WORKING
P A P E R
Medicaid at Birth,
WIC Take Up, and
Children’s Outcomes
MARIANNE BITLER
JANET CURRIE
WR-172
May 2004
A Final Report to the IRP/USDA Small Grants Program
This product is part of the RAND
Labor and Population working
paper series. RAND working
papers are intended to share
researchers’ latest findings and
to solicit additional peer review.
This paper has been peer
reviewed but not edited. Unless
otherwise indicated, working
papers can be quoted and cited
without permission of the author,
provided the source is clearly referred
to as a working paper. RAND’s
publications do not necessarily
reflect the opinions of its research
clients and sponsors.
is a registered trademark.
Medicaid at Birth, WIC Take Up, and Children’s Outcomes
A Final Report to the IRP/USDA Small Grants Program
Marianne Bitler,
RAND Corporation
Janet Currie,
UCLA and NBER
May 2004
The opinions expressed do not necessarily represent those of RAND Corporation or any other institution. We thank
Pinka Chatterji, Julie DaVanzo, Sandy Hofferth, Mark Nord, Karl Scholz, Chris Swann, Laura Tiehen, Shelly Ver
Ploeg, and participants in the IRP-USDA Small Grants Pre-conference and Final Conference, the PAA annual
meeting, and the APPAM annual meeting for helpful comments. We thank Aaron Yelowitz for generously
providing his Medicaid coding, and we thank the staff at the Food and Nutrition Service for providing us with
information about older WIC PC reports. Bitler and Currie gratefully acknowledges the financial support of the U.S.
Department of Agriculture, the Institute for Research on Poverty, and the National Institute of Child Health and
Human Development and National Institute on Aging. Currie also acknowledges the support of Princeton’s Center
for Health and Well Being.
Abstract
The Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) provides
food and nutritional advice to low-income women, and infants and children, who are income
eligible and are nutritionally-at-risk. The effects of WIC on infants have been extensively
studied, but children one to four are the most rapidly growing part of the WIC caseload, and little
information is available about the effects of WIC on this group.
n
o
s
t
c
e
f
f
e
m
r
e
t
g
n
o
l
d
a
h
s
t
n
a
f
n
i
g
n
o
m
a
p
u
e
k
a
t
d
e
t
c
e
f
f
a
t
a
h
t
s
e
i
c
i
l
o
p
d
i
a
c
i
d
e
M
t
a
h
t
w
o
h
s
e
w
r
a
e
ns
d
i
l
a
s
o
e
e
ss
l
b
d
a
i
r
e
g
a
r
i
c
wn
l
e
oi
t
e
ed
dc
a
i
n
a
m
e
c
i
H
s
d
e
e
i
.i
l
Mn
o
m
fi
t
a
o
a
f
p
yi
e
t
i
c
m
s
i
t
o
o
r
r
c
a
e
n
p
n
i
e
ne
r
g
o
h
ec
tg
h
i
a
t
h
p
nm
y
i
sib
e
n
h
s
e
c
a
v
u
e
i
r
mr
c
n
gd
i
n
n
,v
e
i
t
e
s
.
b
a
a
d
r
h
tu
e
t
o
n
t
u
n
o
g
o
c
r
e
a
h
v
yi
t
e
a
Bw
v
h
a
.
nh
y
m
e
t
i
s
r
al
c
d
i
r
i
l
b
g
t
i
i
i
h
o
r
g
c
r
c
i
l
p
e
ge
n
CCo
m
I
o
I
m
WW
s
a
s
e
a
d
n
h
e
,
t
o
s
P
i
t
na
I
a
i
e
H
p
nr
i
c
C
o
c
n
i
i
S
i
t
t
r
a
f
n
a
p
o
e
p
i
r
t
c
l
d
i
C
t
u
l
I
r
i
s
a
hW
e
p
c
r
Using data from the 1996 and 2001 panels of the Survey of Income and Program Participation
(SIPP),
e
h
t
x
e
ns
o
r
so
t
f
c
e
e
f
l
f
i
t
e
n
te
n
c
a
r
c
e
i
p
f
i
h
n
t
5
g
i
8
s
e
dh
n
t
a
n
eh
a
g
t
r
a
r
l
e
t
sa
a
e
h
r
rg
u
I
o
M
f
eB
g
d
a
a
th
a
.
n.
e
o
i
i
(
t
a
t
p
h
i
g
c
i
i
e
t
r
w
a
r
p
e
v
Co
I
f
Wo
tk
a
s
h
i
t
r
sa
t
i
gi
s
n
i
d
d
l
n
i
i
h
f
c
ga
n
t
i
a
k
i
h
r
t
t
s
y
t
tl
i
s
i
o
b
ma
b
ro
u
r
Op
e
h
st
i
f
do
o
s
o
s
f
e
yc
c
h
t
u
l
s
a
e
e
h
h
t
ff
o
o
ne
o
r
i
u
s
i
s
v
a
o
e
r
m
p
t
ln
a
a
u
t
t
r
c
o
a
p
em
h
i
t
ra
n
o
s
ni
o
i
s
t
i
a
h
c
T
u
d
.
e
n
e
nr
o
d
i
l
t
i
i
h
r
t
c
u
n
g
n
eu
h
o
t
ry
e
g
h
n
t
i
o
e
m
ta
a
h
y
t
t
i
ss
t
e
s
e
b
g
o
g
t
u
n
s
e
v
se
i
h
r
Tp
o
.t
)
e
g
g
n
a
i
p
de
l
n
h
a
e
h
t
f
o
e
s
u
a
c
e
b
d
n
a
,
t
a
e
r
h
t
h
t
l
a
e
h
c
i
l
b
u
p
a
s
a
y
t
i
s
e
b
o
f
o
e
c
n
a
t
r
o
p
m
i
e
h
t
f
o
e
s
u
a
c
e
b
m
a
r
g
o
r
p
C
I
W
.
e
f
i
l
n
i
y
l
r
a
e
s
t
i
b
a
h
g
n
i
t
a
e
y
h
t
l
a
e
h
g
n
i
h
s
i
l
b
a
t
s
e
f
o
e
c
n
a
t
r
o
p
m
i
I. Introduction
The Special Supplemental Nutrition Program for Women, Infants and Children (the WIC
program) is a federally-funded, state-run program that provides direct nutritional supplements
and nutritional advice to infants and children, as well as to pregnant, postpartum, and lactating
women. While WIC was originally charged with preventing hunger, there is growing
recognition that today many American children are at risk of obesity. The incidence of obesity is
rising faster among children than among adults and recent estimates suggest that obesity will
cause as many future deaths as smoking (Mokdad, Marks, Stroup, and Gerberding, 2004).
Moreover, since the poor are at higher risk of being overweight, the health consequences will fall
disproportionately on those of low income, exacerbating health inequalities. The Surgeon
General’s report on obesity emphasizes the importance of early intervention to prevent obesity,
since once formed, eating habits are hard to change (U.S. Department of Health and Human
Services, 2001).
Hence, the prevention of overweight and obesity is now an important goal of WIC. WIC
could reduce the risk of obesity either by providing nutritious foods that substitute for less
nutritious foods that are high in calories, or through nutrition education. It is easy to imagine,
for example, that having the WIC staff tell the mother that the child was overweight and at risk
for future health problems could have an impact on the family’s diet.
WIC serves a large share of eligible low-income pregnant women and infants and the
effects of WIC on this population have been extensively studied. However, the most rapid
growth in participation in the WIC program is for children aged one to four, and participation
rates among children are relatively low––Bitler, Currie and Scholz (2003) estimate that while
1
73% of eligible infants take up the program, only 38% of eligible one to four year old children
use WIC. Fifty-seven percent of all one to four year old children are eligible for WIC, so that
future growth in the program will likely come from efforts to enroll larger numbers of these
children. It is important to see what effect such efforts might have on children’s health and
nutritional status.
This report focuses on the determinants of WIC take up among four year old children and
on the effects of childhood WIC participation on children four to six years old. We use rich data
on WIC participation, child anthropometrics, health, and health care utilization from the 1996
and 2001 panels of the Survey of Income and Program Participation (SIPP). We show that
recent changes to the Medicaid program had significant effects on the take up of WIC, and use
these changes as instruments for identifying the effects of WIC on child outcomes.
There are two types of changes to the Medicaid program that may have increased WIC
participation among four year old children. First, higher Medicaid cutoffs for infants are likely
to affect childhood WIC participion because most children who use WIC began using the
program as infants, and Medicaid confers automatic eligibility for WIC. Hence, it is possible
that higher Medicaid cutoffs when children were infants induced their mothers to join both
Medicaid and WIC, and that some fraction of these new child entrants remained on the program
through early childhood.
The second type of Medicaid eligibility changes occurred through the State Child Health
Insurance Program (SCHIP). Under SCHIP, states were given the option of extending public
health insurance to uninsured children either by expanding Medicaid or by creating a separate,
stand-alone program (or some combination). Because categorically eligible Medicaid
2
participants are eligible for WIC, states that used SCHIP to expand Medicaid also expanded
eligibility for WIC among children. This expanded eligibility could have led to expanded WIC
coverage among children of higher income levels than those of the typical WIC participant.
In fact, Besharov and Germanis (2001) argue that expansions of WIC eligibility to
women and children of higher income levels have greatly increased WIC expenditures. Their
argument implies that the increases in WIC participation among children have been largely
driven by children from higher income families. However, we find that while Medicaid cutoffs
for infants have a strong effect on children’s WIC take up, take up among children who became
eligible because of SCHIP was low, and so SCHIP had little impact on WIC caseloads.
In order to examine the impact of WIC on child outcomes, it is necessary to account for
selection into the program. We show that WIC participants are negatively selected even relative
to the population of eligibles so that failure to adequately control for selection into the program
will lead to negative estimated effects of WIC. We draw on the results regarding take up and use
the Medicaid income cutoff when the child was an infant as an instrument for WIC participation
at age four.
However, it is possible that Medicaid coverage at the time of the birth has long-term
effects on child health. Since, as we will show below, the vast majority of children on WIC at
age four were on WIC as infants and had Medicaid coverage as infants (and conversely, take up
of WIC among eligible infant Medicaid recipients is high), it is likely to be difficult to
disentangle the effects of Medicaid coverage at birth and WIC participation over the child’s early
life. We thus interpret our estimates as the combined effect of Medicaid at birth and continuous
WIC coverage up to age four.
3
Our main finding is that WIC participation at age four is associated with a large and
statistically significant reduction in the probability that children are at risk of overweight
(defined as having a Body Mass Index or BMI above the 85th percentile for sex and age). Hence
WIC is successfully preventing overweight in young children, which is likely to have
implications for their future risk of contracting obesity-related diseases. We find no significant
effect of WIC on measures of access to health care, which supports the interpretation that the
effects on weight and obesity are due to the nutritional component of the WIC program rather
than to any links between WIC and current access to medical care.
The rest of this report is laid out as follows. Section II provides necessary background
information about WIC and the Medicaid expansions relevant to WIC participation. Section III
discusses the SIPP data and presents preliminary sample statistics. Section IV provides an
overview of our statistical methods, while results appear in Section V. Section VI presents our
conclusions.
II. Background Information about WIC and Medicaid
a) The WIC program
WIC is entirely federally funded, and provides both nutritional education and food
packages to eligible women, infants, and children. There are both categorical eligibility
requirements and program eligibility requirements. Categorically eligible groups are infants,
children aged one to four and pregnant, breastfeeding, and postpartum women. If categorically
eligible, participants must also be at nutritional risk; though in practice this condition is rarely
binding.
4
WIC has been expanding rapidly over time. Federal program expenditures increased from
$256 million in fiscal year 1977 ($765 million in 2001 dollars) to $4.1 billion in FY 2001 while
participation went from 848,000 per month in 1977 to 7.3 million per month in fiscal year 2001.
Almost half of 2000 participants were children aged one to four.
WIC participants receive vouchers that can be used to purchase specific foods that are
selected because they contain protein, calcium, iron and/or vitamins A and C. The list of
approved foods includes milk, cheese, and peanut butter. Food packages are tailored to
nutritional needs so that they differ for women, infants, and children. Participants receive
nutritional counseling and are encouraged to breast-feed their children (though WIC’s provision
of free infant formula may undermine this advice). Many WIC programs are contracted out to
non-profit agencies and are connected to maternal and child health clinics.
In order to be eligible for the program, individuals must be low income (under 185
percent of the federal poverty level or FPL), nutritionally-at-risk (definitions of this vary by
state) and in one of the categories (pregnant or postpartum women, infants, or children under
five). However, Medicaid, Food Stamp and AFDC/TANF participants are adjunctively eligible,
regardless of their income. Since income cutoffs for Food Stamps and AFDC/TANF are much
lower than 185% of the poverty line, while cutoffs for Medicaid may exceed 185% of poverty,
the major channel for adjunctive eligibility is through the Medicaid program. WIC providers
refer mothers to other health care providers such as Medicaid and to immunization programs and
conversely, these other agencies also refer eligibles to WIC clinics. With the recent Medicaid
eligibility expansions, nearly half of all infants born in the U.S. are eligible for WIC (Bitler,
Currie, and Scholz, 2003).
5
WIC use among children is closely tied to WIC use during the first year of life as we
show below. At the same time, many children leave WIC at age one, when the WIC package of
food changes (to exclude infant formula) and families must have their WIC eligibility
redetermined (Burstein et al. 2000). Infants whose mothers are not exclusively breastfeeding
receive baby formula in their WIC package. The cost of infant formula makes the infant package
valuable: A recent US Department of Agriculture report found that in 2000 a 13 ounce can of
liquid concentrate cost between $2.59 and $3.11 (Olievera, Prell, Smallwood, and Frazeo, 2001)
so that the 403 fluid ounces permitted in an infant package would be worth $80 to $96. By
contrast, the value of the package available to the average four year old WIC participant in our
data is $31. The value of the WIC infant package, along with the strong relationship between
WIC use as an infant and WIC use later, suggests that the Medicaid eligibility threshold for
infants may be a strong predictor of later WIC use.
Most previous evaluations of WIC focus on the effects on birth outcomes and infant
mortality. These studies have found that WIC is associated with a decrease in the risk of infant
mortality, a lower probability of a small-for-gestational-age birth, and reductions in the cost of
maternal and neonatal care, among other beneficial effects.1
As noted above, to be eligible for WIC, women must be nutritionally-at-risk and either
under 185 percent of the poverty line or participating in the AFDC/TANF, Medicaid, or Food
1
Papers about the effect of WIC on birth outcomes and nutrition include Ahluwalia et al.
(1998), Devaney (1992), Devaney et al. (1992), Devaney and Shirm (1993), and Moss and
Carver (1998).
6
Stamp programs. If these women are less well-off along other dimensions than the population at
large, ordinary least squares (OLS) estimates of the impact of WIC are likely to understate the
program's true beneficial effects. Alternatively, if only the more skilled mothers among women
eligible for WIC manage to obtain benefits, OLS estimates of the effects of WIC may be biased
upwards. Moreover, the literature points out that women who have longer pregnancies have
longer to get onto the WIC program. This could result in a non-causal correlation between
gestation (or other outcomes) and having been on the WIC program.
Most of the previous studies do not control carefully for potential biases due to selection
into the WIC program, leading to a concern that their positive findings are driven by positive
selection. Bitler and Currie (2004) provide evidence that compared to eligibles who do not
participate, WIC participants are in fact negatively selected on a wide range of observable
variables, and they confirm that WIC use during pregnancy has a positive effect on infants.
Other recent papers that consider possible selection into the WIC program include Brien and
Swann (2001), Kowaleski-Jones and Duncan (2000, 2002), and Chatterji et al. (2002).
A second weakness of the existing WIC literature is the virtually exclusive focus on
pregnancy and infant health outcomes rather than the outcomes of older children. KowaskeiJones and Duncan (2000) is one of the few papers that looks at the impact of WIC on children
rather than infants. Using data from the NLSY from 1990 to 1996, they find that in fixed-effect
models which compare siblings who received WIC prenatally with other siblings, WIC had a
positive impact on child temperament. However, this finding is not robust to the use of
instrumental variables methods and relies on a small set of families where one child received
7
WIC at a particular age while another did not.2
b) The Interaction Between Medicaid and WIC
The Medicaid program provides public health insurance coverage for low income women
and children. Until the mid 1980s, most children became eligible for Medicaid because their
mothers received cash welfare through the Aid to Families with Dependent Children (AFDC)
program. Beginning in the mid-1980s, Congress enacted a series of laws expanding Medicaid
eligibility to families who did not qualify for AFDC either because of their income or because of
their family structure. By 1991, states were required to cover pregnant women with incomes less
than 1.33 times the poverty line, and states were eligible to receive federal matching funds to
increase the cutoff to 1.85 times the poverty line. Income cutoffs for infants under the Medicaid
program are shown in Table 1 for 1996 and 2002.
While most state Medicaid eligibility thresholds did not rise above the WIC income
cutoff of 1.85 times poverty, Medicaid is a far more valuable program, and hospitals face strong
incentives to enroll eligible pregnant women. Hospitals are required by law to treat women in
active labor. Prior to the Medicaid expansions, uncompensated care for maternity patients was a
significant burden to hospitals. Now, hospitals can get reimbursed for this care by the Medicaid
program. In fact, since 1993, about 37% of deliveries have been paid for by the Medicaid
2
This is partially due to a data issue with the NLSY. The NLSY only asks about any
WIC participation within the family. It is impossible to determine who is receiving WIC in the
family using the NLSY, unlike the SIPP, which details WIC use by any eligible persons.
8
program (National Governor’s Association, 2003). These women and their infants were
automatically eligible for WIC, and did not have to go through any further certification process.
The automatic link between Medicaid coverage and WIC eligibility suggests that changes
in the Medicaid income cutoffs that fell below the WIC income cutoff could still have had an
impact on WIC take up. For example, increases in the Medicaid cutoff from 1.33 times poverty
to 1.5 times poverty brought many more pregnant women onto Medicaid. These women were
always eligible for WIC, but may not have been willing to undertake a separate WIC application.
Once they gained Medicaid coverage, they became automatically eligible for WIC, which is
likely to have increased WIC take up. Bitler and Currie (2004) show that 81% of women whose
deliveries were covered by Medicaid used WIC during their pregnancies.
The SCHIP program was established in the Balanced Budget Act of 1997. SCHIP allows
states to expand public health insurance to children who were previously ineligible for Medicaid
(and also to their parents in some states). Originally, states could extend eligibility to children in
families with incomes up to twice the federal poverty level (or higher in some cases). Some
states have now extended eligibility for public health insurance to children in families with
incomes up to 3 or 3.5 times the federal poverty line.
Table 1 shows the income cutoffs for SCHIP and whether the program was implemented
as a Medicaid expansion or as a separate program.3 This distinction is important because if the
state implemented SCHIP by expanding Medicaid, then they also increased the income cutoff for
the WIC program. For example, New Jersey, the state that has extended coverage to 3.5 times
3
Some states chose to have completely separate SCHIP programs while others had
separate programs for children of some ages but expanded Medicaid for other children (usually
infants or children not covered by the Medicaid expansions of the late 1980s and early 1990s).
9
poverty, implemented SCHIP by expanding its Medicaid program, so that children with incomes
up to 3.5 times poverty are now eligible for WIC.
It is unlikely that states considered the likely impact on their WIC programs when they
made the choice between stand-alone or Medicaid SCHIP programs. More important factors
discussed by states included ease of implementation and the perception that stigma associated
with Medicaid might lead to eligible children remaining uninsured. Thus, in principal, the
implementation of SCHIP can be viewed as a “natural experiment” that is likely to affect WIC
eligibility without having other effects on families, at least once the overall cutoff for public
health insurance has been controlled. Whether or not SCHIP will have important effects hinges
on the number of children in states whose SCHIP-Medicaid threshold exceeded 1.85 times
poverty and on take up rates among young children in families with higher income levels who
did not use WIC as infants.
III. The Data
The SIPP has been conducted by the Census Bureau since 1984. It consists of a series of
short panels that collect information on income, demographics, and national program
participation. The 1996 panel follows a nationally representative sample of persons from early
1996 through early 2000, while the 2001 panel has followed another group of individuals from
early 2001 to the present. Unlike some other data sets such as the Current Population Survey,
the SIPP follows the original household members when they move.
The SIPP identifies
the individual's state of residence for all but a handful of small states (the states which are not
identified and hence are excluded from our analysis include Maine, North Dakota, South Dakota,
10
and Vermont).
Each panel re-interviews families once every four months. At each interview, a core
questionnaire and various “topical modules” are administered. The core questionnaire asks
about WIC use in each month for all individuals in households that have a woman aged 15 to 45
present. Questions about children’s health were included in topical modules administered in
waves 3, 6, 9, and 12 of the 1996 SIPP and in wave 3 of the 2001 SIPP; weight and height
(which are used to create BMI) were only asked in waves 6 and 12 of the 1996 SIPP. The SIPP
is well-suited for this project because of the large sample, monthly WIC and other program
participation measures, relatively high-quality income data, and the availability of a range of
child outcome measures in the topical modules.
Table 2 provides an overview of the data on WIC participation from the 1996 and 2001
SIPP panels for the subsample of four year old children whose family income is below 3.5 times
the federal poverty level4. We focus on this group in our analysis, since children of higher
income are unlikely to be affected by WIC. Table 2 shows that of the 7,310 four year old
children in this income range, a quarter were participating in WIC at age four.
However, given the structure of the SIPP, there are only a few children who are observed
from infancy until age four. Table 2 shows that of the 7,310 children using WIC at age four,
4,731 were in the sample at age three, but only 464 were also in the sample in infancy. What this
means is that we cannot accurately reconstruct the history of WIC (or Medicaid) participation for
4
One would not expect children at income levels higher than 3.5 times the poverty level
to respond to WIC. For example, over the 1991-2001 period, the highest eligibility cutoff
associated with Medicaid or Medicaid SCHIP was 3 times the poverty level.
11
the majority of four year old children in our sample. Yet, some key outcomes of interest are not
observed for younger children.
The final panel of Table 2 focuses on the small group of children for whom we know past
WIC participation and asks how many of these children began participation as infants. The table
shows that 84 percent of children on WIC at age four were on WIC as infants. Given underreporting of WIC participation in the SIPP, the true fraction may be even higher.5 The final row
of Table 2 shows that of the children on WIC at age four and in the sample of infants, 62% were
on Medicaid as infants, illustrating the tight link between participation in the two programs.
Summary statistics for other variables used in our analysis are shown in Table 3 for
children who were on WIC at age four and those who were not. As discussed above, one
argument that is frequently raised about WIC is that participants may be positively selected
relative to other eligibles, and hence that positive measured impacts of WIC could be associated
with the positive characteristics of WIC participants rather than with participation in the program
per se.
Table 3 demonstrates, that on the contrary, children who participate in WIC at age four
are less well-off than other children, even conditional on being in a family with average in-panel
income less than 3.5 times the federal poverty level. The WIC child’s average family income is
lower than that of non-WIC children (1.06 times the FPL vs. 1.83 times the FPL). Children on
WIC also have mothers who are less educated than those of other children, are more likely to be
5
See Bitler, Currie, and Scholz (2003) for a discussion of undercounting of program
participation in the SIPP and CPS.
12
black or Hispanic, live in larger families, and are more likely to be on welfare than other lowincome children.
Finally, Table 4 shows means for the child outcomes we consider.6 We use the first
available measure in the SIPP, after the child in question turned four. Anthropometric measures
we consider include height, weight, BMI, and measures of whether BMI is high or low.
Height is considered to be a longer-run measure of nutritional status, since it reflects the
cumulative impact of nutrition. Weight is a shorter-run measure. BMI is defined as ((weight in
pounds)/(height in inches)2) x 703. Among children, a BMI less than the 5th percentile for child
age indicates that the child is underweight, while a BMI over the 85th percentile indicates that the
child is at risk for overweight.7
While WIC was originally charged with preventing stunting and wasting (i.e., low
height-for-age and low weight-for-height) in children, as well as with the prevention of micronutrient deficiencies such as anemia, there is growing recognition that many American children
are at risk of obesity. The Surgeon General’s 2001 report on obesity in America shows that the
incidence of obesity is rising faster among children than among adults and warns that obesity
will cause as many deaths as smoking. Moreover, since the poor are at higher risk of being
overweight, obesity’s health consequences will fall disproportionately on those of low income,
6
In addition to the measures we consider here, the SIPP has information about a range of
outcomes such as placement in special education for children aged six to eight. However, there
are relatively few children for whom we have measures both of WIC participation at age four
and one of these other outcome measures. We found that in this small sample, Medicaid income
cutoffs at the time of the birth were no longer statistically significant predictors of WIC
participation at age four, so that our TSLS approach to evaluating the effect of WIC on child
outcomes could not be implemented.
7
See http://www.cdc.gov/nccdphp/dnpa/bmi/bmi-for-age.htm.
13
exacerbating health inequalities. The report emphasizes the importance of early community
intervention to prevent obesity, since once formed, eating habits are hard to change (U.S.
Department of Health and Human Services, 2001).
The prevention of overweight and obesity is now an important goal of WIC. WIC could
reduce the risk of obesity either by providing nutritious foods that substitute for less nutritious
foods that are high in calories, or through nutrition education. It is easy to imagine, for example,
that having the WIC clinic point out to a mother that her child was overweight and at risk for
future health problems could have an impact on her family’s diet.
While WIC focuses on child nutrition, WIC programs also refer clients to medical
providers, so WIC could have an impact on health outcomes. A second reason for examining
measures of health in this study is that we use the generosity of Medicaid at the time of the
child’s birth as an instrument for WIC participation. States that are more generous in their
Medicaid coverage of infants also tend to be more generous in their coverage of older children,
so it is possible that the effects we measure actually reflect past participation in Medicaid rather
than participation in WIC. We will show, however, that we find much larger effects of WIC on
nutritional outcomes than on access to health care, which suggests that what we measure actually
is the effect of the nutritional component of the WIC program.
The measures of health care utilization that we focus on are whether the child saw a
doctor in the past 12 month and whether the child was hospitalized in the past 12 months. We
also present estimates for two measures of health status: whether the child has a condition that
limits the ability to walk, run, or play; and whether the child’s mother-reported health status is
fair or poor (compared to good, very good or excellent).
14
Like Table 3, Table 4 indicates that WIC children are less well off than other children.
Children on WIC at age four are more likely to be above the 85th percentile of BMI for sex and
age; are more than twice as likely as other children to have a condition that limits their ability to
walk, run, or play; and are much more likely to be in fair or poor health. On the other hand,
there is little difference between WIC and non-WIC children in measures of current access to
medical care.
IV. Methods
This report ask two questions. First, how did changes in eligibility for Medicaid/SCHIP
affect the take up of WIC? In order to answer this question, we use the SIPP data to examine
the impact of Medicaid income cutoffs on WIC participation. We estimate OLS regressions that
control for family structure, demographics, the availability of public health insurance in the state,
and state and year fixed effects. We focus on two different Medicaid cutoffs.
The first measure is the Medicaid eligibility cutoff for infants in effect at the time of the
child’s birth. This variable will capture the effect of making it easier to get Medicaid at birth on
future WIC participation. The second measure is the state SCHIP-Medicaid cutoff for young
children, which, as we saw in Table 1, was higher than the WIC cutoff in some states. Our
regression models take the following form:
+
T
+
S
(1) WIC = X γ + MEDICAIDδ1 + PUB_HEALTHδ2 +
5
e
h
t
f
o
e
n
o
s
i
D
I
A
C
I
D
E
M
,
r
u
o
f
e
g
a
t
a
C
I
W
d
e
s
u
d
l
i
h
c
e
h
t
f
i
e
n
o
o
t
l
a
u
q
e
r
o
t
a
c
i
d
n
i
n
a
s
i
C
I
W
e
r
e
h
w
c
i
l
b
u
p
r
o
f
f
f
o
t
u
c
t
n
e
r
r
u
c
e
h
t
s
i
H
T
L
A
E
H
_
B
U
P
d
n
a
,
e
v
o
b
a
d
e
b
i
r
c
s
e
d
s
f
f
o
t
u
c
e
m
o
c
n
i
d
i
a
c
i
d
e
M
o
w
t
d
n
a
t
s
a
s
a
d
e
t
n
e
m
e
l
p
m
i
s
a
w
P
I
H
C
S
f
i
t
a
h
t
e
t
o
N
.
n
e
r
d
l
i
h
c
d
l
o
r
a
e
y
r
u
o
f
r
o
f
e
g
a
r
e
v
o
c
e
c
n
a
r
u
s
n
i
h
t
l
a
e
h
15
y
B
.
f
f
o
t
u
c
e
m
o
c
n
i
d
i
a
c
i
d
e
M
t
n
e
r
r
u
c
e
h
t
m
o
r
f
r
e
f
f
i
d
l
l
i
w
H
T
L
A
E
H
_
B
U
P
n
e
h
t
,
m
a
r
g
o
r
p
e
n
o
l
a
n
a
c
e
w
t
a
h
t
o
s
,
e
c
n
a
r
u
s
n
i
h
t
l
a
e
h
c
i
l
b
u
p
o
t
s
s
e
c
c
a
t
n
e
r
r
u
c
r
o
f
l
o
r
t
n
o
c
e
w
,
H
T
L
A
E
H
_
B
U
P
g
n
i
d
u
l
c
n
i
.
n
o
i
t
a
p
i
c
i
t
r
a
p
C
I
W
d
n
a
s
s
e
c
c
a
e
c
n
a
r
u
s
n
i
h
t
l
a
e
h
c
i
l
b
u
p
t
n
e
r
r
u
c
f
o
s
t
c
e
f
f
e
e
h
t
n
e
e
w
t
e
b
h
s
i
u
g
n
i
t
s
i
d
s
e
d
u
l
c
n
i
X
r
o
t
c
e
v
e
h
T
an indicator equal to one if there is an infant in the family, the
number of other children aged one to four in the family, an indicator equal to one if the child’s
mother is in the household, the mother’s age and age squared, controls for race (black, Asian,
other non-white) and ethnicity (Hispanic), controls for the mother’s completed education (high
school graduate, some college, or four year college degree), an indicator for living in an urban
area, an indicator for the child being male, and controls for child age and whether we observe the
e
r
u
t
p
a
c
h
c
i
h
w
s
t
c
e
f
f
e
d
e
x
i
f
e
t
a
t
s
f
o
r
o
t
c
e
v
a
s
.
i
S
child in the data at ages zero, one, two, or three
h
t
l
a
e
h
d
l
i
h
c
d
n
a
l
a
n
r
e
t
a
m
f
o
t
n
e
t
x
e
e
h
t
s
a
h
c
u
s
,
s
e
t
a
t
s
f
o
s
c
i
t
s
i
r
e
t
c
a
r
a
h
c
t
n
a
t
s
n
o
c
y
l
e
v
i
t
a
l
e
r
d
e
v
r
e
s
b
o
n
u
l
a
i
t
n
e
t
o
p
y
n
a
r
o
f
l
o
r
t
n
o
c
o
t
p
l
e
h
h
c
i
h
w
,
s
t
c
e
f
f
e
d
e
x
i
f
h
t
r
i
b
f
o
r
a
e
y
f
o
r
o
t
c
e
v
a
s
i
T
.
e
r
u
t
c
u
r
t
s
a
r
f
n
i
n
o
t
n
e
i
c
i
f
f
e
o
c
e
h
t
,
s
i
t
s
e
r
e
t
n
i
f
o
t
n
e
i
c
i
f
f
e
o
c
y
e
k
r
u
o
e
r
e
H
.
n
e
r
d
l
i
h
c
f
o
s
t
r
o
h
o
c
n
e
e
w
t
e
b
s
e
c
n
e
r
e
f
f
i
d
δ
.
f
f
o
t
u
c
d
i
a
c
i
d
e
M
s
e
t
a
t
s
e
h
t
’
The second question we address is how WIC participation affects outcomes at age 4, 5,
and 6. These regressions are of form:
Y =
WIC + X α + PUB_HEALTHδ3 +
+
T
+
S
(2)
ε,
where Y is a child outcome, WIC is whether the child was on WIC at age four, and the
independent variables are defined in the same way as above. The key coefficient of interest in
(2) is
. We begin by estimating (2) by OLS, but as discussed above, it is often argued that
WIC participants are a selected sample, so that the OLS estimate of
16
may be biased by
omitted variables.
In order to control for this possibility, we need an instrumental variable that predicts WIC
participation at age four, but is uncorrelated with outcomes, at least once one controls for the
other variables included in the model. The discussion of the close link between Medicaid and
WIC suggest the use of Medicaid income cutoffs as possible instruments. As we show below,
the SCHIP-Medicaid cutoffs had little impact on WIC participation, so they cannot be used as
instrumental variables. Medicaid income cutoffs in the child’s first year of life are predictive of
WIC participation at age four, and so this candidate instrument passes this important test.
However, using the Medicaid income cutoff at the time of the birth as an instrument for
WIC participation requires the strong assumption that these cutoffs have no separate effects on
outcomes once one controls for the other variables, such as PUB_HEALTH. We think the best
way to think about the problem is to recognize that Medicaid and WIC are closely linked. The
figures discussed above suggest that 37% of all U.S. infants are covered by Medicaid at birth,
and that almost the same number of infants, approximately 42%, are on WIC. In fact, all of the
evidence suggests that these are largely the same infants, and given the structure of the SIPP we
cannot identify the small group who participated in one program as infants without participating
in the other.
Changes in the Medicaid income cutoff at the time of the birth cause infants to take up
both Medicaid and WIC. Some fraction of these infants stay on WIC as they age. Thus, children
who were induced to be on WIC at age four by the Medicaid expansions also received Medicaid
at birth as well as prior years of WIC assistance, so that a more accurate way to describe them
would be as “Medicaid-at-birth-plus-continuous-WIC children.”
17
The counter-factual highlighted by our instrument is that in the absence of increases in
the Medicaid income cutoff, these children would not have received Medicaid at the birth or
WIC. Thus, we are comparing Medicaid/WIC children to similar children who did not
participate in these programs. An additional problem, then, is that some children induced to join
Medicaid and WIC by the Medicaid expansions drop WIC before age four. These children are
treated as “no WIC” in our models, so that we understate both the effect of the Medicaid income
cutoffs on overall WIC take up and the effects of WIC use on outcomes. Hence, our estimates
should be interpreted as lower bounds on true WIC/Medicaid at birth effects.
All regressions are weighted to be representative of the population since SIPP oversamples blacks and Hispanics. Standard errors are adjusted for arbitrary within state-by-year
group correlations. Each regression only uses one observation per child (the observation taken
as of the first time the child was at least four and the outcome was available).
V. Results
Table 5 presents OLS estimates of the effects of Medicaid income cutoffs and other
factors on the probability of childhood WIC participation. The first column shows that the
income cutoff in effect at the time of the child’s birth has a strong positive effect on the child’s
probability of being on WIC at age four. In the population with incomes less than 3.5 times
poverty, a doubling of the Medicaid income cutoff (say from poverty to two times poverty)
would be expected to increase the share of children on WIC by 7.5 percentage points.
Household composition also plays an important role. Having an infant in the family
increases the probability of WIC participation by other children in the family by 17.8 percentage
18
points, while having a sibling aged one to four increases the probability by 8.1 percentage points.
These results make sense: families with more children receive larger benefits, so if there is a
fixed cost to maintaining WIC coverage, then families with more children will be more likely to
bear this cost. Race, ethnicity, and maternal age also have positive effects on WIC use, as does
lack of maternal education. Finally, urban mothers are less likely to participate in WIC than
others.
Column 2 of Table 6 shows that in contrast to the cutoff at the time of the birth, higher
Medicaid cutoffs due to SCHIP had very little effect on the use of WIC. Note that in this model,
the variable “Eligibility cutoff for Medicaid SCHIP” is different from zero only if SCHIP
resulted in increases in Medicaid income cutoffs for children. The overall effect of higher public
health insurance cutoffs is captured by the variable PUB_HEALTH. A possible reason for this
negative result is that children who do not participate in WIC as infants are unlikely to be
subsequently enrolled. Hence, making older children eligible for Medicaid increases eligibility
for WIC, but has little effect on enrollments.
Table 6 explores this explanation further by examining the effect of SCHIP on eligibility
for WIC and take up rates. These analyses are conducted by dividing the sample into the groups
that one would and would not expect to be affected by SCHIP. For example, WIC eligibility and
participation among families with incomes less than 2 times the poverty line should not be much
affected by SCHIP, since families with incomes less than 1.85 times poverty were always
eligible for WIC. The new eligibles would be concentrated in the higher income group.
Panel C of Table 6 shows that implementing SCHIP by expanding the Medicaid program
did increase eligibility for WIC in this group. In the SCHIP-Medicaid states, 11% of four year
19
old children in families with incomes between 2 and 3.5 times the poverty line were eligible for
WIC, compared to 4% of families at this income level in other states. Some of these families
reported that they had Medicaid (even though their reported incomes are sometimes higher than
the relevant Medicaid cutoff), so we assumed that they were eligible for Medicaid and therefore
for WIC. It is possible that these families’ incomes had increased since they last renewed their
Medicaid coverage.
Table 6 shows that the higher rate of WIC eligibility in the SCHIP-Medicaid states is
balanced by lower take up of WIC: only 16% of eligibles took up the program, compared to 30%
in the stand alone-SCHIP states. On balance then, implementing SCHIP by expanding Medicaid
resulted in only a very small increase in the fraction of higher income families on the program
(1.8% compared to 1.2% in stand-alone SCHIP states), which suggests that take up in this group
was very low. If we reflect on the fact that the WIC package was only worth $31 per month,
then it may not be surprising that families with incomes over about $30,000 did not find it
worthwhile to participate in the program. This result provides evidence that higher income
families are not driving the increased child participation in WIC, as some critics have argued.
Turning to our second question, Table 7 shows our estimates of equation (2), the effects
of WIC participation at age four on child outcomes. Column (1) shows that in OLS regressions,
WIC use at age four did not have any significant effect on anthropometric outcomes or on the
use of medical care. Consistent with Table 4, children on WIC at age four are more likely than
other children to have a health limiting condition and to have fair or poor health status, even
conditional on the observable characteristics of their families. These results may due to negative
selection into the WIC program. That is, children who are less healthy for other reasons may be
20
more likely to enroll in WIC. In order to control for this possibility, we turn to TSLS models.8
Column (2) shows estimates of TSLS models using the Medicaid income cutoff at the
time of the birth as an instrument. These estimates indicate that WIC participation reduces the
child’s BMI and weight. However, there is no significant effect on the probability that the child
is underweight (less than the 5th percentile of BMI for sex and age). All of the decrease in
weight comes from the other end of the distribution, where we see that the probability that a
child is at risk of being overweight (over the 85th percentile for sex and age) is greatly reduced.
The contrast between the OLS and TSLS results indicates that the WIC children have
unobservable characteristics that place them at higher risk of having high BMI than other
observationally similar children and that WIC is successful in reducing the risk of overweight
among young children. Possible pathways for WIC to affect children’s weight are either through
the WIC program’s provision of healthy foods or through the nutrition education component of
WIC.
Column (2) also shows that in contrast to the OLS models, where WIC had a negative
effect on measures of health limitations and health status, there is no significant effect of WIC
participation in the TSLS models. The contrast between the OLS and TSLS results for these
variables confirm that WIC children are negatively selected in terms of health. They are more
likely than other children to have activity limitations and poor overall health status for reasons
that have little to do with participation in the WIC program, and in fact WIC has little causal
impact on these outcomes. These results confirm that selection into the program is an important
8
The first stage for our TSLS was presented in column 1 of Table 5.
21
issue that must be dealt with in evaluations of WIC.
n
o
i
s
u
l
c
n
o
C
.
I
V
,
s
t
n
a
p
i
c
i
t
r
a
p
C
I
W
d
l
i
h
c
g
n
o
m
a
n
o
i
t
c
e
l
e
s
g
n
i
d
r
a
g
e
r
s
n
o
i
s
u
l
c
n
o
c
l
a
i
t
i
n
i
e
m
o
s
s
r
e
f
f
o
t
r
o
p
e
r
s
i
h
T
e
W
.
s
e
m
o
c
t
u
o
d
l
i
h
c
n
o
C
I
W
f
o
s
t
c
e
f
f
e
e
h
t
d
n
a
,
p
u
e
k
a
t
C
I
W
n
o
m
a
r
g
o
r
p
d
i
a
c
i
d
e
M
e
h
t
f
o
s
t
c
e
f
f
e
e
h
t
l
l
a
o
t
e
v
i
t
a
l
e
r
d
e
t
c
e
l
e
s
y
l
e
v
i
t
a
g
e
n
e
r
a
C
I
W
n
o
n
e
r
d
l
i
h
c
,
C
I
W
n
o
s
t
n
a
f
n
i
d
n
a
n
e
m
o
w
e
k
i
l
t
a
h
t
,
t
s
r
i
f
,
d
n
i
f
g
n
i
t
i
m
i
l
h
t
l
a
e
h
e
v
a
h
o
t
d
n
a
h
t
l
a
e
h
r
o
o
p
n
i
e
b
o
t
y
l
e
k
i
l
e
r
o
m
e
r
a
y
e
h
t
,
r
a
l
u
c
i
t
r
a
p
n
I
.
s
e
l
b
i
g
i
l
e
t
i
e
c
n
i
s
,
m
a
r
g
o
r
p
C
I
W
e
h
t
f
o
s
n
o
i
t
a
u
l
a
v
e
e
r
u
t
u
f
n
i
d
n
i
m
n
i
t
p
e
k
e
b
d
l
u
o
h
s
n
o
i
t
c
e
l
e
s
s
i
h
T
.
s
n
o
i
t
i
d
n
o
c
y
l
r
e
p
o
r
p
o
t
l
i
a
f
s
t
s
y
l
a
n
a
f
i
d
e
k
o
o
l
r
e
v
o
e
b
y
a
m
m
a
r
g
o
r
p
e
h
t
f
o
s
t
c
e
f
f
e
l
a
s
u
a
c
e
v
i
t
i
s
o
p
t
a
h
t
s
e
i
l
p
m
i
.
n
o
i
t
c
e
l
e
s
r
o
f
l
o
r
t
n
o
c
t
a
h
t
o
s
,
s
t
n
a
f
n
i
g
n
o
m
a
d
e
k
n
i
l
y
l
e
s
o
l
c
y
r
e
v
e
r
a
e
s
u
C
I
W
d
n
a
d
i
a
c
i
d
e
M
t
a
h
t
w
o
h
s
e
w
,
y
l
d
n
o
c
e
S
n
o
s
t
c
e
f
f
e
m
r
e
t
g
n
o
l
e
v
a
h
o
t
y
l
e
k
i
l
e
r
a
s
t
n
a
f
n
i
g
n
o
m
a
p
u
e
k
a
t
t
c
e
f
f
a
t
a
h
t
s
e
i
c
i
l
o
p
d
i
a
c
i
d
e
M
t
c
e
f
f
e
n
i
e
r
e
w
t
a
h
t
d
i
a
c
i
d
e
M
r
o
f
s
f
f
o
t
u
c
e
m
o
c
n
i
e
h
t
,
r
a
l
u
c
i
t
r
a
p
n
I
.
m
a
r
g
o
r
p
C
I
W
e
h
t
n
i
n
o
i
t
a
p
i
c
i
t
r
a
p
,
t
s
a
r
t
n
o
c
y
B
.
r
e
t
a
l
s
r
a
e
y
r
u
o
f
n
o
i
t
a
p
i
c
i
t
r
a
p
C
I
W
f
o
y
t
i
l
i
b
a
b
o
r
p
e
h
t
s
t
c
e
f
f
a
n
r
o
b
s
a
w
d
l
i
h
c
a
n
e
h
w
t
u
o
h
t
i
w
y
t
i
l
i
b
i
g
i
l
e
C
I
W
d
e
s
a
e
r
c
n
i
n
e
r
d
l
i
h
c
r
e
d
l
o
s
d
r
a
w
o
t
d
i
a
c
i
d
e
M
f
o
y
t
i
s
o
r
e
n
e
g
e
h
t
n
i
s
e
s
a
e
r
c
n
i
f
i
n
e
r
d
l
i
h
c
e
l
b
i
g
i
l
e
l
l
o
r
n
e
o
t
t
l
u
c
i
f
f
i
d
y
l
e
m
e
r
t
x
e
e
b
o
t
s
r
a
e
p
p
a
t
I
.
n
o
i
t
a
p
i
c
i
t
r
a
p
n
o
t
c
a
p
m
i
h
c
u
m
g
n
i
v
a
h
s
e
s
a
e
r
c
n
i
t
a
h
t
s
e
t
a
c
i
d
n
i
o
s
l
a
t
l
u
s
e
r
s
i
h
T
.
d
e
s
s
i
m
s
i
y
c
n
a
f
n
i
g
n
i
r
u
d
t
n
e
m
l
l
o
r
n
e
r
o
f
w
o
d
n
i
w
l
a
u
s
u
e
h
t
”
“
m
o
r
f
n
e
r
d
l
i
h
c
y
b
n
o
i
t
a
p
i
c
i
t
r
a
p
d
e
s
a
e
r
c
n
i
y
b
n
e
v
i
r
d
n
e
e
b
t
o
n
e
v
a
h
n
o
i
t
a
p
i
c
i
t
r
a
p
C
I
W
s
n
e
r
d
l
i
h
c
n
i
’
.
d
e
u
g
r
a
e
v
a
h
s
c
i
t
i
r
c
e
m
o
s
s
a
,
P
I
H
C
S
f
o
t
l
u
s
e
r
a
s
a
e
l
b
i
g
i
l
e
e
d
a
m
s
e
i
l
i
m
a
f
e
m
o
c
n
i
r
e
h
g
i
h
s
t
c
e
f
f
e
t
n
a
c
i
f
i
n
g
i
s
d
n
a
e
g
r
a
l
s
a
h
r
u
o
f
e
g
a
t
a
n
o
i
t
a
p
i
c
i
t
r
a
p
C
I
W
t
a
h
t
s
i
g
n
i
d
n
i
f
g
n
i
k
i
r
t
s
t
s
o
m
r
u
O
22
r
o
f
e
l
i
t
n
e
c
r
e
p
h
t
5
8
e
h
t
n
a
h
t
r
e
t
a
e
r
g
I
M
B
d
a
h
.
e
.
i
(
t
h
g
i
e
w
r
e
v
o
f
o
k
s
i
r
t
a
s
i
d
l
i
h
c
a
t
a
h
t
y
t
i
l
i
b
a
b
o
r
p
e
h
t
n
o
n
o
i
t
i
r
t
u
n
e
h
t
r
e
h
t
i
e
t
a
h
t
s
t
s
e
g
g
u
s
t
I
.
s
f
f
o
t
u
c
I
M
B
t
n
e
r
e
f
f
i
d
g
n
i
s
u
o
t
t
s
u
b
o
r
s
i
t
l
u
s
e
r
s
i
h
T
.
)
e
g
a
d
n
a
x
e
s
n
o
i
t
n
e
v
e
r
p
e
h
t
n
o
t
c
e
f
f
e
t
n
a
t
r
o
p
m
i
n
a
g
n
i
v
a
h
e
b
y
a
m
d
o
o
f
y
h
t
l
a
e
h
f
o
n
o
i
s
i
v
o
r
p
l
a
u
t
c
a
e
h
t
r
o
n
o
i
t
a
c
u
d
e
m
a
r
g
o
r
p
C
I
W
e
h
t
f
o
s
s
e
c
c
u
s
e
h
t
f
o
e
r
u
s
a
e
m
t
n
a
t
r
o
p
m
i
n
a
s
i
s
i
h
T
.
n
e
r
d
l
i
h
c
g
n
u
o
y
g
n
o
m
a
y
t
i
s
e
b
o
f
o
f
o
e
c
n
a
t
r
o
p
m
i
e
h
t
f
o
e
s
u
a
c
e
b
d
n
a
,
t
a
e
r
h
t
h
t
l
a
e
h
c
i
l
b
u
p
a
s
a
y
t
i
s
e
b
o
f
o
e
c
n
a
t
r
o
p
m
i
e
h
t
f
o
e
s
u
a
c
e
b
.
e
f
i
l
n
i
y
l
r
a
e
s
t
i
b
a
h
g
n
i
t
a
e
y
h
t
l
a
e
h
g
n
i
h
s
i
l
b
a
t
s
e
d
n
a
d
i
a
c
i
d
e
M
n
e
e
w
t
e
b
k
n
i
l
e
s
o
l
c
y
r
e
v
e
h
t
o
t
e
u
d
t
a
h
t
s
i
h
c
r
a
e
s
e
r
r
u
o
o
t
t
a
e
v
a
c
t
n
a
t
r
o
p
m
i
n
A
e
h
T
.
d
o
o
h
d
l
i
h
c
g
n
i
r
u
d
C
I
W
f
o
t
c
e
f
f
e
e
h
t
e
t
a
l
o
s
i
o
t
t
l
u
c
i
f
f
i
d
s
i
t
i
,
y
c
n
a
f
n
i
g
n
i
r
u
d
n
o
i
t
a
p
i
c
i
t
r
a
p
C
I
W
e
w
t
a
h
t
s
t
s
e
g
g
u
s
y
l
g
n
o
r
t
s
e
r
a
c
l
a
c
i
d
e
m
o
t
s
s
e
c
c
a
n
a
h
t
r
e
h
t
a
r
s
e
m
o
c
t
u
o
l
a
n
o
i
t
i
r
t
u
n
s
t
c
e
f
f
a
C
I
W
t
a
h
t
t
c
a
f
o
t
y
a
w
a
d
n
i
f
o
t
k
r
o
w
e
r
u
t
u
f
r
o
f
t
n
a
t
r
o
p
m
i
e
b
l
l
i
w
t
i
,
r
e
v
e
w
o
H
.
t
c
e
f
f
e
C
I
W
a
g
n
i
r
u
s
a
e
m
t
c
a
f
n
i
e
r
a
t
c
a
p
m
i
e
h
t
f
o
t
e
n
,
d
o
o
h
d
l
i
h
c
g
n
i
r
u
d
n
o
i
t
a
p
i
c
i
t
r
a
p
C
I
W
f
o
t
c
e
f
f
e
l
a
n
i
g
r
a
m
e
h
t
f
o
e
r
u
t
c
i
p
r
e
r
a
e
l
c
a
d
l
e
i
y
.
y
c
n
a
f
n
i
g
n
i
r
u
d
e
s
u
C
I
W
d
n
a
d
i
a
c
i
d
e
M
f
o
23
s
e
c
n
e
r
e
f
e
R
n
i
w
l
A
d
n
a
;
e
l
l
i
v
o
C
.
R
m
a
i
l
l
i
W
;
d
e
w
a
r
t
S
r
e
m
m
u
l
G
e
c
n
e
r
u
a
L
;
n
a
g
o
H
.
K
a
y
a
j
i
V
;
.
B
u
d
n
I
,
a
i
l
a
w
u
l
h
A
.
2
9
9
1
,
n
a
g
i
h
c
i
M
:
s
h
t
r
i
B
e8
.
g
7
A4
l
a
7
n
3
o
1
i
t
.
a
p
t
p
s
e
,
8
G9
r
9
o
1
f
r
l
e
l
a
b
mm
e
St
p
ne
o
S
n,
o
)
i
9
t(
a
8
p
8
i
c
.
h
i
t
t
r
l
a
a
e
P
H
Cc
I
i
Wb
l
u
fP
o
tf
c
o
e
f
l
f
a
E
n
r
eu
h
o
TJ
n
.c
a
n
i
o
r
s
e
r
e
m
t
e
A
P
”
“
d
n
a
,
s
t
n
a
f
n
I
,
n
e
m
o
W
e
h
t
f.
o
1
n0
o
0
i
2
t
a
)
u
e
l
t
a
u
vi
t
Es
t
n
nI
Ae
:i
s
C
r
I
p
r
We
t
gn
n
E
i
k
n
n
a
i
c
h
i
t
r
e
e
Rm
A
.:
s
i
.
n
C
a
.
mD
rn
e
o
Gt
g
rn
e
t
i
e
h
P
s
a
dW
n
(
a
,,
s
a
m
l
g
a
r
u
g
o
o
Dr
,P
v
n
o
e
r
r
a
d
h
l
i
s
h
e
BC
“
”
h
t
r
i
B
d
n
a
y
c
n
a
n
g
e
r
P
n
o
C
I
W
f
o
s4
.
t
c
0
e
0
f
f
2
,
E
t
n
ee
h
m
Te
?g
k
a
r
n
o
a
WM
Cd
I
n
Wa
y
sc
e
i
l
o
o
DP
f
.o
e
i
l
r
a
r
n
u
r
Cu
to
J
e
n
,
a
g
J
n
di
n
m
a
o
.c
P
h
t
er
n
o
n
F
a
i
r
.
a
s
e
Mm
,o
r
c
e
t
l
t
u
i
O
B
“
”
n
o
i
t
c
e
l
e
S
f
:o
h
t
t
l
n
a
e
e
Hm
t
ta
r
n
a
p
f
e
n
D
I
di
a
n
n
a
i
g
nr
o
i
i
V
t
a
f
p
o
i
c
y
i
t
t
r
i
a
s
r
P
e
v
Ci
I
n
WU
,
lt
a
p
t
i
a
r
n
c
e
s
r
u
P
n
a
.M
n
n
d
a
e
wh
s
Si
l
.b
A
u
rn
p
e
h
U
p
o
t
.
s
s
i
t
r
c
h
e
Cf
.
f
1
dE
0
n
0
d2
a
.ey
x
J
i
a
lF
e
M
l
a
a
h
,
n
s
c
r
c
i
e
i
Ma
t
m
,M
o
n
n
e
o
d
i
c
r
n
B
E
a
d
n
T
a
B
,A
e
c
i
:
r
A
P
rM
e
f
e
o
g
t
d
s
i
i
r
r
b
C
m
,a
m
aC
(
L
,
nn
e
e
Kr
d
,i
l
d
l
h
e
C
f
n
r
C
o
I
KW
tf
r
e
o
b
e
o
l
i
Rf
o
,r
r
e
P
l
l
:
i
t
c
H
e
j
.o
B
r
nP
a
s
d
i
s
r
y
o
l
J
a
,n
x
A
o
F
l
a
ye
r
a
n
Ke
yG
r
a
C
MI
.
,W0
0
y
c
0
.
n
2
a
a
)
Nds
d
e
o
,R
t
n
a
i
i
e
c
do
t
i
s
v
r
s
a
s
u
BDA
”
“
d
n
a
n
o
i
t
.
a
i
2
t
0
i
n
6
I
4
e2
h
1
t
.
do
n
n
a
r
ne
o
p
i
a
t
a
P
p
i
n
c
o
i
i
t
s
r
s
a
u
P
c
Cs
i
I
D
Wy
t
r
.e
b
v
e
o
DP
.n
N
o
dh
n
c
a
r
a
;e
n
s
a
e
wR
ar
h
o
Df
it
e
m
u
i
t
S
i
t
s
;n
k
I
c
u
n
.
g
o
n
Bi
d
ne
e
e
r
f
a
t
Ks
;a.
e
a
r
2
k
B
0
n
i
0
f
P
2
o
,n
y
i
j
r
o
r
a
i
e
tr
u
t
a
t
r
a
b
u
h
e
CDF
f
o
sn
t
c
o
e
i
t
f
i
f
r
t
E
u
eN
h
Td
n
:a
s
e
d
t
a
o
t
o
S
F
ee
v
r
i
u
F
t
l
nu
i
c
i
sr
n
g
r
o
A
b
f
wo
et
Ne
n
dm
i
t
a
r
c
a
i
p
d
e
e
MD
.
g.
S
n
U
o
m:
aA
tV
h
,
g
a
i
i
e
r
wd
n
ha
t
r
x
i
e
B
l
wA
o.
Ln
o
yi
r
t
e
a
p
Vi
c
i
.r
t
a
r
a
a
P
b
r
C
a
BI
,W
y
l
e
a
t
n
a
a
n
v
e
e
r
DP
.
2
9
9
1
h
c
r
a
M
.
n
o
i
t
a
u
l
a
v
E
d
n
a
s
i
s
y
l
a
n
A
f
o
e
c
i
f
f
O
,
e
c
i
v
r
e
S
ed
h
n
Ta
s
:i
s
s
e
y
ma
l
on
c
t
A
u
y
Oc
i
hl
t
o
r
i
P
B
f
do
n
l
a
a
n
sr
t
u
s
o
o
CJ
d.
i
e
a
r
c
a
i
d
C
e
l
Mt
a
a
n
.e
e
r
r
o
P
h
f
c
o
S
e
rs
e
f
U
i
n
e
n
h
.
e
t
2
J
d
9
dn
n
a
3
a
7
n
5
;
ro.
i
e
t
p
ma
p
p
i
i
,
e
c
2
h
i
9
l
t
i
r
9
a
B
1
P
aC
l
l
d
a
n
I
F
i
W
L
,
)
;l(
4
a
a
t
r
1
a
a
1
n
b
.
t
r
e
n
a
r
e
BP
m
fe
,o
y
g
e
sa
n
t
c
n
a
e
a
v
f
e
f
M
DE
e
h
Td
:n
s
a
e
t
a
d
t
o
S
o
eF
v
e
i
r
F
u
t
nl
i
u
si
c
n
r
r
g
o
A
b
w
o
ef
Nn
t
de
i
m
a
t
c
r
i
a
d
p
e
e
.
MD
3
9
g.9
S
n
1
.
o
y
mUa
a:M
yA.
t
V
i
n
l
,
a
o
a
t
i
i
r
t
r
a
o
d
u
Mn
l
a
a
tx
v
n
e
E
a
l
f
A
d
n
n
I
a
.s
.n
i
o
m
s
i
rt
y
i
a
l
h
a
p
c
i
n
c
S
A
i
t
nr
f
a
e
o
l
P
l
e
A
C
c
i
I
dW
f
f
n
O
a
l
,
aa
e
t
r
a
c
a
i
n
b
v
r
e
r
a
r
e
BP
S
fn
,o
y
o
e
st
i
n
t
i
c
a
r
e
v
t
f
u
e
f
N
DE
24
l
a
n
r
u
o
J
.
n
o
i
t
a
p
i
c
i
t
r
a
P
d
n
a
y
t
i
l
i
b
i
g
i
l
E
C
I
W
.
z
l
o
h
c
S
l
r
a
K.
9
n7
h
9
o
J
3
1
d1
n
a
,
)
;S
e
(
i
8
r
r
3
a
,
C3
0
t0
e
n
2
a
s
.
J
e
;r
c
.
P
u
es
o
n
e
n
R
a
i
r
n
a
a
Mm
,u
r
H
e
l
t
f
i
o
B
“
”
”
“
“
”
“
”
“
”
“
”
e
c
n
a
t
s
i
s
s
A
d
o
o
F
C
I
W
e
h
t
n
i
n
o
i
t
a
p
i
c
i
t
r
a
P
f
o
s
t
c
e
f
f
E
.
n
a
c
n
u
D
.
J
g
e
r
G
d
n
a
i
r
o
L
,
s
e
n
o
J
i
k
s
e
l
a
w
o
K
r
e
p
a
P
n
o
i
s
s
u
c
s
i
D
y
t
r
e
v
o
P
n
o
h
c
r
a
e
s
e
R
r
o
f
e
t
u
t
i
t
s
n
I
.
t
n
e
m
p
o
l
e
v
e
D
d
n
a
h
t
l
a
e
H
s
'
n
e
r
d
l
i
h
C
n
o
m
a
r
g
o
r
P
“
.
0
0
0
2
.
0
0
7
0
2
1
.
o
N
”
h
t
a
e
D
f
o
s
e
s
u
a
C
l
a
u
t
c
A
.
g
n
i
d
r
e
b
r
e
G.
.5
L
4
e8
i
l
3
u
2
J
1
dp
.
n
a
p
,4
,
p
u
0
o
0
r
2
t
S
,
)
.0
F
1
(
a1
n
9
n
2
o
.
DA
,M
s
k
A
r
J
a
M
.
.0
S
0
0
s2
e
ms
,
ae
J
t
a
,t
.
S
H
d
ie
l
A
t
i
,n
d
U
a
d
e
k
h
t
o
n
Mi
e
h
t
n.
i
2
y6
t
i
3
l
1
a
t
r
4
o
5
3
M1
t.
n
p
a
p
f
n
,
I
8
9
n9
o
1
dr
i
e
a
b
c
i
m
d
e
e
t
Me
p
dS
n
,
a
)
9
C8
(
I
8
Wh
.
fl
t
o
a
te
c
H
e
f
c
f
i
E
l
b
eu
h
P
Tf
o
.l
r
a
e
n
v
r
r
u
a
o
CJ
nn
e
a
r
c
a
i
r
Ke
dm
n
A
a
.
E
.
s
ye
c
t
n
a
t
a
NS
d
,e
s
t
s
i
o
n
MU
e
n
u
J
)
en
mo
ot
i
c
a
n
i
I
c
o
w
s
os
LA
rs
o
r
f
o
er
n
g
e
a
r
v
e
o
v
G
o
l
Ca
n
ho
t
l
i
aa
t
e
HN
e:
.
t
C
a
t
.
S
D
:n
2
o
0
t
0
g
2
n
i
eh
t
s
a
a
d
p
W
U(
Hs
,
Cn
t
Me
r
a
.P
n
o
d
i
n
t
a
a
i
c
n
o
e
r
s
s
d
l
Ah
i
sC
r,
o
n
n
e
r
e
m
v
o
o
GW
lt3
.
n
a
a
n
0
n
o
0
i
g
t
2
e
a
r
,
NP
9
“
’
’
”
e
c
i
v
dr
n
e
a
S
sh
e
c
c
i
r
r
a
P
e
s
ae
l
R
u
mi
c
rm
o
F
o
n
to
n
a
c
f
E
n
I
e
r
u
.t
o
l
e
u
z
c
a
i
r
r
F
g
hA
t
f
e
o
b
a
t
n
z
i
e
l
m
E
t
r
da
n
p
a
e
,D
d
o
.
o
S
.
wU
l
l
a
.
ms
s
Se
r
dg
i
v
n
a
o
DC
,o
l
l
t
e
t
r
r
P
o
p
ke
r
a
R
Ma
l
,i
n
r
o
F
t
c
:
i
y
V
t
i
l
,b
i
a
r
a
i
l
e
i
v
a
i
v
l
A
O
”
.
1
0
0
2
r
e
b
o
t
c
O
.
1
0
0
2
N
A
F
E
.
1
t0
n
0
e
2
v
e
)
r
l
P
a
r
e
on
t
e
nG
o
i
t
n
c
o
Ae
g
or
t
u
lS
l
a
e
Ch
t
sf
lo
a
r
e
e
c
n
i
f
e
f
GO
n.
:
o
D
e
.
g
r
M
u
S
e
l
l
e
i
hv
Tc
k
.o
s
R
e
(
c
i
1
v
0
r
e
0
S
2
,
ny
a
t
i
ms
e
ub
HO
dd
n
n
a
a
ht
t
l
h
a
g
e
i
e
Hw
fr
o
e
tv
n
O
e
ms
e
t
r
a
a
e
r
p
c
e
e
DD
.d
S
.
n
U
a
’
25
c
i
l
b
u
P
f
o
l
na
o
n
r
mu
ao
r
J
g
n
o
r
a
P
c
i
r
Ce
I
m
WA
e
h
t
.
h
nt
i
u
no
o
Y
i
t
f
a
o
p
i
y
c
i
e
t
v
r
r
a
u
P
S
fl
o
a
si
n
t
c
d
e
u
f
t
f
i
E
g
n
o
.L
n
a
l
c
a
n
n
u
o
Di
t
.a
J
.
N
4
g
e
0
e
h
8
r
t
G
9
9
dm7
o
n
r
a
.
f
p
i
e
p
r
c
o
,
Ln0
2
e
,d
0
s
i
e
v
2
n
E)
,
o
:
J
5
t(
h
i
2
g
k
9
i
s
e
e
.
h
l
w
t
a
l
ha
wr
t
e
oi
H
KB
“
”
”
“
“
”
“
Tables, Final USDA Report
May 28, 2004
Table 1: Variation in Medicaid and separate SCHIP thresholds and Medicaid eligibility threshold while an
infant for children aged 0–4
State/Year
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
DC
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
Tennessee
Texas
Utah
Virginia
Washington
West Virginia
Wisconsin
Wyoming
1996
1.33
1.33
1.36
1.34
1.58
1.33
1.74
1.52
1.52
1.52
1.52
1.52
1.33
1.33
1.39
1.52
1.39
1.52
1.33
1.74
1.53
1.54
2.08
1.52
1.52
1.33
1.39
1.33
1.72
1.85
1.85
1.85
1.85
1.33
1.59
1.33
1.79
2.22
1.85
1.85
1.85
1.33
1.33
1.94
1.50
1.66
1.33
Medicaid as an infant
2002
1.33
1.91
1.40
2.00
2.00
1.33
1.85
1.87
1.99
1.97
1.85
1.94
1.50
1.97
1.50
1.93
1.50
1.85
1.70
2.00
2.00
1.85
2.80
1.85
2.94
1.33
1.85
1.33
2.94
1.85
2.33
1.85
1.85
1.80
1.85
1.33
1.85
2.50
1.85
1.85
1.85
1.33
1.33
2.00
1.50
1.95
1.33
Medicaid SCHIP
2002
2.00
2.00
–
2.00
2.50
–
3.00
–
2.00
1.99
–
2.00
1.50
1.87
2.00
1.99
–
2.00
2.00
3.00
1.99
2.00
2.80
2.00
3.00
–
1.85
–
2.98
3.50
2.35
2.50
–
2.00
1.85
–
–
2.50
1.50
2.00
2.00
–
–
–
–
2.00
–
Separate SCHIP
2002
–
–
2.00
–
–
1.85
–
2.00
–
–
2.35
–
–
–
–
–
2.00
–
–
–
–
–
–
–
–
1.50
–
2.00
–
–
–
–
2.00
–
–
1.70
2.35
–
–
–
–
2.00
2.00
2.50
2.00
–
–
Table shows means in eligibility thresholds for SCHIP and Medicaid for children aged 0–4 across states in the 1996
and 2002 SIPP. Thresholds expressed as percent of the FPL. Columns 1 and 2 show the state level Medicaid eligibility
threshold for children 0–4 as infants. Column 3 shows the maximum Medicaid SCHIP eligibility threshold for 2002
conditional on your state having a Medicaid SCHIP program. Column 4 shows the maximum separate SCHIP
eligibility threshold for 2002 conditional on your state having a separate SCHIP program. No states had SCHIP
programs in 1996. Statistics are weighted by the population of children 0–4 in the state.
Table 2: Summary statistics for WIC use, 1996 and 2001 SIPP, average family income <350% of FPL
Number in sample at age 4
7310
Number on WIC while 4
1822
Conditional on being 4
Number in sample at 3
4731
Number in sample at 2
2659
Number in sample at 1
1270
Number in sample as infant
464
Number on WIC while 4 and in sample as infant
102
Conditional on being on being 4, on WIC at 4, and in the sample as an infant
On WIC as infant
0.84
On Medicaid as infant
0.62
Table contains either the number of observations or means for children with average family income
< 3.5 times the federal poverty level from the 1996 and 2001 SIPP. Statistics are weighted. Row 1
is the number of children who were 4 at some point in the sample. Row 2 is the number of 4 year
olds on WIC while 4. Rows 3–6 are the number of 4 year olds who were in the sample at ages 3, 2,
1, or 0 respectively. Rows 7 is the number of 4 year olds on WIC while 4 who were in the sample
as infants. Row 8 is the share of children who were in the sample as infants and at 4, were on WIC
at 4, and who got WIC while aged 0. Row 9 is the share of children who were in the sample as
infants and at 4, were on WIC at 4, and who got Medicaid while aged 0.
Table 3: Summary statistics for other variables, 1996 and 2001 SIPP, average family income < 350% of
FPL
WIC while 4
0.56
No WIC while 4
0.20
On AFDC/TANF now
0.17
0.06
Average family income to needs ratio
1.06
1.83
Elig. cutoff for Medicaid while an infant, share of FPL
1.69
1.65
Elig. cutoff for Medicaid SCHIP, share of FPL
0.88
0.86
Elig. cutoff for public HI, share of FPL
1.80
1.79
An infant in the family
0.18
0.08
Number of other children 1–4 in family
0.50
0.33
Black
0.26
0.18
Hispanic
0.33
0.19
Asian
0.02
0.04
Other non-white
0.02
0.02
Mother absent, age/education missing
0.06
0.07
Father absent from household
0.43
0.30
27
28
Urban
0.76
0.80
Mother is high school dropout
0.31
0.15
Mother is high school grad., no college
0.34
0.32
Mother has some college, no 4 year degree
0.23
0.30
Mother is 4 year college graduate
0.04
0.13
Male child
0.52
0.51
On Medicaid now
Mother’s age (=0 if no mother in HH)
N
1822
5488
Table contains means (standard deviations) for children with average family income 3.5 times the federal
poverty level from the 1996 and 2001 SIPP. Outcomes for children aged 4 in first wave they are at least 4.
Statistics are weighted.
Table 4: Summary statistics for children’s outcomes, 1996 and 2001 SIPP, family income < 350% of FPL
WIC while 4
Anthropomorphic measures (measured at ages 4 or 5)
Child’s Body Mass Index
17.9
No WIC while 4
N total
17.7
3399
Child’s weight (lbs)
45
44
3399
Child’s height (inches)
42
42
3399
BMI < 5th percentile for sex and age
0.06
0.05
3399
BMI > 85th percentile for sex and age
0.17
0.14
3399
Health care utilization (measured at ages 4 or 5)
Saw a doctor last year
0.66
0.69
5854
In the hospital last year
0.03
0.03
5854
Health status
Condition limits ability to walk, run, or play
0.07
0.03
3507
(measured at ages 4 and 5)
Health status fair/poor
0.05
0.03
5854
(measured at ages 4, 5, or 6)
Table contains means (standard deviations) for children with income < 3.5 times the federal
poverty level from the 1996 and 2001 SIPP. Outcome for children first survey wave that they are
at least 4 and relevant outcome is reported. Statistics are weighted.
Table 5: OLS regression of determinants of WIC use while 4, 1996 and 2001 SIPP, average family income
< 350% of FPL
Elig. cutoff for Medicaid while an infant, share of FPL
Elig. cutoff for Medicaid SCHIP, share of FPL
0.075∗∗∗
(0.021)
0.014
(0.009)
Elig. cutoff for public HI, share of FPL
-0.005
-0.008
(0.020)
(0.024)
An infant in the family
0.178∗∗∗
0.177∗∗∗
(0.021)
(0.021)
Number of other children 1-4 in fam.
0.081∗∗∗
0.081∗∗∗
(0.012)
(0.012)
Male child
0.004
0.004
(0.011)
(0.011)
Black
0.129∗∗∗
0.129∗∗∗
(0.017)
(0.017)
Hispanic
0.124∗∗∗
0.125∗∗∗
(0.019)
(0.020)
Asian
-0.025
-0.024
(0.028)
(0.028)
Other non-white
0.088∗
0.087∗
(0.049)
(0.049)
Mother absent, age/education missing
-0.039
-0.040
(0.034)
(0.034)
Mother’s age (=0 if no mother in HH)
0.008∗∗∗
0.008∗∗∗
(0.002)
(0.002)
Mother’s age squared
-0.000164∗∗∗
-0.000164∗∗∗
(0.000044)
(0.000044)
Urban
-0.061∗∗∗
-0.061∗∗∗
(0.016)
(0.016)
∗∗∗
Mother is high school grad., no college
-0.095
-0.095∗∗∗
(0.019)
(0.019)
∗∗∗
Mother has some college, no 4 year degree
-0.141
-0.141∗∗∗
(0.021)
(0.021)
Mother is 4 year college graduate
-0.237∗∗∗
-0.237∗∗∗
(0.024)
(0.024)
N
7310
7310
2
Adjusted R
0.184
0.109
Coefficients (SEs) from OLS regressions of the determinants of WIC use while aged 4 for
children with average family income 3.5 times the federal poverty level in the first survey
wave they turn 4. Each column contains the results of one regression. State and year of
birth fixed effects and controls for the child being in the sample at 0, 1, 2, or 3 years of age
also included. Regression is weighted, and SEs clustered for state and year. ∗∗∗ , ∗∗ , and ∗
denote significance at the 1, 5, and 10% levels.
Table 6: Variation in WIC eligibility and takeup by family income and the type of SCHIP expansion in
place for children aged 4 in 1996 and 2001 SIPP
State has:
No SCHIP
Panel A: Child’s family income < 133% FPL
Any WIC eligibility
1.00
N
4058
Medicaid SCHIP
Stand-alone SCHIP
1.00
800
1.00
2858
0.32
800
0.30
2858
Panel B: 133% FPL ≤ child’s family income < 200% FPL
Any WIC eligibility
0.85
N
1762
0.91
380
0.81
1257
WIC takeup rate for eligibles
N
0.17
345
0.17
1028
Panel C: 200% FPL ≤ child’s family income < 350% FPL
Any WIC eligibility
0.04
N
5130
0.11
1078
0.04
4409
WIC takeup rate for eligibles
N
0.16
118
0.30
215
WIC takeup rate for eligibles
N
0.31
4058
0.20
1500
0.31
239
Panel A shows the variation in WIC eligibility and takeup for children aged 4 with family income less
than 133% of the federal poverty level. Panel B shows the variation in WIC eligibility and takeup for
children aged 4 in families with income between 133% and 200% of the federal poverty level from the 1996
and 2001 SIPP. Panel C shows the variation in WIC eligibility and takeup for children aged 4 in families
with income between 200% and 350% of the federal poverty level from the 1996 and 2001 SIPP. Column
1 contains means for children in states without any SCHIP, column 2 means for children in states which
have implemented Medicaid SCHIP, and column 3 means for children in states with stand-alone SCHIP.
Table 7: OLS and TSLS estimates of the impact of WIC use at 4, 1996 and 2001 SIPP, average family
income < 350% of FPL
Coefficient on any WIC while 4
N
Child’s Body Mass Index
OLS
TSLS
OLS/TSLS
Anthropomorphics (measured at ages 4 or 5)
-0.0610
-8.4157∗∗
3399
(0.2133)
(3.6655)
Child’s weight (lbs)
0.1806
(0.4922)
-19.7435∗∗
(9.8184)
3399
Child’s height (inches)
0.1847
(0.1917)
-1.6189
(3.5681)
3399
BMI < 5th percentile for sex and age
0.0015
(0.0103)
0.1003
(0.1928)
3399
BMI > 85th percentile for sex and age
0.0099
(0.0156)
-0.6905∗∗
(0.3058)
3399
Saw a doctor last year
In the hospital last year
Health care utilization (measured at ages 4 or 5)
0.0163
0.7340
5854
(0.0152)
(0.4939)
0.0043
(0.0069)
0.0556
(0.1692)
5854
Condition limiting ability to walk, run, or play
(measured at ages 4 or 5)
Health status
0.0342∗∗∗
(0.0098)
0.1476
(0.1398)
3507
Health status fair/poor
(measured at ages 4, 5, or 6)
0.0249∗∗∗
(0.0086)
0.2329
(0.1988)
5854
Table contains coefficient on any WIC use while 4 from OLS, TSLS, and TSIV regressions of the determinants
of various health outcomes (columns 1 and 2). Columns 3 contains sample sizes. Sample is all children with
income < 3.5 times the federal poverty level from the 1996 and 2001 SIPP. Outcomes are for children in the
first survey wave they are at least 4 and questions were asked unless otherwise noted. Columns 1 and 2 are
estimated on the sample for which any WIC use while 4 is known. Regressions also control for the cutoff at
which eligibility for public health insurance ends, whether there is an infant in the family, the number of other
children 1–4 in the family, mother’s age and age squared, indicators for the child’s race (black, Asian, other), age,
Hispanic ethnicity, education (high school graduate, some college, four year college degree, and education missing),
urban residence, the child being male, and controls for the child being in the data at 0, 1, 2, or 3 years of age.
Instrument for first stage of TSLS is the average Medicaid eligibility threshold in the child’s first year. All regressions
also include state, age, and year-of-birth fixed effects. All estimates are weighted and cluster at the state-by-year level.