The Impact of Legalized Abortion on Crime

This article is about a paper on the subject of abortion and crime. For the subject in general, see legalized abortion and crime effect.

"The Impact of Legalized Abortion on Crime" is a controversial paper by John J. Donohue III of Yale University and Steven Levitt of University of Chicago that argues that the legalization of abortion in the 1970s contributed significantly to reductions in crime rates experienced in the 1990s. The paper, published in the Quarterly Journal of Economics in 2001, offers evidence that the falling United States crime rates of the 1990s were mostly caused by the legalization of abortion due to the Roe v. Wade court decision of 1973 .

Abstract

"We offer evidence that legalized abortion has contributed significantly to recent crime reductions. Crime began to fall roughly eighteen years after abortion legalization. The five states that allowed abortion in 1970 experienced declines earlier than the rest of the nation, which legalized in 1973 with Roe v. Wade. States with high abortion rates in the 1970s and 1980s experienced greater crime reductions in the 1990s. In high abortion states, only arrests of those born after abortion legalization fall relative to low abortion states. Legalized abortion appears to account for as much as 50 percent of the recent drop in crime."

History of argument

The 1972 Rockefeller Commission on Population and the American Future is one of the better known early versions of this claim, but it was surely not the first. The Commission cited research purporting that the children of women denied an abortion "turned out to have been registered more often with psychiatric services, engaged in more antisocial and criminal behavior, and have been more dependent on public assistance." A study by Hans Forssman and Inga Thuwe was cited by the Rockefeller Commission and is probably the first serious empirical research on this topic. They studied the children of 188 women who were denied abortions from 1939 to 1941 at the hospital in Gothenburg, Sweden. They compared these "unwanted" children to another group – the next children born after each of the unwanted children at the hospital. The "unwanted" children were more likely to grow up in adverse conditions, such as having divorced parents or being raised in foster homes and were more likely to become delinquents and engaged in crime. Levitt and Donohue revived discussion of this issue with their paper, "The Impact of Legalized Abortion on Crime."

Responses

Sailer

In 1999, before the paper was published, a debate was held between magazine writer and Internet columnist Steve Sailer and Steven Levitt at Slate.com. (See this link for Levitt's opening, this link for Sailer's response, this link for Levitt's rebuttal, and this link for a final Sailer response.) Sailer argues that the end of the crack wars was more significant than abortion -- and that, contrary to what Levitt's thesis would suggest, "the murder rate for 1993's crop of 14- to 17-year-olds (who were born in the high-abortion years of 1975 to 1979) was a horrifying 3.6 times that of the kids who were 14 to 17 years old in 1984 (who were born in the pre-legalization years of 1966 to 1970)."

Lott and Whitley

In 2001, John Lott and John Whitley argued that Donohue and Levitt assume that states which completely legalized abortion had higher abortion rates than states where abortion was only legal under certain conditions (many states allowed abortion only under certain conditions prior to Roe) and that CDC statistics do not substantiate this claim. In addition, if abortion rates cause crime rates to fall, crime rates should start to fall among the youngest people first and then gradually be seen lowering the crime rate for older and older people. In fact, they argue, the murder rates first start to fall among the oldest criminals and then the next oldest criminals and so on until it last falls among the youngest individuals. Lott and Whitley argue that if Donohue and Levitt are right that 50 percent of the drop in murder rates during the 1990s is due solely to the legalization of abortion, their results should be seen in these graphs without anything being controlled for, and that in fact the opposite is true.

On May 15, 2005, Levitt replied to Sailer and Lott in his Freakonomics blog at the New York Times, and argued that two of the key assumptions underlying their alternative hypothesis are false.

Joyce

Ted Joyce made a number of arguments against the abortion and crime hypothesis in his 2004 paper "Did Legalized Abortion Lower Crime?". He claimed that legal abortions in the early 1970s were just replacing illegal abortions, that there was no measurable impact of abortion between 1985 and 1990, that cohorts born before 1973 had roughly the same crime rates as cohorts born after 1973 in the states where abortion was legalized in 1973, and that omitted variables are driving the results.

Donohue and Levitt respond to each point that Joyce makes in a reply paper and conclude that none of Joyce's arguments cast doubt on the original hypothesis presented in their 2001 paper. They also introduce an updated version of their dataset which had better measures of abortion (given to them by Stanley Henshaw of the Alan Guttmacher Institute after their initial paper was published).

Foote and Goetz

In November 2005, Federal Reserve Bank of Boston economists Christopher Foote and Christopher Goetz released a working paper, "Testing Economic Hypotheses with State-Level Data: A Comment on Donohue and Levitt (2001).", in which they argued that Donohue and Levitt's study did not estimate the regressions that Donohue and Levitt had claimed that they examined. In particular, Foote and Goetz said that, despite their claims that they had done so, the 2001 Donohue and Levitt study failed to control for influences that varied within a state from year to year (such as the effect of crack-cocaine). Foote and Goetz also point out that Donohue and Levitt accidentally used the total number of arrests, not the arrest rate, to explain the murder rate. Using the total number of arrests does not establish the unwantedness mechanism Donohue and Levitt propose, only that the total number of arrests has changed. After making these two corrections, Foot and Goetz interpreted their results as evidence that violent crime actually increases with more abortions and that property crime is unrelated to abortions. This study received press coverage in The Wall Street Journal and The Economist.

Donohue and Levitt admit the programming error made in the original version of the paper and then go on to address the two points that Foote and Goetz make (see here for the reply). Donohue and Levitt contend that even though the Foote and Goetz analysis was doing what Donohue and Levitt claim that they were originally doing, the Foote and Goetz analysis produces heavy attenuation bias (the reason they find no statistical relationship between abortion and crime). To remedy this, Donohue and Levitt use the improved abortion measures (that Lott and Whitley originally used) and they make other changes that they now argue are necessary. Donohue and Levitt claim that with these new changes the results are smaller, but still statistically significant.

Nevin

Rick Nevin suggests an alternative explanation for the correlation observed by Donohue and Levitt. His study, which included several other countries in addition to the United States, suggests that variations in estimated early-childhood blood lead levels (from exposure to leaded gasoline and paint) can better explain the variation in violent crime rates over the years. For example, it better explains why crime rates rose in the first place, and continued to rise in the UK (who continued to use leaded gasoline after the USA phased it out) even though abortion was legalized. It also explains why New York (especially NYC) and California (which legalized abortion before Roe) saw crime rates drop earlier than the rest of the nation, since they were also more proactive in reducing lead exposure. The theory is biologically plausible, given that lead is a known neurotoxin which has long been linked to impulsivity, delinquency, violence, and reduced IQ.

In Levitt's reply in his Freakonomics blog at the New York Times, he notes that while a 2007 study by Jessica Reyes also found that lead had a large effect on violent crime, the same study found that the effect of legalized abortion remained similar in magnitude to that in Levitt's study, and statistically and practically significant (albeit as one of the lesser factors), even after controlling for blood lead levels and several other confounders.

See also

• Legalized abortion and crime effect
• Roe effect
• William Bennett
• Freakonomics