Abortion and Crime, Revisited (Update)

Episode Summary

Executive Summary: The episode revisits Steve Levitt and John Donohue’s controversial claim that legalized abortion contributed substantially to the 1990s crime decline. It walks through the original hypothesis, the evidence and critiques, a later correction that strengthened the results, and a new analysis extending the pattern through 2014. The discussion ends by stressing multicausal explanations, better data literacy, and the policy limits of using this research to judge abortion law.

Main Topics: Roe v. Wade, abortion politics, and current policy shifts (Priority: 5/5): Dubner frames the episode in the context of the post-Roe landscape: state bans, state ballot measures, and competing claims from Harris and Trump about national policy. The original abortion-crime hypothesis (Priority: 5/5): Levitt and Donohue explain how legalized abortion may have reduced unwanted births, which in turn lowered later crime rates among cohorts born after legalization. Evidence, methodology, and the 2001 controversy (Priority: 5/5): The episode details their collage of evidence, the publication of the 2001 paper, and the intense backlash, including critiques over coding and table errors. Lead exposure as a competing explanation (Priority: 4/5): Jessica Reyes’s research is used to show that lead removal also plausibly explains part of the crime decline, reinforcing the idea that multiple causal factors can coexist. Updated analysis of crime through 2014 (Priority: 5/5): Levitt and Donohue’s later update finds the relationship still holds and may be even stronger than before, implying abortion access had large long-run crime effects. Policy, ethics, and the danger of single-cause narratives (Priority: 4/5): The episode argues that the research should inform understanding of crime and child welfare, but not be used simplistically to settle moral questions about abortion.

Key Arguments: Legalized abortion likely reduced the number of unwanted children, and unwantedness is associated with worse long-run outcomes, including higher criminality. Because abortion access changed dramatically across states and cohorts, it provided a natural experiment that could be studied indirectly through crime patterns. The original 2001 paper found that higher-abortion states experienced larger later crime declines, and age-cohort analysis supported the same conclusion. Critics were right to challenge an error in the paper, but the correction weakened the estimate only slightly and did not overturn the hypothesis. Lead removal is another important factor in the crime decline; the evidence supports multivariate causality rather than a single explanation. The updated analysis found that high-abortion states saw crime fall about 60% more than low-abortion states from 1997 to 2014, suggesting a very large cumulative effect. Even if the research is empirically compelling, it should not be used to make a simple moral argument for or against abortion because the ethical trade-off depends on one’s values. A more useful policy takeaway may be that society should reduce unwanted pregnancies and improve child welfare rather than focus only on punishment or abortion law.

Data Points: States banning abortion: 13 states - Current post-Roe landscape described in the opening segment States with more limited abortion restrictions: 8 states - Current post-Roe landscape described in the opening segment States with abortion ballot measures: 10 states - Election Day measures aimed at protecting abortion access Year Roe v. Wade decided: 1973 - Supreme Court ruling that legalized abortion nationwide Abortions at peak: 1.5 million per year - Levitt describes the scale of abortion at its peak Live births at peak: roughly 4 million per year - Compared with annual abortions at the peak Abortion ratio at peak: 345 abortions per 1,000 live births - Magnitude used in Levitt and Donohue’s back-of-the-envelope estimate Crime decline: more than 30% - U.S. violent crime fall from 1991 to 2001 New York homicides in 1990: more than 2,200 - Illustrative city-level example of the earlier crime era New York homicides in 2023: fewer than 400 - Illustrative city-level example of the crime decline Share of 1990s crime drop explained by abortion in original paper: as much as 50% - Stated finding from the 2001 Quarterly Journal of Economics paper Crime difference between high- and low-abortion states by 1997: roughly 30% - Original cross-state comparison in the paper Updated relative crime decline in high-abortion states: about 60% more - New analysis comparing 1997 to 2014 Estimated total share of long-run crime decline explained by abortion: about 80% to 90% - Levitt’s description of the updated cumulative effect Teen pregnancy decline over 30 years: more than 75% - Used as a broader indicator of falling unwanted pregnancies Estimated homicides prevented: 5,000 to 10,000 fewer homicides - Levitt’s rough policy-relevant estimate tied to abortion access Lead phase-out period: 1975 to 1985 - EPA lead removal timeframe discussed by Reyes

Pivotal Quotes: "I think maybe legalized abortion might have reduced crime." — John Donahue: The moment the central hypothesis first emerged in Levitt’s office "So, the abortion hypothesis is quite unusual... it makes really strong and quite straightforward predictions about what should happen in the future." — Steve Levitt: Explaining why the team revisited the study decades later with updated data "If you're pro-life and you believe that the fetus is equivalent in moral value to a person, well, then the trade-off is awful." — Steve Levitt: Describing why the empirical result does not resolve the ethical abortion debate

Implications: The episode suggests abortion policy, crime policy, and child-welfare policy should not be debated as single-cause issues. It also highlights the need to distinguish empirical evidence from moral judgment and to communicate social-science findings with nuance.