foster children relative to either host or biological siblings. Results in columns 3 and 6 show that
younger foster children are 12.0 and 11.8 percent more likely to be enrolled after the fostering
exchange compared to host and biological siblings, respectively.
These results indicate that, after conditioning on the child’s type, there are no longer any
positive enrollment impacts following the fostering, with the exception of younger foster children.
This provides evidence that the biological parents, in selecting which child to send out, are probably
taking into account these factors that are unobservable to the researcher yet clearly influence how
the foster child fares in the host household. Even after controlling for the child’s type, the impact
of fostering, while not positive, is still not negative as many individuals believe.
4.3 Comparisons with Children of Non-Fostering Households
Comparing foster children with their host and biological siblings is important, but it is also necessary
to compare them with children from non-fostering households in order to understand the benefit to
families of being able to reallocate resources by sending away children. Table 7 presents an overview
for children from non-fostering households with information similar to Tables 1,2, and 3. In Panels
A and B, these children appear similar to the biological siblings, with an average enrollment rate
of 18.3 percent, an average age of 9.4 years, and a school enrollment transition rate in which 76.4
percent of them are not enrolled in either year and 2.8 percent discontinued enrollment in 2000.
Panel C presents difference in differences results comparing foster children with children from non-
fostering households using pre and post-fostering enrollment for the foster children and 1999 and
2000 enrollment for the other children, with similar results using 1998 and 1999 enrollment for
these other children. There is no statistical difference between foster children and children from
non-fostering households in terms of enrollment after foster children are sent away, but again the
table does not use all available information and is presented only as a comparison to Table 3.
18