The 1960s were a time of dramatic health improvement in the United
States, especially for African-Americans. In the late 1960s, the mortality rate
among Black infants—a common bellwether of population health—fell from
more than 40 deaths per 1,000 live births to 30. In the process, the rate of
infant death among Black infants converged on the lower mortality rate among
Whites. The late 1960s was the only period of convergence in the mortality
gap between Black and White infants in the post-World War II period.
Conditions of fetal and early-life health may exert a large effect on subsequent
adult health. A previous study focused on the role of fetal undernutrition
in “programming” chronic health conditions in adulthood, such as heart
disease. A natural question, therefore, is whether the large improvement in
infant health among Blacks improved adult health during the 1980s and
1990s. This pattern is indeed observed. When the authors use a dataset of
mothers giving birth in the United States, they find that both health and
education of African-Americans born in the late 1960s is substantially
improved relative to health and education of infants born in the early 1960s.
The improvement by birth cohort is much smaller among White mothers
and is not observed for Black mothers who were not born in the United
States. Therefore, infant health substantially improved in the late 1960s as
did the adult health of the same infants 20-35 years later.
What lead to the historic improvement in Black health during the late
1960s? This question is difficult to answer given the sweeping policy and
social changes of this period. These changes include the inception of both
Medicare and Medicaid, as well as the passage of the 1964 Food Stamp Act,
which provided $300 million to improve nutrition among the poor.
The goal of this paper is to assess the causal impact of the Food Stamp
Program (FSP) on health, with a particular focus on the infant health of
African-Americans. As the late 1960s witnessed major policy changes that
could confound estimates of FSP effects, the major empirical challenge is to
identify unique and exogenous FSP variation. Two basic approaches are
used to evaluate impacts of the FSP. Both of these approaches analyze the
initial rollout of the program by individual U.S. counties and compare
health outcomes immediately before and after introduction of the FSP.
The first approach looks at health outcomes before passage of the national
FSP in 1964. In the early 1960s, the first official act of the Kennedy Administration
was to establish FSP pilot projects. Health outcomes in the eight
counties that received a pilot project in 1962 are compared with (1) health
outcomes in 1961 in the same eight counties and (2) health outcomes in
counties neighboring the eight pilot counties that did not receive a pilot
project. The result is a conventional “differences in differences” estimate of
the effect of FSP pilot programs on health. The initiation of projects during
this period permits control for fixed factors that might affect infant health in
different counties (i.e., county-fixed effects).
Analysis of pilot projects permits isolation of the effects of the FSP from
Federal health initiatives that began during the subsequent Johnson Administration.
However, the small scale of the pilot programs makes analysis of
relatively rare events, like infant mortality, difficult. Therefore, this second
approach focuses on the national rollout of the FSP. To distinguish FSP
impacts from concurrent programs of the “Great Society,” the discrete
timing of program initiation by county is used. In particular, the study uses
data collected in previous research to identify the month FSP began in each
U.S. county. This second approach evaluates whether exposure to FSP
during the prenatal period has an effect on infant mortality.
The two analytic approaches revealed that infant mortality fell with
FSP exposure, especially for deaths occurring within the first month of
life. Moreover, mortality reductions were larger among Black infants than
among White infants. The study of pilot projects includes 33 counties in
both 1961 and 1962. Neonatal mortality fell nearly 2 deaths per 1,000 live
births with initiation of the FSP. This estimate is significant at the 10-
percent level of significance, while effects for the post-neonatal period are
approximately one-tenth as large and not significant at conventional levels.
A preponderant effect on neonatal mortality is consistent with a primary role
of improved prenatal conditions.
Data from the national rollout of the FSP permits analysis of the discrete
timing of program initiation. Effects are again found for measures of
newborn health, and birthweight in particular. Both White and Black infants
were less likely to be born at low birthweights (below 2,500 grams) once
the FSP began operation. The likelihood of low birthweight fell about 2
percent for Black infants and slightly under 1 percent for White infants.
Analysis of the FSP has been hampered by its regularity. Other major entitlement
programs, notably AFDC, varied substantially at the State level,
permitting analysis of State experiments with the program. The FSP
program, by contrast, is relatively monolithic. From a research perspective,
it is fortunate that the initial rollout of the FSP was not so regular. Counties
had to wait until Congress raised the FSP appropriation to levels sufficient
for nationwide coverage. In this process, counties queued for their turn. This
gradual phase-in of the modern FSP generated ready “treatment” and
“control” groups to the benefit of empirical analysis.
The modern FSP began at a time when many of America’s poorest were
starving. As late as 1964, 1,400 people died each year from hunger in the
United States. Results of this study indicate that the FSP had substantial
health benefits in reducing mortality, particularly among African-Americans
infants. This success is more notable given the persistently high levels of
infant mortality among African-American infants.
Moreover, the fetal-origins hypothesis predicts that the improved infant
health generated by the FSP would have persistent effects on the adult
health of the 1960s birth cohorts to this day, suggesting that the FSP has
additional “multiplier” effects that have yet to be measured. Future research
should identify datasets of adult health outcomes with information on the
county and date of birth and analyze these potential long-term effects using
the then-staggered phase-in of the modern FSP.