03.10.2014 Views

Schooling, Family Background, and Adoption: Is it Nature or is ... - Etla

Schooling, Family Background, and Adoption: Is it Nature or is ... - Etla

Schooling, Family Background, and Adoption: Is it Nature or is ... - Etla

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

If we repeat the analys<strong>is</strong> w<strong>it</strong>h 1992 income, parental IQ effects remain<br />

similar, <strong>and</strong> income effects become somewhat higher f<strong>or</strong> sons. Our nature<br />

estimates show that w<strong>it</strong>h respect to IQ transfers <strong>and</strong> educational outcomes<br />

44 percent <strong>is</strong> in the genes f<strong>or</strong> boys. Our 102 percent estimate f<strong>or</strong> girls shows<br />

that <strong>it</strong> <strong>is</strong> all genetics. W<strong>it</strong>h 1992 income that <strong>is</strong> generated by market luck<br />

we find that the nature component f<strong>or</strong> sons <strong>and</strong> daughters drop <strong>and</strong> become<br />

32 <strong>and</strong> 90 procent respectively.<br />

In the end, however, all likelihood ratio tests indicate that th<strong>is</strong> model<br />

<strong>and</strong> the model we estimate in Table 6 are stat<strong>is</strong>tically identical (cr<strong>it</strong>ical value<br />

<strong>is</strong> set at 14.1). Hence, in all four different specifications the estimates cannot<br />

reveal that boys <strong>and</strong> girls are affected differently by family background<br />

variables such as family income <strong>and</strong> parental IQ, <strong>or</strong> that the environment<br />

treats boys <strong>and</strong> girls differently.<br />

6 Selectiv<strong>it</strong>y, adopting families <strong>and</strong> adopted children<br />

While our nature <strong>and</strong> nurture estimates suggest that genes are rather dec<strong>is</strong>ive,<br />

we should treat our estimates w<strong>it</strong>h care. Since we do not observe<br />

abil<strong>it</strong>y of the natural parents of adopted children, the estimates may still<br />

suffer from abil<strong>it</strong>y bias. In fact, we are qu<strong>it</strong>e convinced that such a bias<br />

ex<strong>is</strong>ts. To determine sources of th<strong>is</strong> bias, <strong>it</strong> <strong>is</strong> instructive to return to our<br />

model once m<strong>or</strong>e. If e ∗ t−1 represents the parental abil<strong>it</strong>ies of biological parents<br />

of adopted children, the c<strong>or</strong>rected abil<strong>it</strong>y mobil<strong>it</strong>y relation <strong>is</strong> defined<br />

as<br />

e t = b 0 + (b 1 − b g1 δ t )e t−1 + b g1 δ t e ∗ t−1 + v t (6.1)<br />

which implies that the human cap<strong>it</strong>al function reads as<br />

h t = c 0 + b 0 c 2 + c 1 y t−1 +<br />

b 1 c 2 e t−1 − b g1 c 2 δ t e t−1 + b g1 c 2 δ t e ∗ t−1 + w t + c 2 v t (6.2)<br />

W<strong>it</strong>h th<strong>is</strong> in mind, we briefly outline some of the potential dangers of abil<strong>it</strong>y<br />

bias.<br />

• Selection in genes <strong>and</strong> adopted children.<br />

Children who are given up f<strong>or</strong> adoption are m<strong>or</strong>e likely to have less<br />

fav<strong>or</strong>able socio-economic backgrounds. The mechan<strong>is</strong>m to explain why<br />

18

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!