In the journal Economica , Leo Kahane, David Paton and Rob Simmons offer an analysis of the supposed link between abortion rates and the reduction of crime rates in the UK (the article is entitled, “The Abortion–Crime Link: Evidence from England and Wales”). The authors challenge the findings of several articles by J. J. Donohue and S. D. Levitt (D&L), whose argument they Kahane, et. al., summarize as follows:

“The authors argue that the legalization of abortion some fifteen to twenty years earlier may be a large part of the answer. Their hypothesis (discussed in greater detail below) is that legalized abortion ultimately reduced the birth of children who, had they have been born, would have been at greater risk of committing crimes when they reached their teenage years.”

There are two crucial claims in the D&L thesis: “The D&L hypothesis that changes in abortion rates may lead to changes in crime rates some years later encompasses two propositions: that changes in the abortion rate will change the size of birth cohorts, and that they will change the composition of the cohort. Only one of these propositions needs to be valid for changes in abortion rates to have an effect on crime rates. The first proposition straightforwardly states that if the abortion rate increases then, ceteris paribus , the birth cohort will be smaller, thereby leaving fewer individuals to commit crimes later. But it is well known that abortion in the United States probably had only minimal effects on the number of children born, calling this first proposition into question. Hence it is the second proposition that D&L stress in their paper, and this is clearly more controversial. A strong link between abortion and crime may be revealed if a selection effect is operational. ”

Based on their study of the UK, Kahane and his co-authors find that the evidence for such a link is not compelling: “Based on the above analysis, we are unable to say that abortion legalization in the United Kingdom significantly reduced crime in England and Wales some twenty years thereafter. We come to this conclusion by first noting, as we did earlier, that total recorded crime in the United Kingdom began to decrease at about the same time as in the United States, despite the fact that abortion legalization occurred here about five years earlier. Thus, we have a discrepancy in the timing of the potential effect of abortion on crime between the United States and the United Kingdom. Further, crime in England and Wales did not decrease relative to areas in which abortion remained illegal throughout the time period.”

They admit that “regression models linking effective abortion rates in the United Kingdom and subsequent recorded crime suggest the same negative and significant correlation between the two variables (at least for total recorded crime and some sub-categories) as that reported for the United States by D&L. ” But they conclude that the pattern of results does not demonstrate that abortion is a significant cause of a reduction of crime rates: “We hypothesized above that this pattern of results might be explained in one of two ways: first, that abortion reduced crime in both the United Kingdom and the United States but that the timing of the relationship in the United Kingdom is obscured by other variables; or, second, that abortion did not reduce crime in either the United Kingdom or the United States and that the negative correlation observed in the regressions was a spurious one.”

Further, they fear that concentrating on the abortion-crime link may result from an illegitimate narrowing of possible causal factors in the falling crime rate and in abortion rates: “one of the key weaknesses that we believe exists not only in the D&L paper, but in other papers attempting to link abortion rates to crime rates some fifteen to twenty years hence is the distinct likelihood that abortion is endogenous to crime, owing to omitted variables bias. For example, we can consider factors such as education and income levels of women, which may affect the fertility decisions of women and, in particular, the likelihood that an unwanted pregnancy will be terminated. According to research by Finer and Henshaw on unintended pregnancy in the United States in 2001, women below the poverty level had an abortion ratio (relative to unintended births) of 0.93, whereas the figure for women with income more than twice the poverty level was 1.64. As for education, women with less than high school education had an abortion ratio of 0.90, while women who were college graduates had an abortion ratio of 1.60.” In short, “To the extent that these same measures—income and education—would have an effect on the human capital endowment of their children, and ultimately the propensity for these children to engage in criminal activity, the exclusion of such measures would bias the estimated coefficient to abortion.”