Using data from the rise and fall of communism in Romania, a new paper suggests that abortion's effect on crime is minimal, once its effect on population size is properly accounted for.
The paper, authored by economists Randi Hjalmarsson, Andreea Mitrut, and Cristian Pop-Eleches, looks at sudden changes in Romanian abortion law during and after the Ceaușescu regime. It shows that while overall numbers of crimes rose and fell in line with abortion, this effect vanishes when adjusting for the size of the population. This conclusion challenges prevailing explanations, including the idea that abortion reduces the number of "unwanted" children who might otherwise commit crimes.
The proposed link between abortion and crime has a long, and fraught, history. Back in 2001, economists John Donohue and Steven Levitt (the latter of Freakonomics fame) released a paper arguing that the legalization of abortion — first in specific states, and then nationwide via Roe v. Wade — drove the massive decline in crime from 1995 onwards, as children who would have been in their teens and early-to-mid 20s—a high-crime age range—simply were never born. They contended that this abortion effect accounts for up to 50 percent of the overall drop.
Donohue and Levitt argued that abortion affects crime through two mechanisms. One is cohort size — if there are fewer kids, there will be fewer crimes in absolute terms. The other one, that the pair described as "far more interesting," is selection, i.e. "the possibility that abortion has a disproportionate effect on the births of those who are most at risk of engaging in criminal behavior."
In other words, the legalization of abortion may have led to the abortion of children who were otherwise more likely to grow up to commit crime. This hypothesis — which implies that abortion benefits society by removing crime-prone people, who are in turn more likely to be low-socioeconomic status — has been derided by critics as eugenicist and racist.
Unsurprisingly, Donohue and Levitt's paper attracted a great deal of criticism, and a complex debate about methodological choices ensued. One of the criticisms, from economists Christopher Foote and Christopher Goetz, argues that among other issues Donohue and Levitt look only at absolute levels of crime, not rates of crime per capita.
Donohue and Levitt also recently updated their original paper, claiming similarly strong effects of abortion on crime using the same contested methodology.
According to Hjalmarsson and her team, when Foote and Goetz used crime rates, they found that "there is no evidence of a selection effect of abortion on crime, and one cannot even conclude that there is evidence of a cohort size effect." Subsequent research found selection effects for crime rates as well as levels, but conceded that "most of that relationship appears to reflect cohort size effects rather than selection."
With this debate still ongoing, the new paper makes a novel contribution by looking at another country entirely, Romania. In October of 1966, shortly after taking power, communist dictator Nicolae Ceaușescu banned abortion throughout the country. Several decades later, following Ceaușescu’s untimely demise and the end of communist rule, abortion was re-legalized. Today, Romania still has one of the highest abortion rates in Europe.
This set of discontinuities, Hjalmarsson et al. write, "provides a unique opportunity to study the effect of abortion on crime." Each legal change created large shocks to the number of births, which doubled after criminalization and fell by a third after legalization. In fact, just a year after legalization, Romania had the highest abortion rate in the world.
The basic question is, why? Did mothers of more crime-prone children abort, or were there simply fewer children overall, or was some other factor at play? To answer this question, the study authors collected data on births and later-in-life crime levels and rates, then used those data to examine how being born immediately before or immediately after a change in law changed a given birth cohort's propensity to commit crime. Because they had month-level birth data, they were able to look precisely at the effect of the timing of each change.
The results are surprising: while crime levels rose following criminalization and dropped following legalization, overall crime rates — i.e. crimes per number of cohort members — remained constant. This finding is clear over a number of different crime categories, and persists even given a variety of statistical controls.
"Our main finding is that abortion policy in Romania has had a large and significant impact on the number of crimes and hospitalizations for crime-related behaviors," the authors explain. "But this impact is proportionate to the change in the size of the population, such that there is no significant effect on crime or hospitalization rates. Moreover, this pattern of significant level but insignificant rate effects is seen across multiple crime outcomes and opposing reforms, increasing the external validity of these results."
What this means is that the entirety of the effect of abortion on crime in Romania comes from there simply being more/fewer people to commit crime pre-/post-legalization. This finding is not that surprising: taken to its extreme, all it is showing if that no children were born, the crime rate would drop to zero, because there would be no criminals. But at the same time, it means that abortion's "real" impact on crime — i.e. impact on the amount of crime given how large a country is — is fairly minimal.
This conclusion directly challenges the Donohue–Levitt idea that abortion's effect on crime works by reducing births within social groups that are more crime-prone. At least in Romania, abortion did not reduce crime by getting rid of so-called "super-predators," as Levitt alludes to in Freakonomics. Instead, it just reduced the size of society, cutting the number of victims and victimizers alike.
For the purpose of U.S. readers, there are obvious limitations to this finding. Romania is not the United States, and just because there are large cohort effects there does not mean that selection does not play a role here. The two nations' abortion regimes are also different, with abortion being legal in Romania only through the first 14 weeks, compared to 20+ weeks in the United States.
Still, the data originally used by Donohue and Levitt have been examined to death, with the conclusion that different, equally valid methodological choices can produce wildly different conclusions. Hjalmarsson et al.'s exploitation of a new, different data set should at least revitalize the discussion, especially in challenging the popular notion that abortion is a powerful anti-crime tool, rather than just a means to shrink society.