Our main empirical analysis examines whether women with low socioeconomic
status are more responsive to differences in the level of income inequality
in terms of their childbearing and marital outcomes. We use individual-level data
from the National Survey of Family Growth to look across U.S. states. We also use
individual-level data from the Fertility and Family Survey, conducted by the United
Nations, to look across a set of roughly a dozen developed countries. An illustrative
example of the results of our analysis is provided in Figure 5, focusing on crossstate
variation in teen births within the United States. We use the level of education
attained by the mother of each teen to separate them into different categories of
socioeconomic status and then we divide the states into high-, middle-, and lowinequality
categories based on the 50/10 ratio of household income (the ratio
of the income at the median of the income distribution to the income at the
10th percentile of the income distribution). Among teens with high socioeconomic
status whose mothers attended college, we observe no difference in the likelihood
of giving birth as a teen across these states, despite the reasonably large number of
these women who do so. Among teens with lower socioeconomic status, though,
there is a clear pattern of teen fertility across inequality categories. Teens in the
highest- inequality states are roughly 5 percentage points more likely to give birth
as a teen than teens in the lowest-inequality states. We fifi nd the opposite pattern
when we focus on abortions as a teen—much less-frequent abortions among teens
with low socioeconomic status in high-inequality states—and no pattern like this
when we repeat this exercise for sexual activity. We have also conducted a similar
exercise at the county level and obtained similar results. We also obtain analogous
results in a cross-country analysis.
One potential concern in an analysis like this is that income inequality at
the state level may be capturing any other state fifi xed factor that happens to be
correlated with it, suggesting that inequality may not be the causal determinant of
teen fertility. Although it is impossible to rule out this alternative completely, we
have experimented with including other conditions that could lead to economic
“despair,” such as poverty concentration, the incarceration rate, absolute levels of
deprivation, and others. We also consider other potentially confounding factors
that would be outside the proposed model, such as measures of state religiosity,
political preferences, and a measure of social capital. None of these additional
factors are found to alter the estimated relationship between inequality and teen
fertility among women with low socioeconomic status.
Thus, we conclude that women with low socioeconomic status have more teen,
nonmarital births when they live in higher-inequality locations, all else equal. The
proximate mechanism driving this fifi nding is that conditional on getting pregnant,
more of these girls choose to carry their pregnancy to term. Indeed, our estimates
suggest that income inequality can explain a sizable share of the geographic variation
observed in the teen childbearing rate, on the order of 10 to 50 percent. We believe
these results are consistent with the large body of work in other social science
disciplines arguing that social marginalization and hopelessness are to blame for
young, nonmarital childbearing. To the extent that greater levels of inequality are
associated with a heightened sense of economic despair and marginalization, our
empirical fifi ndings support this claim. Certainly, additional research into this link is warranted. This explanation is one of the fifi rst that has the potential to explain any sizable share of the geographic variation in teen childbearing.