The United States has relatively high rates of low-weight
births, preterm births, and infant mortality
compared with other Western industrial countries. A
number of Federal programs—WIC, Food Stamp,
Medicaid, and AFDC—have provided benefits to help
improve birth outcomes. In spite of the fact that many
women participate in more than one of these
programs, previous research has only considered the
effect of participation in one program at a time. Such
analysis may lead to misleading conclusions if women
participate in multiple programs and if some programs
are effective while others are not. It also ignores the
possibility of synergies among programs. In this paper,
Brien and Swann consider whether participation in
more than one program improves birth outcomes. They
allow for possible synergies among programs in their
analysis, and attempt to control for nonrandom selection
into the programs.
Brien and Swann used data from the National
Maternal and Infant Health Survey, conducted in 1988.
The survey includes information about welfare
program participation, birth outcomes, and sociodemographic
characteristics of mothers and families for
almost 19,000 women. The authors restrict their
analysis to 3,451 low-income, single women for whom
the dataset contained complete information on all the
The authors use a number of techniques to evaluate the
impact of program participation on birth outcomes,
including simple descriptive statistics comparing
average birth outcomes of women who participate in
different combinations of programs; ordinary least
squares regressions controlling for observed characteristics
but not for the selection of women into the
various programs; and a more complex model of the
decision to participate in each of the four programs
and the resulting birth outcome. Highlights of their
findings include the following:
- The descriptive analysis shows that WIC recipients
can expect better birth outcomes than nonrecipients.
Women who participate in the other programs can
generally expect worse birth outcomes than nonparticipants,
suggesting possible adverse selection into
- After controlling for observed characteristics such
as age and education, WIC participation continues
to improve birth weights by approximately 230
grams, on average.
- When allowing for synergies among programs, WIC
continues to be effective. Though there appear to be
some synergies, there is no consistent pattern across
all bundles of choices.
- The positive effect of WIC participation on birth
weights is statistically insignificant after controlling
for nonrandom selection into the programs.
In future work, Brien and Swann plan to refine their
method of determining program eligibility and attempt
to better understand the participation decisions for the
possible bundles of programs.