Abortion and Crime: An Update

In Chapter Two of Beyond Choice, on pages 66-67 in the hardback, I discussed, in the section on eugenics, the abortion/crime controversy, citing the 2001 study by Donohue and Levitt, which found that the legalization of abortion resulted, twenty or so years later, in a reduction in the crime rate because potential criminals were being aborted rather than born. I also cited contradictory studies that found no effect, or the opposite effect, of abortion on crime. I concluded saying that “The best that can be said is that the case for the alleged causal connection between the legalization of abortion and a decrease in crime rates is unproven.”

Since the publication of Beyond Choice, there has been much heat and somewhat less light on the issue. Most notably there was the publication of Freakonomics by Levitt and Dubner in 2005. A reader suggestedthat I should revisit the entire issue based on Freakonomics. So, here goes.

I will spare you the gory econometric details of the argument. While I have an MBA and studied statistics and did regression analyses, that was many years ago and on a computer that took up an entire room. Fortunately for me, the American Enterprise Institute conducted a symposium on this issue on March 28, 2006. A transcript is available on the AEI website. Much of it is accessible to the non-economist.


Since that symposium there have been papers and responses written by many of the symposium participants and by a few new entrants intothe fray. This battle is not over.

Levitt was not present at the AEI symposium, but John Donohue was, where he defended his and Levitt’s 2001 paper as well as two subsequent papers which corrected errors in the 2001 paper that had been pointed out by his critics. Virtually every economist (or their co-author) who has studied the issue was present. None agreed with Levitt and Donohue entirely, and some disagreed completely and came to the opposite conclusion. All of theten panelists were male, a comment perhaps on the state of the economics profession and/or what kind of economist wants to enter this particular debate.

The lag time between a woman’s decision whether or not to have a child (or give it up for adoption) and the time when criminal behavior becomes apparent is between 15 and 25 years. This is the first problem in trying to identify causation versus correlation. There’s many a slip between the cup and the lip during 25 years. Other antisocial or delinquent behaviors of unwanted children may be manifested earlier, such as poor school performance, alcohol and drug abuse and health problems. The challenge for policy makers andacademics is to isolate one factor― the legalization of abortion―and to calculate its impact, if any, in the subsequent anti-social manifestationsof being unwanted. The Levitt model reportedly had over 1100 different variables, including one imagines, poverty, single parenthood, peer pressure, neighborhood, family, church attendance, social programs, father’s involvement, sibling influence ― the list is almost endless.

Donohue and Levitt argue that legalizing abortion would affect the crime rate twenty years hence through two mechanisms: 1) the cohort size effect, i.e., fewer children being born and thus fewer potential criminals being alive twenty years later and 2) the selection effect, i.e., abortions will be more common for children who were unwanted. Unwanted means in this case that the parents (or mother) did not have the means or disposition to care for the child, and that there would be less of an investment (time, money, effort) in the child, who would therefore have a higher propensity to become a criminal.

On its face, the argument seems to make sense. Women who have abortions do not want to be a mother at that time, if ever. They want to invest in the children they already have or in later children born at a time when they can be a better parent. Abortion, by definition, can affect either the absolute number of children born or their timing, or both. Earlier research, though not by economists, seemed to show that unwanted children did have a higher propensity for poor school performance and delinquent behaviors.Advocates of legal abortion have in the past used this as an argument for notcriminalizing the procedure.

There is a problem with the terms “unintended” and “unwanted” however. When do we make this judgment and who makes it? Is a potential child “unintended” and “unwanted” at the time of sex, at conception, or at three or six months gestation? Is it “unwanted” at birth, at age two, at age 13? Different parents may have different answers at different times. And isn’t it possible that a parent’s idea of wantedness can change, perhaps multiple times, during a child’s prenatal and postnatal life? And which parent are we talking about? Mother? Father? Both?

Aside from upbringing, genes and all the other factors that effect a child’s decisions as to their life course, there are societal factors influencing crime levels: in the case of the US in the late 1980’s and early 1990’s there were efforts to put more police on the street, longer jail terms, better policing, the good economy and the devastating Crack Epidemic. It is a challenge to control for all these intervening effects.

In addition, there are problems of measuring abortion both before and after legalization. Pre-Roe,and pre-1970 in New York and a few other states, legal abortion was difficult to access, though there were therapeutic abortions available to a greater or lesser degree. Criminal abortion was a major enterprise. How to measure the extent of legal and illegal abortions before it was decriminalized is a major problem since statistics weren’t kept. And even after legalization, not every state has accurate records.

Then there is the people problem. People move. A lot. People go across state lines to get abortions and move to other states to live. It is hard to measure a state’s abortion rate in the first place, even the legal abortion rate after Roe, and hard tokeep track of the families who move in and out of state and to discern in all cases if a crime is committed by a person born in the state or elsewhere.

On top of these data problems, there is the issue that the legalization of abortion does more than decriminalize the procedure. Some academics argue, and I cite this in Beyond Choice, that abortion can act as an insurance policy and lead to more risky sex, thus more pregnancy, more abortion and unwanted children. These academics argue that legalizing abortion leads therefore to an increase in crime.

One economist from the Federal Reserve Bank of Boston argued that state crime levels were converging in the 1990’s and that abortion had nothing to do with it. Their regression analysis showed that it was the high crime states that were seeing their crime rates drop, not the high abortion states.They also found that property crime levels increased even as violent crime was falling.

Other analysts noted that young males between the ages of 17 and 25 commit the majority of crimes. If abortion did reduce crime, crime rates would have dropped first among young people, but they didn’t. The number of crimes committed by older people dropped first in the 1990’s. Furthermore, while the rate of homicide committed by young men dropped, the rate of aggravated assaults among the young increased, and the rate of homicides committed by young females — which should have been equally affected by abortion as males – did not drop.

The economists at the AEI symposium seemed to agree that there were all these data issues, modeling issues and econometric issues, and others too arcane to discuss here. Donohue and Levitt believe they have solved these issues. The rest disagreed. In sum, most economists present agreed that the evidence for an abortion/crime link is “pretty weak”, “really inconsequential”, not “statistically significant”, and “rather bleak”. The model that Levitt used is what economists calls “sensitive”, not “robust,” meaning that anytime one thing changes, the coefficients of the abortion/crime link change dramatically.

Nonetheless, Donohue and Levitt are sticking to their argument and producing new papers answering their critics. The moderator of the event said later, trying to be diplomatic, that: “I think the consensus position is that the abortion effect probably explains some of the crime decrease but most likely not nearly as much as Donohue and Levitt estimate and no one has much confidence in the precise size of the effect.”

My conclusion remains what it was in 2004 in BeyondChoice. The proposition that there is a connection between legalizing abortion and a subsequent reduction in the crimerate remains unproven.

It was pointed out at the AEI symposium that there is perhaps reluctance on both the Left and Right to validate the Levitt thesis: the Right because it gives societal legitimacy to abortion and the Left because it smacks using racial profiling and eugenics to support the legalization of abortion.

I framed my discussion of the abortion/crime link in BeyondChoice by saying that “eugenics disguised as social engineering wasn’t dead yet.” This comment followed a long discussion of the Norplant saga where the Philadelphia Inquirer, after stating that those having the most children are the least capable of supporting them, suggested incentives for the poorer members of society to use Norplant. And after my discussion of the abortion/crime link, I said that “At its worst, this argument is eugenics in new clothing.”

One reader has suggested that I maligned Levitt by associating him with eugenics.

The AEI conference was notably free of policy recommendations by the panelists, even coded ones. Levitt denies that his theory has racial implications. A moderator (not Levitt) did raise policy questions by saying that the debate over the abortion/crime link could inform the debate as to whether or not the states should cover abortion in their Medicaid program or whether parental consent laws should be enacted, as both these provisions affect the ease of access by the poor and the young to abortion, those whom, under Levitt’s thesis, would be most likely to give birth to future criminals. I have not read anything where Levitt or his co-author make any policy recommendations, nor any statements that sound like eugenics.

That said, the American Enterprise Institute is not an academic, non-partisan think tank. They want to affect policy in Washington. I doubt they would sponsor a symposium that would conclude with an endorsement of Medicaid-funded abortions. Economists too have political opinions, as do we all. Eugenics and the fear of eugenics lurk all around the abortion/crime debate. Only rarely does it surface explicitely.

Former Education Secretary William Bennett, hardly a supporter of legal abortion, entered the fray on his radio show in 2005. Here is the transcript of a section where he and a caller discuss Freakonomics:

BENNETT: All right, well, I mean, I just don’t know. I would not argue for the pro-life position based on this, because you don’t know. I mean, it cuts both — you know, one of the arguments in this book Freakonomics that they make is that the declining crime rate, you know, they deal with this hypothesis, that one of the reasons crime is down is that abortion is up. Well–

CALLER: Well, I don’t think that statistic is accurate.

BENNETT: Well, I don’t think it is either, I don’t think it is either, because first of all, there is just too much that you don’t know. But I do know that it’s true that if you wanted to reduce crime, you could — if that were your sole purpose, you could abort every black baby in this country, and your crime rate would go down. That would be an impossible, ridiculous, and morally reprehensible thing to do, but your crime rate would go down. So these far-out, these far-reaching, extensive extrapolations are, I think, tricky.

If that isn’t eugenics in new clothing, I don’t know what is.

I end up where I began. I don’t see much of a role for economists analyzing the after-effects of biological imperatives, especially effects not seen for a quarter century. It may sell books and get academics tenure and speaking fees and air time on cable TV, but I don’t think the debateon an abortion/crime link adds much of relevance to the real world that women especially find themselves in when they need to decide between reproducing now or later or not at all. Women do the best they can in difficult circumstances. Society should be trying to make their circumstances less dire, less difficult, less fraught. Now that’s a topic for a symposium.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s