In the last several years, many causes have been enthusiastically advanced to explain the dramatic drop in crime in the United States over the past decade: innovative policing, tougher sentencing laws, the waning of the crack epidemic, better enforcement of gun laws and an improvement in the economy. Many scholars have leaped to promote a pet theory, which often coincided with a favorite policy recommendation. But one theory has come along that no one, not even the authors, is eager to embrace as policy: abortion.
At first glance, abortion as crime prevention may sound like a modern equivalent of ”A Modest Proposal,” in which Jonathan Swift, tongue in cheek, suggested that the Irish solve their famine by eating their own children. In fact, when word of the scholars’ findings was first made public two years ago, the reaction was shock and fury. Some people accused them of promoting abortion, others of supporting racist eugenics.
Now, the authors of the theory, John J. Donohue III of Stanford Law School and Steven D. Levitt, a professor of economics at the University of Chicago, are publishing their completed research for the first time next month in Harvard University’s Quarterly Journal of Economics. And they insist, however controversial its implications, that it is impossible to ignore what they say is a striking link between the introduction of legalized abortion in the early 1970’s and the drop in crime about 18 years later.
But as other experts have had their first chance to scrutinize the research in detail, almost every aspect of the theory has been attacked, from its assumptions to its conclusion that abortion and crime are connected, not separate trends that happened to surface at the same time. Indeed, given all the influences on the crime rate — including the economy, the availability of guns and drugs, and policing strategies — some critics doubt whether it is ever possible to figure out the precise relationship between the two, let alone to assert that abortion might be responsible for a 50 percent drop in crime. ”My instinct was nothing in the social sciences accounts for 50 percent of anything,” said Ted Joyce, an economist who has examined Mr. Donohue and Mr. Levitt’s data and is now about to publish his own counterstudy.
Mr. Levitt and Mr. Donohue insist their paper is purely descriptive and has no policy recommendations. Studies cited by them indicate that unwanted children are twice as likely to engage in criminal behavior as those who are wanted. ”The thesis is about as simple as you could have,” Mr. Levitt said. ”A difficult home environment leads to an increased risk of criminal activity. Increased abortion reduced unwantedness and therefore lower criminal activity.”
But that’s not all. Children of poor teenage mothers, unmarried women and black women are statistically at a higher risk for crime when they grow up. And it is these groups who have higher rates of abortion, according to the Centers for Disease Control and Prevention in Atlanta. Studies show that children of both teenage mothers and unmarried women generally, for example, are twice as likely to commit crimes; one-third of all abortions are performed on teenagers, Mr. Levitt and Mr. Donohue note.
But the critics say it’s not at all simple. As summed up by Alfred Blumstein, a leading criminologist at Carnegie Mellon University and the author of major studies on the drop in crime: ”The problem with these kinds of econometric analyses is that if a couple of the assumptions change, you can come up with very different results.”
For instance, is the assumption that unwanted children are more likely to commit crimes valid?
Still, even many skeptics concede that Mr. Levitt’s and Mr. Donohue’s study deserves serious attention; that even if they are wrong on some specifics, the subject of abortion has been wrongly ignored in discussions of crime. ”I don’t think many people think it could account for 50 percent of the reduction in crime,” said Mr. Blumstein, ”but they have convinced me it’s a factor that needs to be considered.” Indeed, the next meeting of the American Society of Criminology in November will dedicate a session to arguing both sides of the issue.
Before the Supreme Court overturned bans on abortions in 1973, about 700,000 were performed each year; the number reached 1.6 million in 1978, remained fairly constant through the 80’s and has begun to decline slightly in the last decade. ”When I first learned of the magnitude of abortions performed in the late 1970’s, I was staggered by the numbers and thought, ‘How could this not have had some dramatic social impact?’ ” Mr. Levitt says.
”When you put that together with the studies of crime among unwanted children and the fact that legalization suddenly made abortion available to millions of women who were not in the economic elite, you have a picture that screams out for further analysis,” he continued.
Mr. Levitt adds that working on the abortion issue made him lean personally toward a right-to-life position. And he says that one could easily decide, based on these findings, that improving the education and life conditions of the poor is the most effective way of fighting crime.
As he was beginning to think about the problem a few years ago, he had a talk with Mr. Donohue. ” ‘I have this wacko idea,’ I told John, and he told me he had been thinking about the same thing,” Mr. Levitt said. ”I think the thing that cemented it for us was the literature on the extreme importance of early family environment.”
Mr. Levitt and Mr. Donohue needed a way to test their hypothesis, a difficult task in the social sciences, where experiments cannot be controlled and repeated as in a laboratory. But the two thought that the way in which abortion was legalized in the United States could provide them with a good comparison.
Five states — New York, California, Hawaii, Alaska and Washington — legalized abortion three years before the Supreme Court’s decision in Roe v. Wade made it a national phenomenon. ”The five states that legalized abortion in 1970 saw drops in crime before the other 45 states,” the scholars write. ”States with high rates of abortion have experienced a roughly 30 percent drop in crime relative to low-abortion regions since 1985.” While caution is necessary in extrapolating results, they continue, ”the estimates suggest that legalized abortion can account for about half the observed decline in crime in the United States between 1991 and 1997.”
Immediately, scholars began to focus on several potential problems with the thesis. First: was the comparison valid? Hawaii and Alaska are hardly representative states; moreover, women from all over the country traveled to states like California and New York to obtain legal abortions. Mr. Blumstein points out other problems: ”Of the five states that legalized abortion, two of them, New York and California, totally dominate and skew the sample. The crack epidemic started in New York City and Los Angeles. That big city effect could be driving the crime rate up and then down.”
In other words, states with large urban populations are likely to show the strongest trends in terms of both abortion rate and crime, and a correlation between them does not prove that one causes the other. Other factors are bound to play a role.
Mr. Joyce agrees: ”Crack and the spread of guns are the great confounders. There is no question that the introduction of crack and the spread of handguns among kids 15 to 24 years old between 1985 and 1991 played a role in the rise of homicide. After that, crack begins to decline and there is a major campaign to check the spread of handguns among young people. My concern is that Levitt and Donohue are simply picking up the correlations in the huge downturn in crack and handguns.”
In fact, Mr. Joyce said, when he looked more specifically at homicides among young people in the early 1990’s in New York, the data showed the opposite of what the Donohue-Levitt thesis would have predicted. ”If their theory were correct,” Mr. Joyce said, ”you would expect to see a drop. Instead, murders hit an all-time high.”
Mr. Donohue and Mr. Levitt respond that the high murder rate was just a temporary spike created by the crack epidemic. The fact that the overall crime rate was going down even during the crack epidemic and has continued to do so since crack use abated, they say, suggests that something larger and more long-lasting is at work. Moreover, they say, their research shows a steady decline in places where there was no crack epidemic and where innovative police strategies were not put into place.
Another potential hole, said Mr. Joyce, is that no one can prove that there is a one-to-one correspondence between the number of abortions and the number of unwanted children whose birth was avoided.
But Mr. Donohue argues that the data strongly suggest that a link does exist. He said that the birth rate fell by 13 percent among teenagers, a group whose children are twice as likely to commit crimes, and that the birth rate fell by 12 percent among black women, a group whose children are nine times more prone to commit homicides, according to police and Centers for Disease Control statistics. And he says that the drop in the number of adoptions during the 1970’s from 170,000 a year to 130,000 indicates a drop in unwanted children.
But is it really true, Mr. Joyce asks, that unwanted children are twice as likely to commit crimes? ”The main study they use,” he said, ”involves children who were institutionalized for at least four months in their first year of life, which means it’s a pretty exceptional pool.” Mr. Levitt insists that this is more typical than one might think: ”A lot of poor teenage mothers have poor prenatal care and have drug and alcohol problems that lead to birth complications and hospitalization for the children.”
Such questions make it nearly impossible to pinpoint the relationship between abortion and crime, other scholars say. ”There are a lot of different things going in this period at the same time that plausibly fit these same facts,” said Lawrence W. Sherman, a criminologist at the University of Pennsylvania. ”The waning of the crack epidemic. A major crackdown on handguns in cities. An improving economy. So while I think Levitt and Donohue are brilliant, no amount of brilliance will overcome a lack of data. So maybe rather than propose new theories, we should spend our money gathering better information.”
Mr. Joyce said: ”The notion that fertility control could have an impact on child well-being is a reasonable one. But to show it you would need to study 50,000 women, half of whom terminated their pregnancies, half of whom wanted to but did not, and follow them and their children over time.
”Can you tease out that effect from crude aggregate data on crime 15 to 20 years later? I don’t think so.”
– April 14, 2001
Published at The New York Times