods at home. The assumption that
spending on child care increases achievement depends on
two links: first, that spending increases the quality of care
and, second, that quality improves childhood outcomes.
Existing research provides some evidence that these links
exist. Higher quality appears to lead to better outcomes,
including cognitive skills, based on both observational and
experimental evidence on child care uptake (National
Institute of Child Health and Human Development Early
Child Care Research Network [NICHD ECCRN], 2000; C.
Ramey et al., 2000; Ruzek, Burchinat, Farkas, & Duncan,
2014; Schweinhart, Barnes, & Weikart, 1993). Attending
center-based formal care is also positively related to achieve-
ment (NICHD ECCRN & Duncan, 2003). However, effect
sizes are small, and children of mothers with low levels of
education seem to benefit more from attending high-quality
child care than children of mothers with high levels of edu-
cation, suggesting fewer benefits of child care for high-
income children (Barnett & Boocock, 1998; NICHD
ECCRN & Duncan 2003). Higher costs—which translate
into higher parental spending—are associated with higher
quality, as indicated by smaller child-teacher ratios, small
child group size, and more educational programs (Helburn
& Howes, 1996; Powell & Cosgrove, 1992). Thus, shifts in
child care spending might signal shifts in the quality of care
and its effects on young children.
Similarly, for enrichment goods, increased spending
would need to increase either the quantity or quality of these
goods, and quantity or quality should improve childhood
outcomes. Researchers have measured the presence of toys,
games, books, and other items in the home that may stimu-
late child achievement, using the Home Observation for
Measurement of the Environment Scale (Totsika & Sylva,
2004), and find that these predict better childhood outcomes
(Bradley, Corwyn, McAdoo, & Coll, 2001; Danziger &
Waldfogel, 2000). Although parents can purchase these
items, continuous increases in, for example, the number of
books in the home may not always lead to improvements in
Parental Characteristics and Investments
Understanding the sources of changes in the gap in spend-
ing is also important. One likely reason parents spent more
over time is that top incomes increased over the past 40
years. However, other family characteristics associated with
spending have also changed, including wives’ labor force
participation and the size of families. In addition, parents
today may be more interested in organized activities than in
the past, and some of these shifts may be attributable to
shifts in parental education (Lareau, 2003). Below, I discuss
Inequalities in Parental Spending
family characteristics that predict parental spending on chil-
dren; later in the article, I examine the effect of changes in
these characteristics on changes in spending on children.
Household income is important for parental expenditures
on children because income sets bounds on expenditures.
Because the goal in this article is to investigate whether gaps
in parental spending on children mirror gaps in test scores,
and Reardon (2011) examines spending at the 10th and 90th
percentiles of the income distribution, I focus much of the
analysis on relative income—membership in the top and
bottom decile—rather than absolute income.
Maternal Employment and Earnings
One of the largest categories of expenditures for young
children is child care. Although parents have a range of
motives for spending on child care, its use is often necessary
when there are no parental or other familial caregivers avail-
able. Shifts in labor force participation among women with
young children since the early 1970s may help explain shifts
in spending on children. I thus investigate the effects of
women’s participation in part-time and full-time work on
spending on young children.
Maternal labor force participation may also influence
spending because it shifts the distribution of income in the
household. Women’s share of earned income is typically
associated with greater household spending on services that
might replace women’s household labor (Cohen, 1998; de
Ruijter, Treas, & Cohen, 2005). In England, when control of
money for a child benefit changed from the father to the
mother, households spent more on children (Lundberg,
Pollak, & Wales, 1997). This suggests that women are more
likely than men to direct earnings toward children, so I
investigate whether women’s increased share of earnings is
related to shifts in parental investments in children.
Family Structure and Size
Single-parent families are more common, and families
have fewer children now than in the 1970s. Even net of
income, family structure likely influences parental spending
through a combination of capabilities and needs. Two-parent
families can pool resources and share other expenses, free-
ing up funds for investment in children. Indeed, existing
research finds that married-couple households spend more
than other types of households on children (Ziol-Guest,
Kalil, & DeLeire, 2004).
Children’s characteristics can also influence parents’
behaviors. The number of children is related to children’s
outcomes, like educational achievement and parental invest-
ment (Kuo & Hauser, 1997; though see Guo & VanWey,
1999). Smaller family sizes could lead parents to increase
investments per child as they have more resources per child
available. Beyond the number of children, the gender of
children may also influence parental decision making. Since
parents typically do not select children’s gender, the gender
of children offers a pseudoexperimental comparison, as par-
ents react differently to boys and girls (Pollard & Morgan,
2002). In the 1970s, parents spent more on boys than on girls
(Kornrich & Furstenberg, 2013). Much of this difference
resulted from spending on higher education, so it is unclear
whether this holds for younger children.
Finally, parental education may lead parents to choose
more intense investments in children. Education may be
associated with tastes for investments and use of formal
child care. Rubin (1976) argued that working-class house-
holds distrusted many organized settings for their children,
as they worried about what types of lessons children would
learn in these settings. Similarly, recent research suggests
the growing importance of ideologies of intensive parenting
and “concerted cultivation,” leading middle- and upper-class
parents to devote substantial resources to their children,
often in organized activities (Hertz, 1997; Lareau, 2003). If
these ideologies are new, this should imply that parents
increase spending not only because they have more educa-
tion now than in the past but also because education is more
strongly associated with spending on children. A similar
explanation is that highly educated parents are concerned
about increased competition for college spots, which may
explain increases in time spent with children (G. Ramey &
Data, Measures, and Methods
I use data from the CES, a nationally representative sur-
vey of spending conducted by the Bureau of Labor Statistics.
The CES has been conducted quarterly between 1980 and the
present day. Before 1980, surveys were conducted sporadi-
cally, with the most recent wave from 1972 to 1973. Although
there are waves of the CES at earlier points in time, for exam-
ple 1960 to 1961, these waves do not contain sufficiently
detailed expenditure data to compare them with present data.
For example, spending on child care—both nursery school
and other domestic service—is combined with other house-
hold spending, such as on ice and paper supplies. Thus, I use
data from the 1972-to-1973 survey and then rely on data
aggregated for each year from 1980 to 2010.
Responses in the CES are collected in a diary and an
interview format. I use data from the interview survey, in
which households are asked about expenditures over the pre-
vious 3 months. The interview survey is useful for capturing
expenses, like child care, which are large or regularly
occurring even if they are infrequent because the 3-month
recall period typically will include an instance of spending if
parents spend at all, and the large amount implies that they
will remember spending even despite a long recall period. In
addition, Bureau of Labor Statistics interviewers visit a
household 3 months before the first data are recorded to ask
the household to keep spending records. Thus, the interview
survey should also capture spending on enrichment goods
for children, even though expenses are smaller.
I use a sample of all households that have nonmissing
values for total expenditures and that have any children pres-
ent under the age of 6.
Spending in the interview survey is measured by self-
reports of expenditures over the past 3 months. I examine
child care, which includes both day care and babysitting, and
enrichment spending for children, which includes books,
toys, games, and fees for enrichment activities largely
intended for children. The appendix contains a list of these
goods and services and the CES codes associated with them.
A challenge in measuring spending on children is the
assignment of expenditures to family member, because the
CES does not indicate the intended recipient of expendi-
tures. This difficulty is minimized for child care because
much, although not all, child care spending is for younger
children. For spending on some goods, like books, toys, and
games, assignment is more problematic since spending is
likely to occur for both older and younger children. One pos-
sibility would be to restrict the sample to households that
have no children age 6 or older. This would exclude a sub-
stantial share of the under-6 population. Preliminary analy-
ses suggest that trends and levels of spending are similar
when examining households with only children under the
age of 6. I thus retain households with older as well as
For descriptive purposes, I present both spending per
child and total spending. For spending per child, I use the
number of children under age 6 in the denominator. I use the
Consumer Price Index Research Series (CPI-U-RS) to
express expenditures in 2012 dollars (Sahr, 2013).1
Income. The CES includes measures of earned and unearned
income as well as income before and after taxes. I use mea-
sures of total income before taxes after 1980, and the closest
comparable measure—total family income—for 1972-to-
1973 data. Because these measures are total income, they
include welfare benefits, such as food stamps, which results
in some equalization of income levels. Relying on after-tax
income rather than pretax would likely result in greater
equality due to progressivity in U.S. income taxes. I choose
the pretax measure of income because I expect that reporting
will be more reliable than after-tax income. As with spend-
ing variables, I use the CPI-U-RS to inflate income to 2012
dollars. To ensure confidentiality, the CES censored data
near the top and bottom of the distribution for 1972 to 1973.
Thus, estimates of incomes for that year are not exact but are
a rough average taking censoring into account. Few house-
holds have censored outcomes—only roughly 10% of house-
holds within either the top or the bottom decile in the
1972-to-1973 data have censored incomes.
Income is one of the most frequently missing variables in
the data. In a given year, roughly 10% to 20% of respondents
do not report income. In the descriptive results I present, I
exclude these households and generate income decile cut
points using only households that reported income. Because
respondents report a range of correlated variables, including
total expenditures, I use multiple imputation for missing
income for the regression-based analyses. To impute miss-
ing household and individual incomes, I use PROC MI as
implemented in SAS using Markov chain Monte Carlo
imputation. To impute missing incomes, I include measures
of husband’s and wife’s age, education, weeks and hours
worked, total children, and total household income (in the
case of missing individual incomes). I also include total
household expenditures, which are highly correlated with
income and provide useful additional information about
missing income data. In generating imputations, it is often
useful to set a “seed” to begin the imputation process so that
results are identical in each imputation, rather than relying
on the seed set by the system clock. I use the number 88888
as a seed, and I generate five imputations.
Women’s work status. I use two dichotomous variables to
control for wife’s time in addition to her monetary contribu-
tions. These variables measure whether a woman is at work
part- or full-time, with the reference category being a house-
hold in which the woman reports no paid work.
Women’s share of income. To gauge the effect of women’s
provision of income to the home, I measure the proportion of
reported earned income from women. For single-mother
households, I set the measure to 100%. For single-father
households, I set the measure to 0%.
Family structure. I use three dichotomous variables to
examine family structure, using two-parent households as
the reference category: one for single-mother families, one
for single-father families, and a final category for all other
families. The last category includes, among others, house-
holds in which multiple generations reside in one
Children’s characteristics. I control for a number of charac-
teristics of children. I include a measure of the age of the
Inequalities in Parental Spending
youngest child and a squared term to capture nonlinearities,
as the relationship may not be linear. I also include a mea-
sure for the total number of children, because more children
may create more demands on parental income. Since parents
are now likely to support children even after they leave the
home, this measure includes children ages 0 to 24. I also
include a measure of the gender of children in the home.
Education. Because education may change parental incen-
tives to spend on children, I also control for parents’ educa-
tional level. I rely on mother’s education where possible. For
the 1972-to-1973 data, only the father’s education is listed
when both parents are present, so I use father’s education for
those years. For single-parent households, I use the educa-
tion of the parent in the household. I include variables for
some college and for a college degree or higher, with the
reference category those who completed high school or less.
I do not differentiate between the completion of college and
advanced degrees because the latter category does not exist
in the 1972-to-1973 data.
I first present descriptive evidence on spending over time
for households at different points in the income distribution.
To extend these descriptive results, I investigate why spend-
ing changes in the richest and the poorest households. To do
so, I use a variant of a regression-based decomposition anal-
ysis (Blinder, 1973). Decomposition analysis allows the
examination of sources of differences between two samples.
My primary interest is in how changes in household charac-
teristics have produced changes in spending. I use data from
all years to generate a regression model for spending on chil-
dren and estimate the contribution of changes in characteris-
tics between the early 1970s and the late 2000s. I do so by
subtracting the average characteristics for families in the lat-
ter period from the earlier period and multiplying this differ-
ence by the regression coefficient from the pooled model.
Spending: Trends Over Time and Sources
I begin by showing trends in parental spending on young
children from 1972 to 2010, combining spending on babysit-
ting, day care, and enrichment goods in Figure 1. Panel A
shows spending per child, Panel B shows total spending, and
Panel C shows total spending on a logged scale to better
highlight percentage changes rather than changes in the total
amount spent. Figure 1 shows spending for six groups:
among households in the top decile, the ninth decile, and
each of the quintiles below. Because there are not substantial
differences in trends over time for the bottom eight deciles, I
cluster these into four quintiles.
For both total spending and spending per child, the most
striking pattern is the separation between rich and poor
households in spending on young children. Households in
the top decile triple their total spending, increasing from
$3,000 in the early 1970s to $9,000 in 2010, with steady
increases across the entire period. Spending in the second
decile and second quintile also increased, although mostly
between the early 1970s and the early 1990s. After the
1990s, spending fluctuated between roughly $4,500 and
$6,000 for the second decile and between $3,000 and $4,000
for the second quintile. For the remainder of the income dis-
tribution, spending increases are smaller and appear to occur
only before the mid-1980s. For the third and fourth quintile,
spending increased from the early 1970s through the mid-
1980s, after which spending remained at roughly $2,200 and
$1,500, respectively. These increases are more visible in
Panel C, which highlights percentages changes due to the
change in scale. The bottom quintile, finally, shows practi-
cally no increase over time, although spending is admittedly
higher in the early 1980s than in the 1970s.
The effect of the Great Recession also seems visible
throughout the income distribution. Spending declined
between 2006 and 2008 for those at the top of the income
distribution, as measured by total spending, whereas the tim-
ing was somewhat later across the rest of the income distri-
bution, but there were noticeable decreases between 2008
and 2010 for nearly all other income groups. These patterns
hold for both spending per child and total spending, although
the trend is more pronounced when examining total spend-
ing. The sharper increase in total spending compared to
spending per young child suggests that, over time, higher-
income parents have relatively more young children present
compared to middle-income and lower-income parents.
Figure 2 shows changes in which goods rich parents spent
more on over time compared to poorer parents. To illustrate
these changes, I compare parents in the top decile to those at
the bottom decile, highlighting the differences on day care,
enrichment goods, and babysitting between those at the top
and bottom of the income distribution. Increased spending
on day care and enrichment goods nearly equally account for
increases in spending from the 1970s until the late 1990s.
After the late 1990s, enrichment spending fluctuates,
whereas spending on day care seems to continue to increase.
In the early 1970s, rich parents spent little more than poorer
parents on day care. Day care did not constitute the largest
share of the gap until the early 1990s. After this, spending by
richer parents on day care continued to increase, and the gap
between rich and poor households grew. Over time, it
appears parents made a large switch from babysitting to day
care, with substantial growth between the early 1970s and
the mid-1980s. One potential interpretation is that parents
are more likely to report care services for children as “day
care” despite few structural changes in the care setting.
However, shifts in the number of workers in care occupa-
tions suggest that actual changes occurred in care. For exam-
ple, Kornrich (2012) finds that the number of workers
FIGURE 2. Components of 90-10 spending gap per child.
FIGURE 1. Spending on children by income decile.
identified as babysitters or home care workers declined from
roughly 500,000 in 1970 to about 250,000 in 1980, but the
number of non-home-based child care workers increased
from about 230,000 to nearly 1 million.
It is useful to compare shifts in spending to estimates of
change in the income-based achievement gap among young
children. Reardon (2011) shows that the income-based
achievement gap increased from the 1970s until roughly
2000, largely because of increased achievement among
those at the top of the income distribution. Yet between 1998
and 2010, income-based achievement gaps declined slightly
but significantly (Reardon & Portilla, 2015). Although it is
suggestive at best, trends in spending on enrichment goods
seem to match these patterns slightly better than those on
day care, as the gap in spending on enrichment goods
increases until near the late 1990s and then appears to decline
or at least remain steady, whereas spending on day care con-
tinues to increase. Obviously, other factors, such as shifts in
parenting styles, family structure, or parental education,
could also account for shifts in test scores. Nonetheless,
these data on trends in spending gaps are useful for under-
standing how young children’s environments have changed
over time and across the income distribution.
To understand why the gap in spending increases, I exam-
ine sources of change in the spending patterns of those at the
bottom and top of the income distribution. First, I highlight
differences between the characteristics of high-income and
low-income households in the sample. Table 1 shows snap-
shots of household characteristics from the bottom quintile
and top decile of the income distribution in the early 1970s
and late 2000s.
The two groups have very different characteristics. First,
income differences between the top decile and the bottom
quintile are high and have increased over time. For those in
the bottom quintile, average incomes were around $19,000
in the early 1970s, and declined to roughly $10,000 by the
late 2000s. In contrast, top incomes increased by nearly
$60,000. Family structure differences are also striking. In
both periods, nearly all households in the top decile of earn-
ers are two-parent families. For households in the bottom
decile, two-parent families were rare, at 52% in the early
1970s, and became rarer, at 37% in the 2000s. The number
of single-mother families increased only slightly among
households in the bottom decile. There was a larger increase
in families that did not fit into standard classifications but
instead fell into the other category, which includes house-
holds with other nonspouse adults living in the household.
This could include cohabiting partners but also adult chil-
dren living with their parents or other household arrange-
ments, such as roommates.
Finally, rich households today are more likely than in the
past to have college degrees or higher. Low-income house-
holds also have higher completed levels of education,
although they are still unlikely to have completed college
degrees. Thus, the average characteristics of low-income
households with young children are more different from
high-income households with young children today than in
the past. It is worthwhile emphasizing the sharp contrasts in
income, household structure, and a range of other household
characteristics across the income distribution.
Changes in high-income parents’ characteristics are sug-
gestive of reasons they increased spending. Decomposition
analysis allows a more precise estimate of the importance of
different characteristics. Results from decomposition analy-
ses are presented in Tables 2 and 3 for households in the top
decile and the bottom quintile of the income distribution.
Characteristics of Households in the Bottom and Top Income Deciles
Characteristic Bottom decile Top decile Bottom decile Top decile
Income (US$) 12,145 124,246 8,495 206,464
Two-parent family 41.7 99.2 30.2 93.9
Single father 1.5 0.3 2.3 0.8
Single mother 54.2 0.5 44.9 0.4
Other family 2.6 0 22.6 4.7
Wife’s work hours 12.3 15.9 15.7 29.5
Husband’s work hours
32.3 39.7 27.8 46.5
No high school degree 56.8 11.0 37.6 1.7
High school graduate 34.8 20.2 28.2 8.0
Some college 5.6 18.4 26.0 18.4
College and greater 2.8 50.4 8.2 71.8
These decomposition analyses show only the effects of
changes in average group characteristics. They do so by esti-
mating the change that would occur based on a regression
coefficient estimated using pooled data from 1972 to 2010. I
do so since the effects of changes in household characteris-
tics are of greatest interest.
Table 2 shows coefficients for a regression estimated
within the top decile of households for all years from 1972
to 2010, followed by the group mean for various characteris-
tics in the early 1970s and 2010. I then show the change
attributable to a change in the mean of these variables and
the portion of the overall changes in spending attributable to
each characteristic. As Table 2 shows, the most important
change in the characteristics of high-income households has
been in levels of household income. The increase of nearly
$60,000 between the 1970s and the late 2000s is predicted to
increase spending by $1,300 over this time period, which
accounts for 34% of the total increase in spending, which
was $3,820. The second largest shift comes from the increase
in the share of earnings that comes from the wife, which
increases from near 10% to roughly 31% of household
income and accounts for near 12% of the increase in spend-
ing. Three remaining variables explain around 6% or 7%
each of the increase in spending: increase in wife’s full-time
work, college completion, and decreases in the number of
children in the home. The remaining variables contribute
only negligible amounts to the increasing gap. A substantial
portion of the increase—nearly one third—remains unex-
plained by the variables in this model. One possible interpre-
tation of increases in spending beyond those explained by
composition is that norms surrounding spending caused
similar households to spend more. This may reflect changes
in beliefs about the value of early childhood education or the
importance and availability of enrichment goods. Of course,
as with all regression analyses, these figures represent asso-
ciations rather than causal estimations, and so the decompo-
sition analysis can only suggest which changes in independent
variables are associated with changes in spending.
The decomposition for households in the bottom quin-
tile of the income distribution, in Table 3, shows substan-
tial differences. First, the total change in spending to be
explained is much lower: Spending increased only $152
between 1972 to 1973 and 2008 to 2010 for households in
the bottom quintile. Several variables predict substantial
proportions of this change: wife’s increased share of earn-
ings predict a roughly $140 increase; increases in wife’s
education predict an additional $130, combining the
increases due to attending some college and completing
college; and declines in the number of children predict an
increase of roughly $35. These variables predict increases
greater than the actual increase in spending among this
group. This occurs in part because several changes lowered
Effects of Changing Characteristics on Changing Spending in the Top Decile of Earners, 1972 to 2010
Change due to
change in means
Percentage of spending
difference ($3,820) Rank
Age of youngest child 1598.58 2.48 2.43 –88.21 –0.75 (combined) 13
Age of youngest child
–227.44 9.12 8.85 59.42
Household income (in
22.92 124.25 181.02 1300.92 34.06 1
Percentage of earnings
2189.31 0.10 0.31 458.61 12.01 2
Wife works part-time 1361.45 0.27 0.32 68.11 1.78 6
Wife works full-time 1633.50 0.17 0.32 245.59 6.43 5
Reference person attended
–252.55 0.18 0.22 –8.55 –0.22 10
Reference person is college
1578.20 0.50 0.70 303.65 7.95 3
Single-mother household 1977.05 0.01 0.01 3.06 0.08 9
Single-father household 2397.35 0.00 0.01 13.72 0.36 7
Other household –215.31 0.00 0.06 –13.76 –0.36 11
Mixed gender of children –224.13 0.20 0.28 –19.08 –0.50 12
Children only girls –50.23 0.60 0.47 6.47 0.17 8
Number of children in home –318.52 2.95 2.17 246.94 6.46 4
Total explained 67.46
spending: Household income drops in this sample, leading
to predicted lower spending of $44, and declines in wife’s
likelihood of working part-time are associated with lower
spending. Even with these taken into account, the regres-
sion model still overpredicts spending by nearly 50%. This
may be attributable to other unexplained increases in
demands on households that lower similar households’
ability to spend on young children over time.
Tables 2 and 3 show that many of the determinants of
spending on young children are similar for rich and poor
households: Wife’s education, labor force status, and earn-
ings are all positive. Yet the effects are much larger for high-
income households. This likely reflects greater income
availability, which increases a wife’s ability to make choices
about how to spend in response to these household charac-
teristics. Overall, the regressions suggest both the direct
importance in changing household income for the rich,
because it is the single largest predictor of increased spend-
ing and because it may make spending choices possible for
the highest-income households that lower-income house-
holds are not able to make.
The goal of this article was to investigate whether parental
investments in children had become more unequal. The article
found that inequality across the income distribution in parental
spending on young children has grown steadily over the years
since the early 1970s, largely because of greater spending by
the rich, particularly on enrichment goods and day care. Results
from a decomposition analysis suggest that the largest share of
change is attributable to increased income at the top of the
income distribution, although wife’s increased share of income,
full-time work, and education are also important determinants.
This article thus extends existing research in two ways. First,
it shows that inequality in parental spending on young children
increases similarly to parental spending on older children. One
key difference is that Kornrich and Furstenberg (2013) find that
spending per child increases across the income distribution,
whereas this article suggests change primarily in the top two
income deciles. Second, it investigates which independent vari-
ables are responsible for these increases. Although this article
examine household characteristics, changes in context may also
be important. Today’s parents have more choices for how to
Effects of Changing Characteristics on Changing Spending in the Bottom Quintile of Earners, 1972 to 2010
Change due to
change in means
Percentage of spending
difference ($152) Rank
Age of youngest child 178.36 1.91 2.09 32.90 12.52 (combined) 5
Age of youngest child
–18.70 6.38 7.12 –13.83
Household income (in
5.07 18.74 10.01 –44.23 –29.02 13
Percentage of earnings
377.23 0.06 0.42 138.28 90.73 1
Wife works part-time 295.08 0.19 0.06 –38.38 –25.18 12
Wife works full-time 676.71 0.01 0.03 8.33 5.46 6
455.21 0.08 0.26 82.87 54.37 2
Reference person is
1139.54 0.04 0.08 47.79 31.35 3
–92.14 0.35 0.37 –1.78 –1.17 10
217.63 0.01 0.02 2.50 1.64 7
Other household –170.18 0.12 0.24 –20.10 –13.19 11
Mixed gender of
–13.62 0.27 0.26 0.15 0.10 9
Children only girls –19.08 0.58 0.47 2.02 1.33 8
Number of children in
–76.70 2.85 2.39 34.99 22.96 4
Total explained 151.90
spend on young children. There are specialized Montessori
schools, bilingual schools, and other programs focused on ath-
letic, musical, or other skills for young children.
This article began by noting that changes in spending are of
interest because they might offer at least a partial explanation
for changes in the income-based achievement gap. To the extent
that high-income children’s test scores at entry to school have
increased, and spending on children during early childhood has
increased among the same group, there is a plausible link
between the two. This may be particularly salient for spending
on enrichment goods and services, which began to stagnate or
decline at a similar point (the late 1990s), as did high-income
children’s test scores (Reardon & Portilla, 2015). For lower-
income children, whose test scores have not grown as quickly,
trends in spending are also roughly flat. Of course, there could
be other reasons that children of high-income parents do better
now than in the past. Parents appear to be more interested in
strategies of concerted cultivation (Lareau, 2003) and also
spend more time in direct interaction with their children than
they did in the past (Bianchi, 2000; Gauthier, Smeeding, &
Furstenberg, 2006). Future research that can attempt to investi-
gate the effects of additional spending on future child outcomes
might offer useful insights on some of these issues.
One limitation is that measuring investment through
spending may misrepresent investment, since higher spending
might not always mean more investment. Two alternatives to
spending stand out: Parents might rely on kin care, which may
be high quality, and low-income parents may take advantage
of subsidized programs, like Head Start, that do not entail
spending. For kin and other unpaid care, the evidence above
on the use of child care suggests that, on average, high-quality
center-based child care will lead to better outcomes. In the
case of subsidized programs, although it is not the main focus
of this article, increases in the percentage of low-income
young children in these programs might explain why there is
little increase in spending. Indeed, the size of Head Start has
expanded, from 379,000 spots in 1972 and 1973 to slightly
over 900,000 spots in 2008 through 2010, and this has
expanded coverage from roughly 10% of children under age 6
in the early years to near 15% of children in this age group in
recent years. If these programs offer the same high-quality
care as those that high-income children attend, then increased
high-income spending on child care would obviously offer
little leverage for understanding increasing test scores for
high-income children. Whether the quality of public and pri-
vate programs has remained similar over time is another
important area for investigation.
Codes for Expenditures Included in
Home Enrichment Spending
290420 Infants’ furniture
320130 Infants’ equipment
590211 Magazine subscriptions
590212 Magazines, nonsubscription
590220 Books through book clubs
590230 Books not through book clubs
600210 Ping-Pong, pool tables, other similar recreation
room items, general sports equipment, and health and exer-
600410 Camping equipment
600420 Hunting and fishing equipment
600430 Winter sports equipment
600901 Water sports equipment
600902 Other sports equipment
610110 Toys, games, hobbies, tricycles, and battery-pow-
610120 Playground equipment
610130 Musical instruments, supplies, and accessories
620211 Admission fees for entertainment activities, includ-
ing movie, theater, concert, opera, or other musical series
(single admissions and season tickets)
620221 Admission fees to sporting events (single admis-
sions and season tickets)
620310 Fees for recreational lessons or other instructions
660310 Encyclopedia and other sets of reference books
690111 Computers, computer systems, and related hardware
for nonbusiness use
690112 Computer software and accessories for nonbusiness
A previous version of this article was presented at the conference
“Income, Inequality, and Educational Success: New Evidence
About Socioeconomic Status and Educational Outcomes.” The
author would like to thank the participants at that conference,
Noam Lupu, John Pothen, and Timothy Dowd for their helpful
1. The Consumer Price Index Research Series is a new Consumer
Price Index series incorporating methodological improvements,
such as the use of rental equivalence for homeowner costs and
quality adjustments for prices (Stewart & Reed, 1999).
Bailey, M. J., & Dynarski, S. M. (2011). Inequality in postsec-
ondary education. In G. J. Duncan & R. J. Murnane (Eds.),
Whither opportunity? Rising inequality, schools, and children’s
life chances (pp. 117–132). New York, NY: Russell Sage
Barnett, W. S., & Boocock, S. S. (Eds.). 1998. Early care and edu-
cation for children in poverty: Promises, programs, and long-
term results. Albany: State University of New York Press.
Becker, G. S. (1975). Human capital: A theoretical and empirical
analysis, with special reference to education (2nd ed.). Chicago,
IL: University of Chicago Press.
Inequalities in Parental Spending
Bianchi, S. (2000). Maternal employment and time with children:
Dramatic change or surprising continuity? Demography, 37,
Blinder, A. (1973). Wage discrimination: Reduced form and struc-
tural estimates. Journal of Human Resources, 8, 436–455.
Bradley, R. H., Corwyn, R. F., McAdoo, H. P., & Coll, C. G.
(2001). The home environment of children in the United States
Part I: Variations by age, ethnicity, and poverty status. Child
Development, 72, 1844–1867.
Cohen, P. N. (1998). Replacing housework in the service economy:
Gender, class, and race-ethnicity in service spending. Gender &
Society, 12(2), 219–231.
Danziger, S., & Waldfogel, J. (Eds.). (2000). Securing the future:
Investing in children from birth to college. New York, NY:
Russell Sage Foundation.
De Ruijter, E., Treas, J. K., & Cohen, P. N. (2005). Outsourcing the
gender factory: Living arrangements and service expenditures
on female and male tasks. Social Forces, 84(1), 305–322.
Farkas, G. (1996). Human capital or cultural capital? Ethnicity
and poverty groups in an urban school district. New York, NY:
Farkas, G., & Vicknair, K. (1996). Appropriate tests of racial wage
discrimination require controls for cognitive skill: Comment on
Cancio, Evans, and Maume. American Sociological Review, 61,
Fischer, C. S., Hout, M., Jankowski, M. S., Lucas, S. R., Swidler,
A., & Voss, K. (1996). Inequality by design: Cracking the bell
curve myth. Princeton, NJ: Princeton University Press.
Folbre, N. (2008). Valuing children: Rethinking the economics of
the family. Cambridge, MA: Harvard University Press.
Gauthier, A. H., Smeeding, T. M., & Furstenberg, F. F. (2004).
Are parents investing less time in children? Trends in selected
industrialized countries. Population and Development Review,
Guo, G., & VanWey, L. K. (1999). Sibship size and intellectual
development? Is the relationship causal? American Sociological
Review, 62, 169–187.
Hao, L., & Yeung, W.-J. J. (2015). Parental Spending on school-
age children: Structural stratification and parental expectation.
Demography, 52, 835–860.
Helburn, S. W., & Howes, C. (1996). Child care cost and quality.
The Future of Children, 6(2), 62–82.
Hertz, R. (1997). A typology of approaches to child care: The
centerpiece of organizing family life for dual-earner couples.
Journal of Family Issues, 18(4), 355–385.
Jencks, C., & Phillips, M. (1999). Aptitude or achievement:
Why do test scores predict educational attainment and earn-
ings?” In S. E. Mayer & P. E. Peterson (Eds.), Earning and
learning: How schools matter (pp. 15–48). Washington, DC:
Kaushal, N., Magnuson, K., & Waldfogel, J. (2011). How is
family income related to investments in children’s learning?
In R. M. Murnane & G. Duncan (Eds.), Whither opportu-
nity? Rising inequality and the uncertain life chances of
low-income children (pp. 187–206). New York, NY: Russell
Kornrich, S. (2012). Hiring help for the home: Household services
in the twentieth century. Journal of Family History, 37(2),
Kornrich, S., & Furstenberg, F. (2013). Investing in children:
Changes in spending on children, 1972 to 2007. Demography,
Kuo, H.-H. D., & Hauser, R. M. (1997). How does size of sibship
matter? Family configuration and family effects on educational
achievement. Social Science Research, 26, 69–95.
Lareau, A. (2003). Unequal childhoods: Class, race, and family
life. Berkeley: University of California Press.
Lundberg, S., Pollak, R., & Wales, T. (1997). Do husbands and
wives pool resources? Evidence from the U.K. child benefit.
Journal of Human Resources, 33, 463–480.
Neckerman, K. M., & Torche, F. (2007). Inequality: Causes and
consequences. Annual Review of Sociology, 33, 335–357.
NICHD Early Child Care Research Network. 2000. The relation
of child care to cognitive and language development. Child
Development, 71, 960–980.
NICHD Early Child Care Research Network & Duncan, G. J.
(2003). Modeling the impacts of child care quality on children’s
preschool cognitive development. Child Development, 74(5),
Piketty, T., & Saez, E. (2003). Income inequality in the United
States, 1913–1988. Quarterly Journal of Economics, 88(1), 1–39.
Piketty, T., & Saez, E. (2013). Tables and figures updated to
2012. Retrieved September 12, 2013, from http://elsa.berkeley.
Pollard, M. S., & Morgan, S. P. (2002). Emerging parental gender
indifference? Sex composition of children and the third birth.
American Sociological Review, 67, 600–613.
Powell, I., & Cosgrove, J. (1992). Quality and cost in early
childhood education. Journal of Human Resources, 27(3),
Ramey, C. T., Campbell, F. A., Burchinal, M., Skinner, M. L.,
Gardner, D. M., & Ramey, S. L. (2000). Persistent effects of
early childhood education on high-risk children and their moth-
ers. Applied Developmental Science, 4, 2–14.
Ramey, G., & Ramey, V. A. (2010). The rug rat race. Brookings
Papers on Economic Activity, 42(1), 120–199.
Reardon, S. F. (2011). The widening achievement gap between
the rich and the poor: New evidence and possible explanations.
In R. M. Murnane & G. Duncan (Eds.), Whither opportunity:
Rising inequality, schools, and children’s life chances (pp. 91–
116). New York, NY: Russell Sage Foundation.
Reardon, S. F., & Portilla, X. A. (2015). Recent trends in socio-
economic and racial school readiness gaps at kindergarten
entry (Working Paper 15-02). Stanford, CA: Stanford Center
for Education and Policy Analysis.
Rubin, L. B. (1976). Worlds of pain: Life in the working-class fam-
ily. New York, NY: Basic Books.
Ruzek, E., Burchinat, M., Farkas, G., & Duncan, G. J. (2014). The
quality of toddler child care and cognitive skills at 24 months:
Propensity score analysis results from the ECLS-B. Early
Childhood Research Quarterly, 29(1), 12–21.
Sahr, R. C. (2013). CPI conversion factors to convert to 2012 dol-
lars. Retrieved from http://oregonstate.edu/cla/polisci/sites/
Schoeni, R. F., & Ross, K. (2005). Material assistance received
from families during the transition to adulthood. In R. A.
Settersten, F. F. Furstenberg, & R. G. Rumbaut (Eds.), On
the frontier of adulthood: Theory, research, and public policy
(pp. 396–416). Chicago, IL: University of Chicago Press.
Schweinhart, L. J., Barnes, H. V., & Weikart, D. P. (1993).
Significant benefits: The High/Scope Perry Preschool Study
through age 27. Ypsilanti, MI: High/Scope Press.
Stewart, K. J., & Reed, S. B. (1999, June). CPI research series
using current methods, 1978-98. Monthly Labor Review, 29–38.
Totsika, V., & Sylva, K. (2004). The Home Observation for
Measurement of the Environment revisited. Child and
Adolescent Mental Health, 9(1), 25–35.
Van de Werfhorst, H., & Salverda, W. (2012). Consequences of
economic inequality: Introduction to a special issue. Research
in Social Stratification and Mobility, 30, 377–387.
Ziol-Guest, K. M., Kalil, A., & DeLeire, T. (2004). Expenditure
decisions in single-parent households. In A. Kalil & T. DeLeire
(Eds.), Family investments in children’s potential: Resources
and parenting behaviors that promote success (pp. 181–208).
Mahwah, NJ: Lawrence Erlbaum.
SABINO KORNRICH is an assistant professor of sociology at
Emory University, 1555 Dickey Dr., Atlanta, GA 30322; korn-
email@example.com. His research focuses on inequalities within and
between families, particularly focusing on household expenditures
and household labor.