The unintended: negative outcomes over the life cycle

Abstract

We quantify the impact of abortion legalization on the incidence of unintended births. While underlying much of the literature on abortion legalization, this effect had only been approximated by previous work. We find a strong decline in the prevalence of unintended births. Moreover, we find that this decline is mainly driven by “pro-choice” women. We then propose an empirical strategy to recover the effect of being “unintended” on life cycle outcomes. We use the differential timing of abortion legalization across states interacted with the mother’s religion (which facilitates or hinders legal abortion take up) to instrument for endogenous pregnancy intention. We find that being unintended causes negative outcomes (higher crime, lower schooling, lower earnings) over the life cycle. Our paper provides an initial step towards quantifying this key mechanism behind many of the well-documented long-term effects associated with changes in reproductive health policy

Appendix

Using changes in adoptions to measure the effects of abortion legalization on unintended births is may not provide an accurate estimate because of the following three reasons: First, Bitler and Zavodny use changes in adoptions granted to non-relative petitioners. This measure is an equilibrium observation from the adoptions market and may not closely track the changes in the supply of unintended babies put up for adoption. Second, only a small fraction of unintended babies are put up for adoption. Third, the impact of abortion legalization on unintended births that are not put up for adoption is likely to be different from that on unintended births put up for adoption. We formalize these last two points here.

Let $b_{\text {t}}=\frac {B_{\text {t}}}{F_{\text {t}}}\times 1,000$ be the birth rate at time t defined as the number of live births per 1,000 women. We can decompose this birth rate into its “intended” and “unintended” components

• $i_{\text {t}}=\frac {{B_{\text {t}}^{\text {I}}}}{F_{\text {t}}}\times 1,000=$ intended birth rate _t

• $u_{\text {t}}=\frac {{B_{\text {t}}^{\text {U}}}}{F_{\text {t}}}\times 1,000=$ unintended birth rate _t

Moreover, we can further distinguish within unintended births, ${B_{\text {t}}^{\text {I}}}$, those who will be put for adoption, ${B_{\text {t}}^{\text {A}}}$ from those who will not be put for adoption, ${B_{\text {t}}^{\text {N}}}$. The corresponding rates are

• $a_{\text {t}}=\frac {{B_{\text {t}}^{A}}}{F_{\text {t}}}\times 1,000=$ for adoption birth rate _t

• $n_{\text {t}t}=\frac {{B_{\text {t}}^{\text {N}}}}{F_{\text {t}}}\times 1,000=$ not for adoption birth rate _t

Note that

$$ b_{\text{t}}=i_{\text{t}}+u_{\text{t}} $$

(6)

$$ u_{\text{t}}=a_{\text{t}}+n_{\text{t}} $$

(7)

Let’s denote by α ^y the percent causal effect of abortion legalization on a generic birth rate y for y=b,i,u,a

$$ y_{\text{legal}}=y_{\text{illegal}}\left( 1+\alpha^{\text{y}}\right) $$

(8)

As explained in the body of the paper, one of the missing links in the study of the effects of abortion legalization is its (percent) causal impact, α ^u, on the rate of unintended births. We expect α ^u < 0. Note that since u _illegal = a _illegal+n _illegal we have

$$ \alpha^{\text{u}}=\frac{{B_{\text{t}}^{\text{A}}}}{{B_{\text{t}}^{\text{U}}}}\alpha^{\text{a}}+\frac{{B_{\text{t}}^{\text{N}}}}{{B_{\text{t}}^{\text{U}}} }\alpha^{\text{n}} $$

(9)

This makes clear that while the impact on adoption rates, α ^a (which Bitler and Zavodny (2002) and Levine (2004) identify) is an important part of the story, we still need two key parameters to estimate the impact of abortion legalization on the rate of unintended births α ^u:

• the share of unintended births that are being put up for adoption, $\frac {{B_{\text {t}}^{\text {A}}}}{{B_{\text {t}}^{\text {U}}}},$ and

• the impact of abortion legalization on the number of unintended births that are not being put up for adoption, α ⁿ

Say, only 1.5 out of 10 unwanted births are put for adoption. Then, $\frac { {B_{\text {t}}^{\text {A}}}}{{B_{\text {t}}^{\text {U}}}}=0.15$ and the actual impact of abortion legalization on the unintended birth rate would range from −5 to −90 %²⁹, depending upon the extent of decline on unintended births not relinquished for adoption, α ⁿ. Clearly, this range is not very informative. One option is to assume that α ⁿ=α ^a. Alternatively, one could conjecture that legalization had a stronger effect on unintended births put for adoption than on unintended births not put for adoption (i.e., |α ⁿ|<|α ^a|). In this case, the impact on the overall rate of unintended births would be lower, |α ^u|<|α ^a|. The Bitler-Zavodny-Levine estimate of a 32 % decline would then be an upper bound for |α ^u|. In particular, this would be true if unintended births who are put for adoption are “more unwanted” than unintended births not put for adoption. And, the higher the degree of unintendedness in a pregnancy, the higher the probability that it will be terminated by abortion. Alternatively, one could argue that the more responsible parents, or those who have moral problems with the use of abortion, are the ones who put unintended children up for adoption. In this case, legalization would have a stronger effect on unintended births not put for adoption. In this case, the Bitler-Zavodny-Levine estimate would be a lower bound for |α ^u|.