This dissertation consists of two essays.
The first essay explores the child support enforcement policies and
their implications. The child support enforcement (CSE) policies, aimed at protecting
out-of-wedlock children from financial disadvantages, brought unexpected
changes in individuals' marriage and fertility behaviors during the
1980s and the 1990s. Our estimates from state-year panel data show
that in states with strict CSE there has been a significant decrease
in non-marital births and a significant increase in marital births. Taking into account all these changes, what are the effects of CSE
on children's welfare? To answer this question, we build a heterogeneous-agent
model that features endogenous marriage and child-investment decisions.
Exploiting the state-level variation in enforcement, we estimate it
using the National Vital Statistics Report data. We find that men's
increased willingness to marry is the driving force behind the shift
from non-marital births to marital births. As evidence for the mechanism,
we show that the number of marriages has risen in the states with strict CSE during the same period, consistent with the model's implication.
Our model predicts that a large increase in child investment comes
through a secondary effect of CSE: the shift from non-marital births
to marital births increases child investment through its income effect.
In the second essay, we ask to what extent changes to the age and sex structure of the
population account for the changes in the marriage behavior observed
in the last century (from 1900 to 1980). The decrease in mortality,
especially for women, and the changes in immigration patterns have
increased the female to male ratio. With respect to marriage, there
has been i) an increase in its incidence, ii) a reduction in the gender
gap of the median age at first marriage, and iii) an increase in the
divorce rate. We pose a model of marriage and divorce in which preferences
over spouses depend on their age and on love (an idiosyncratic shock)
and where frictions make it difficult to get new partners. We estimate
our model using marital and population patterns of the 1950-1959 birth
cohorts. Using the preference parameters estimated on the 1950's cohort
and the population patterns of the 1870's cohorts, we find marriage
patterns are quite similar to those observed in the earlier period.
By making divorce costly for the 1870's cohort, the resemblance is
more stronger. In particular, we find that these features account
for i) 94.5% of the increase in the incidence of marriage ii) 140.8%
of the shrink the gender age gap in the median age at first marriage.