children come from households that have 19.6 percent lower enrollment rates, which is consistent
with previous results in which receiving households are shown to be better off (Akresh, 2004).
Controlling for the gender of the child indicates boys are 2.8 percent more likely to be enrolled
than girls. In all of the regressions, to capture any level effect on enrollment due to a child’s age, I
include dummy variables for each age. All the regressions in this table also include year dummies
intended to capture secular time trends in school enrollment.
In column 2, I analyze whether the fostering impact varies based on the child’s gender by in-
cluding an interaction of the variable EverFosteredij * AfterFosteringjt with a variable indicating
if the child is male. Foster boys are 6.1 percent more likely to be enrolled than foster girls, yet this
difference is not significant. Column 3 measures whether the fostering impact varies based on the
child’s age by including an interaction of the variable EverFosteredij * AfterFosteringjt with age
variables indicating if the child was young (aged 5, 6, or 7), middle (aged 8 to 11), or old (over age
12). Young foster children are 18.4 percent more likely to be enrolled after the fostering relative to
older foster children, and the coefficient is statistically significant at the 5 percent level. Children
aged eight to eleven are 10.0 percent more likely to be enrolled after the fostering compared to older
foster children, but the coefficient is not significant at standard levels (t-statistic is 1.54). These
results indicate that for certain children, fostering is a strong positive experience (at least in terms
of school enrollment), while for other children (in particular older children), the results are more
mixed and potentially negative.
Panel B of Table 4 presents household fixed effects regressions comparing foster children with
their biological siblings. The regression in column 4 indicates foster children are 3.6 percent more
likely to be enrolled after the fostering than the biological siblings who remained behind. This is
evidence that while foster children may not be better off compared to their host siblings (column 1),
they are more likely to be enrolled after the fostering compared to their biological siblings. Column
15