THE
QUARTERLY JOURNAL
OF ECONOMICS
Vol. CXVI May 2001 Issue 2
THE IMPACT OF LEGALIZED ABORTION ON CRIME*
J
OHN
J. D
ONOHUE
III
AND
S
TEVEN
D. L
EVITT
We offer evidence that legalized abortion has contributed signicantly to
recent crime reductions. Crime began to fall roughly eighteen years after abortion
legalization. The ve states that allowed abortion in 1970 experienced declines
earlier than the re st of the nation, which legalized in 1973 with Roe v. Wade.
States with high aborti on 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 ap-
pears to account for as much as 50 percent of the recent drop in crime.
I. I
NTRODUCTION
Since 1991, the United States has experienced the sharpest
drop in murder rates since the end of Prohibition in 1933. Homi-
cide rates have fallen more than 40 percent. Violent crime and
property crime have each declined more than 30 percent. Hun-
dreds of articles discussing this change have appeared in the
academic literature and po pular press.
1
They have offered an
array of ex planations: the increasing use of incarceration, growth
* We would like to thank Ian Ayres, Gary Becker, Carl Bell, Alfred Blumste in,
Jonathan Caulkins, Richard Craswell, George Fisher, Richard Freeman, James
Heckman, Christine Jolls, Theodore Joyce, Louis Kaplow, Lawrence Katz, Joh n
Kennan, John Monahan, Casey Mulliga n, Derek Neal, Eric Posner, Richard
Posner, Sherwin Rosen, Steve Sailer, Jose
´
Scheinkman, Peter Siegelman, Kenji
Yoshino, and seminar participants too numerous to mentio n for helpful comments
and discussions. Craig Estes and Rose Francis provided exceptio nally valuable
research assistance. Correspondence can be addressed to either John Donohue,
Crown Quadrangle, Stanford Law School, Sta nford, CA 94305, or Steven Levitt,
Department of Economics, University of Chicago, 1126 E. 59th Street, Chicago, IL
60637. Email: jjd@stanford.edu; slevi tt@midway.uchicago.edu.
1. For a sampling of the academic literature, see Blum stein and Wallman
[2000] and the articles appearing in the 1998 Summer issue (Volume 88) of the
Journal of Criminal Law and Criminology, especially Blumstein and Rosenfeld
[1998], Kelling and Bratton [1998], and Donohue [1998]. See Buttereld [1997a,
© 2001 by the President and Fellows of Harvard College and the Massachusetts Institute of
Technology.
The Quarterly Journal of Economics, May 2001
379
in the number of police, improved policing strategies such as
those adopted in New York, declines in the crack cocaine trade,
the strong e conomy, and incr eased expenditures on victim pre-
cautions such as security guards and alarms.
None of these factors, howeve r, can provide an entirely sat-
isfactory explanation for the large, widespread, and persistent
drop in crime in the 1990s. Some of these trends, such as the
increasing scale of imprisonment, the rise in police, and expendi-
tures on victim precaution, have been ongoing for over two de-
cades, and thus cannot plausibly explain the recent abrupt im-
provement in crime. Moreover, the widespread nature of the
crime drop argues against explanations such as improved policing
techniques since many cities that have not improved their police
forces (e.g., Los Angeles) have nonethe less se en enormous crime
declines. A similar argument holds for crack cocaine. Many areas
of the cou ntry that h ave never had a pronounced crack trade (for
instance, suburban and rural areas) have nonetheless experi-
enced substantial decreases in crime. Finally, although a strong
economy is supercially consistent with the drop in crime since
1991, previous research has established only a weak link between
economic performance an d violent crime [Fre eman 1995] and in
one case even suggested that murder rates migh t vary proc ycli-
cally [Ruhm 2000].
While acknowledging that all of these factors may have also
served to dampen c rime, we consider a novel explanation for the
sudden crime drop of the 1990s: the decision to legalize abortion
over a quarter century ago.
2
The Supreme Court’s 1973 decision
in
Roe v. Wade
legalizing abortion nationwide potentially ts the
criteria for e xplaining a large, abrupt, and continuing decrease in
crime. The shee r magnitude of the number of abortions performed
satises the rst criterion that any shock underlying the recent
drop in crime must be substantial. Seven years after
Roe v. Wade,
over 1.6 million abortions were being performed annually—al-
most o ne abortion for ever y two live births. Moreover, the legal-
1997b, 1999] for a selection of articles appearing in The New York Times and
Fletcher [2000] for a recent article in The Washington Post.
2. We are unaware of any schola rly article that has examined this effect. We
have recently learned, however, that the former police chief of Minneapolis has
written that abortion is “arguably the only effective crime-prevention device
adopted i n this nation since the late 1960s” [Bouza 1990]. In his subsequent 199 4
gubernatorial campaign, Bouza was attacked for this opinion [Short 1994]. Im-
mediately after Bouza’s view was publicized just prior to the election, Bouza fell
sharply in the polls .
380 QUARTERLY JOURNAL OF ECONOMICS
ization of abortion in ve states in 1970, and then for the nation
as a whole in 1973, were abrupt legal developments that might
plausibly have a similarly abrupt inuence 15–20 years later
when the cohorts born in the wake of liberalized abortion would
start re aching their high-crime years. Finally, any inuence of a
change in abortion would impact crime cumulatively as succes-
sive affecte d cohorts entered into their high-crime late adolescent
years, providing a reason why crime has continued to fall year
after year.
Legalized abortion may lead to reduced crime either through
reductions in cohort si zes or through lower per capita offending
rates for affected cohorts. Th e smaller cohort th at re sults from
abortion legalization means that wh en that cohort reaches the
late teens and twenties, there will be fewer young males in their
highest-crime years, and thus less crime. More interesting and
important is the possibility that children born after abortion
legalization may on average have lower subsequent rates of crim-
inality for either of two reason s. First, women who have abortions
are those most at risk to give birth to children w ho would eng age
in criminal activity. Teenagers, unmar ried women, and the eco-
nomically disadvantaged are all substantially more likely to seek
abortions [Levine e t al. 1996]. Recent studies have found childr en
born to these mothers to be at higher risk for committing crime in
adolescence [Comanor and Phillips 1999]. Gruber, Levine, and
Staiger [1999], in the paper most similar to ours, document that
the early life cir cumstances of those children on the mar gin of
abortion are difcult along many dimensions: infant morta lity,
growing up in a single-parent family, and experiencing poverty.
Second, women may use abortion to optimize the timing of child-
bearing. A given woman’s ability to provide a nurturing environ-
ment to a child can uctuate over time depending on the woman’s
age, education, and income, as well as the pre sence of a father in
the child’s life, whether the pregnancy is wanted, and any drug or
alcohol abuse both in utero and after the birth. Consequently,
legalized abortion provides a woman the opportunity to delay
childbearing if the current conditions are suboptimal. Even if
lifetime fertility remains constant for all women, children are
born into better environments, an d future criminality is likely to
be reduced.
A number of anecdotal empirical facts support the existence
and magnitude of the crime-redu cing impact of abortion. First, we
see a broad consistency with the timing of legalization of abortion
381LEGALIZED ABORTION AND CRIME
and the subsequent drop in crime. For example, the peak ages for
violent crime are roughly 18 –24, and crime starts turning down
around 1992, roughly the time at whic h the rst cohort born
following
Roe v. Wade
would hit its criminal prime. Second, as we
later demonstra te, the ve states that legalized abortion in 1970
saw drops in crime before the other 45 states and the District of
Columbia, which did not allow abortions until the Suprem e Court
decision in 1973.
Third, our more formal analysis shows that higher rates of
abortion in a state in the 1970s and early 1980s are strongly
linked to lower crime over th e period from 1985 to 1997. This
nding is true after controlling for a variety of factors that inu-
ence crime, such as the level of incarceration, the number of
police, and measures of the state’s economic well-being (the un-
employment rate, inco me per capita, and poverty ra te). The esti-
mated magnitude of the impact of legalized abortion on crime is
large. According to our estimates, as shown on Table II, states
with high rates of abortion have experienced rou ghly a 30 percent
drop in crime relative to low-abo rtion regions since 1985. While
one must be cautious in extrapolating our results out of sample,
the estimates suggest that legalized abortion can account for
about half the observed decline in crime in the United State s
between 1991 and 1997.
A number of factors lead us to believe that the link betw een
abortion and crime is causal. First, there is no relationship be -
tween abortion rates in the mid-1970s and crime changes be-
tween 1972 and 1985 (prior to the point when the abortion-
affected cohorts have reached the age of signicant criminal
involvement). Sec ond, virtually all of the abortion-related crime
decrease can be attributed to reductions in crime among the
cohorts born after abortion legalization. There is little change in
crime among older cohorts.
We should emphasize that our goal is to understand why
crime has fallen sharply in the 1990s, and to explore the contri-
bution to this decline that may have come from the legalization of
abortion in th e 1970s. In attempting to identify a link between
legalized abortion and crime, w e do not mean to suggest that such
a link is “good” or “just,” but rather, merely to show that such a
relationship exists. In short, ours is a purely positive, not a
normative analysis, although of course we recognize that there is
382 QUARTERLY JOURNAL OF ECONOMICS
an active debate about the moral and ethical implications of
abortion.
3
The structure of the paper is as follows: Section II reviews the
literature and provides a brief history of abortion. Section III
describes how the legalization of abor tion can inuence crime
rates by changing the proportion of high-risk children entering
the high-crime late adolescent years, and examines the likely
magnitude of these effects based on past research ndings. Sec-
tion IV presents the basic empirical evidence that supports the
proposed negative relationship between abortion and crime. Sec-
tion V provides eviden ce that the reduction in crim e comes pre-
dominantly from the lower crime rates of those born after the
legalization of abortion. Section VI concludes. A Data Appendix
with the so urces of all variables used in the analysis is also
provided.
II. B
RIEF
O
VERVIEW OF THE
H
ISTORY OF
L
EGALIZED
A
BORTION
Under the governing principles of English common law, abor-
tion prior to “quickening” (when the rst movements of th e fetus
could be felt, usually around the sixteenth to eighteenth week of
the pregnancy) was lawful. This co mmon law rule was in force
throughout America until the rst law in the United States
restricting abortions was adopted in New York in 1828 [David et
al. 1988, pp. 12–13]. Over the next 6 0 years, more and more states
followed the lead of New York, and by 1900 abortion was illegal
throughout the country.
The rst modest efforts at abortion liberalization began to
emerge between 1967 and 1970 when a number of states began to
allow abo rtion under limited circumstances.
4
Legal abortion be-
3. For example, Paulsen [1989, pp. 49, 76–77] considers legalized abortion to
be worse than slavery (since it involves death) and the Holocaust (since the 34
million post-Roe abortions are numerically greater than the six million Jews killed
in Europe). Despite these claims, the Sup reme Court has ruled that women have
a fundamental constitutional right of privacy to abort an early-term fetus and that
the state ca nnot unduly burden this right.
4. The 1962 amendments to the Model Penal Code provided for legal abor-
tions to prevent the death or grave impairment of the physical and mental health
of the woman, or if the fetus would be bo rn with a grave physical or mental defect
or in the case of rape or incest. These provisions were adopted in 1967 in Colorado,
North Carolina, and California, in 1968 in Florida, Georgia, and Maryland, in
1969 i n Arkansas, Kansas, New Mexico, and Oregon, and in 1970 in Delaware,
South Carolina, and Virginia—a total of thirteen states. For e xcellent reviews of
state and federal abortion s laws, see Merz, Jackson, and Klerman [1995] and Alan
Guttmacher Institute [1989].
383LEGALIZED ABORTION AND CRIME
came broadly available in ve states in 1970 when New York,
Washington, Alaska, and Hawaii repea led their antiabortion
laws, and the Supreme Court of California (ruling in late 1969)
held that the state’s law banning abortion was unconstitutional.
Legalized abortion was suddenly exte nded to the entire United
States o n January 22, 1973, with the landmark rulin g of the
United States Supre me Court in
Roe v. Wade.
The Supreme Co urt in
Roe
explicitly considered the c onse-
quences of its decision in stating:
The detriment that the State would impose upon the pregnant woman by
denying this choice altogeth er i s apparent. Specic and direct harm medi-
cally diagnosable even in early pregnancy may be involved. Maternity, or
additional offspring, may force upon the woman a distressful life and future.
Psychological harm may be imminent. Mental and physical health may be
taxed by child care. There is also the distress, for all concerned, associated
with the unwanted child, and there is the problem of bringing a child into a
family already unable, psychologically and oth erwise, to care for it.
5
The available data suggest that the number of abortions
increased dramatically following legalization, alth ough there
is little direct evidence on the number of illegal abortions
performed in the 1960s. As Figure I illustrates, the total num-
5. Roe v. Wade, 410 U. S. 110, 15 3 (1973).
F
IGURE
I
Total Abortions by Year
Source: Alan Guttmacher Institute [1992].
384 QUARTERLY JOURNAL OF ECONOMICS
ber of documented abortions rose sh arply in the wake of
Roe,
from und er 750,000 in 1973 (when live b irths t otaled 3.1 mil-
lion) to over 1.6 milli on in 1980 (when live births totaled 3.6
million).
6
If illegal abortions were already being performed in
equivalent numbers, one would not expect a seven-year lag in
reaching a steady state. Moreover, the costs of an abortion—
nancial and otherwise— dropped consid erably after legaliza-
tion. Kaplan [1988, p. 164] notes that “an illegal abortion
before
Roe v. Wade
cost $400 to $500, while today, thir teen
years af ter the decision, the now legal procedure can be pro-
cured for as lit tle as $80.”
7
The c osts of nding and traveling to
an illegal abortionist and any att endant cost of engaging in
illegal and therefore riskier and socially disapproved conduct
were also reduced by legalization.
Perhaps the most convincing evidence that legalization
increased abortion comes from Michael [1 999], who nds abor-
tion rates to be roughly an order of magni tude higher after
legalization using self-reported data on pregnancy outcome
histories. Thus, the rst prerequi site f or legal ization to ha ve
an impact on crime is met—legalization increased the rate of
abortion.
Consistent with this nding is a dramatic decline in the
number of children put up for adoption after abortion became
legal. According to Stolley [1993], almost 9 percent of premarital
births were placed for adoption before 1973; that number fell to 4
percent for births occurring between 1973 and 1981. The total
number of adoptions rose from 90,000 in 1957 to over 170,000 in
1970; by 1975 adoptions had fallen to 130,000.
6. In o ur analysis we use Alan Guttmacher Institute (AGI) data on abortions.
Although Michael [1999] argues that the AGI may substantially overstate true
abortion rates, “it is generally acknowledged [that AGI data provide] the most
accurate count of induced abortions in the United States.” Apparently, “reporti ng
is less complete for nonwhites than for whites, and overall reporting . . . has
declined over time” [Joyce and Kaestner 1996, p. 185].
7. The cost to the mother also depends on the availab ility of public funding,
which was affected by the Hyde Amendme nt, which cut off federal funding of
abortion for Medicaid recipients. The Hyde Amendm ent became law on September
30, 1976. The Hyde Amendment has been subject to a series of revisions and
restraining orders since that time. No consensus exists as to the impact of the
Hyde Amendment o n the number of abortions or births, although most recent
research suggests any impact is now s mall [Joyce and Kaestner 199 6; Kane and
Staiger 1996].
385LEGALIZED ABORTION AND CRIME
III. T
HE
M
ECHANISM BY
W
HICH
A
BORTION
L
EGALIZATION
L
OWERS
C
RIME
R
ATES
In this section we explore in detail the theoretical link be-
tween legali zation of abor tion in t he early 1970s and subsequent
drops in crime fteen to twenty years later. We identify a n umber
of alterna tive pathways through whic h abortion can affect crime.
We then generate “back -of-the-envelope” calculations as to the
likely m agnitude of the various channels based on previous re-
search ndings.
The simplest way in which legalized abortio n reduces crime
is through smaller cohor t sizes. When those smaller cohorts reach
the high-crime late adolescent years, the re are simply fewer
people to commit crime. Levine et al. [1996] nd that legalization
is associated with roughly a 5 percent drop in birth rates.
8
As-
suming that the fall in births is a random sample of all births,
total crime comm itted by this cohort would be expected to fall
commensurately.
Far more interesting from our pers pective is the possibility
that abortion has a disproportionate effect on the births of those
who are most at r isk of engaging in cr iminal behavior.
9
To the
extent that abortion is mor e frequent among those parents who
are least willing or able to provide a nurturing home environ-
ment, as a large and growing body of eviden ce suggests, the
impact of legalized abortion on crime might be far grea ter than its
effect on fertility rates.
10
This is particularly tr ue given that 6
percent of any birth cohort will commit roughly half the crime
8. This decline is broadly consistent with survey responses by moth ers in
1973 who report that approximately 13 pe rcent of lifetime births were unwanted
[Statistical Abstract of the United States 1980, p. 65, table 99]. Note, however,
that the decline in births is far less than the number of abo rtions, suggesting that
the number of conceptions increased substantially—an example of insurance
leading to moral hazard. The insurance that abortion provides against unwanted
pregnancy induces more sexual conduct or diminished protections against preg-
nancy in a way that substanti ally increases the number of pregnancies. Another
possible explanation for the gap between aborti on rates and fertility rate changes
is th at ille gal abortion was already suppressing the birth rate by 15–20 percent
and legalization reduced it another 5–10 percent, but th is would imply a higher
gure for the number of illegal abortions than we think is likely, as discussed
above.
9. As noted earlier, this effect can occur either because of lower lifetime
fertility rates among high-risk groups, or because women delay childbearing until
conditions are more favorable for successfully raising children.
10. In addition, with an estimated number of over 150,000 rapes in 1973
(often thought to be a conservative estimate), it is possible that 10,000 to 15,000
conceptions occurred that year as a result of rape, and one might expect a
substantial proportion of these high-risk conceptions would end in abortion [Bu-
reau o f Justi ce Statistics 1985, p. 230, Table 3.2].
386 QUARTERLY JOURNAL OF ECONOMICS
[Wolfgang, Figlio, and Sellin 1972; Tracy, Wolfgang, and Figlio
1990].
11
Prior to the legalization of abortion, there was a very strong
link between the number of unwanted births and low maternal
education over the period from 1965 through 1970 [Commission
on Populatio n Growth and the American Future 1972, p. 98].
Levine et al. [1996] found that the drop in births associated with
abortion legalization was not uniform across all groups. They
estimated that the drop in births was roughly twice as great for
teenage and nonwhite mothers as it was for the nonteen, white
population.
12
In the years immediately following
Roe v. Wade,
data from the Centers fo r Disease Control [1994] indicate that
almost one-third of abortions were performed on teenagers. An-
grist and Evans [1996] found that while abo rtion reforms had
relatively modest effects on the fertility of white women, “black
women who were exposed to abortion reforms experienced large
reductions in teen fertility and teen out-of-wedlock fertility.”
A number of studies have shown that the availability of
abortion improves infant outcomes by reducing the number of low
birthweight babies and neonatal mortality [Grossman and Jaco-
bowitz 1981; Corman and Grossm an 1985; Joyce 1987; Grossman
and Jo yce 1990]. Moreover, Gruber, Levine, and Staiger [1999, p.
265] conclude that “the average living circumstances of co horts
born immediately after abor tion became legalized improved sub-
stantially relative to preceding cohorts.” They go on to note that
“the marginal children who were not born as a result of abortion
legalization would have systematically been born into less favor-
able cir cumstances if the pregnan cies had not been terminated:
they would have been 60 percent more likely to live in a single-
parent household, 50 percent more likely to live in poverty, 45
percent more likely to be in a house hold collecting welfare, and 40
percent more lik ely to die during the rst year of life.”
Previous research has found that an adverse family environ-
ment is strong ly linked to future criminality. Both Loeber and
11. The high concentration rates of crime among a relatively small number of
offenders makes it more likely that legalized abortion would have larger effects on
crime than on other social outcomes such as high school dropo ut rates or unem-
ployment rates. A given child who has failed to complete schoo l or secu re a job
counts as only one event in measuring school dropout or unemployment rates.
Conversely, a single child may commit hundreds of crimes and thereby contribute
far more powerfully to a higher crime rate.
12. This is not surprising since in the late 1960s the “pill” and other birth
control mechanisms were far more readily available to married, educated, and
afuent women [Goldin and Katz 2000].
387LEGALIZED ABORTION AND CRIME
Stouthamer-Loeber [1986] and Sampson and Laub [1993] present
evidence that a varie ty of unfavorable parental behaviors (e.g.,
maternal rejection, erratic/harsh behavior on the part of parents,
lack of parent al supervision) are among the best predictors of
juvenile de linquency. Raine, Br ennan, and Medick [1994], and
Raine et al. [1996] argue that birth complications combined with
early maternal rejection predispose boys to violent crime at age
eighteen. Rasanen et al. [1999] nd that the risk of violent crime
for Finnish males born in 1966 is a function of (in descending
order of impact): mother’s lo w education, teenage mother, single-
parent family, mother did not want pregnancy, and mother
smoked during pregnancy. It is possible that abortion could re-
duce the number of children born under all these circumstances:
teenagers who have ab ortions can get more e ducation before they
give birth and may delay childbearing until they are married or
want a c hild or both. In addition, women who inadvertently
become pregnant may have engaged in behavior such as smoking,
drinking, or using drugs that elevate the prospect of future crim-
inality of their offspring.
A number of studies have looked at cases of women, living in
jurisdictions in which governmental approval to have an abortion
was required, who sought to have an abortion, but were denied
the right to do so [David et al. 1988; Posne r 1992, p. 283].
13
Dagg
[1991] reports that these women overwhelmingly kept their ba-
bies, rather than giving them up for adoption, but that they often
resented the unwanted children and were far less likely than
other mothers to nurture, hold, and breastfeed these children. In
an array of studies in Eastern Europe and Scandinavia, Dagg
found that the children who were born because their mothers
were denied an abo rtion were substantially more likely to be
involved in crime and have poorer life prospects, eve n when
controlling for the income, age, education, and health of the
mother. This literature provides stron g evidence that unwanted
children are like ly to be disproportionately involved in criminal
activity, which may be the causal pathway from greater avail-
ability of a bortion to lower rates of crime.
Evidence from prisoner surveys further reinforces the link
between a difcult home environment as a child and later crim-
13. David et al. [1988] review the ndings of separate studies of the effects of
denied abortion for cohorts born in Goteberg, Sweden in 1939–1942, Stockholm in
1948, all of Sweden in 1960, and Prague in 1961–1963.
388 QUARTERLY JOURNAL OF ECONOMICS
inality [Beck et al. 1993]. In 1991, 14 perce nt of prisoners re-
ported growing up with neither parent present , and 43 percent
reported having only one parent (compared wi th 3 percent and 24
percent, respectively, for the ove rall population). Thirty-eight
percent of prisoners report that their parents or guardians
abused alcohol or dr ugs; almost one-third of female inmates re-
port being sexually abused before the age of eighteen.
A. The Expected Magnitude of the Impact of Abor tion
Legalization on Crime
Before presenting our empirical estimates in the next sec-
tion, we present “back-of-the-envelope” estimates of the plausible
magnitude of the impact of legalized abortion on crime. Previous
researchers have studied (1) how legalized abortion affects birth
rates across different groups, and (2) crime rates across groups.
By combin ing these two sets of estimates, we can obtain a crude
prediction of the impact of legalized abortion on crime.
This ana lysis considers four factors: race, teenage mother-
hood, unmarried motherhood, and unwantedness. Beginning
with the rst three of these factors, we use the 1990 Census to
determine the proportion of children in each of the eight possible
demographic categories (e.g., white children born to teenage
mothers growing up in a single-parent household, or black chil-
dren born to nonteenage mothers growing up in two-parent
households). We then use the estimates of Levine et al. [1996] to
determine what those proportions might have been in the absence
of legalized abortion. Using Rasanen et al. [1999] and observed
frequencies of crime by race in the United States, we generate
category-specic crime rates corresponding to each of the eight
cells. Combining these crime rates with the cha nge in th e number
of births in each category due to abortion provides an estimate of
the hypothetical reduction in crime. Finally, under the assump-
tion that 75 percent of unwante d births are aborte d (this number
appears consisten t with data from self-reported pregnancy histo-
ries), we estimate the contribut ion to lower crime from fewer
unwanted births.
14
It is important to no te that our calculations
below isolate the
marginal
contribution of race, teenage mother-
hood, unmarried motherhood, and unwantedn ess. Thus, when
14. A full description of the assumptions and calculations is available from
the authors on request.
389LEGALIZED ABORTION AND CRIME
computing the impact of race, we net out any racial differences in
those other characteristics in order to avoid double counting.
The results of this exercise for homicide are as fo llows. All
values reported are the hypothetical reduction in total homicides
committed by membe rs of a given cohort. Thro ugh a purely me-
chanical relationship, the 5.4 percent overall postlegalization de-
cline in cohort size obtained by Levine et al. [1996] translates into
a 5.4 percent reduction in homic ide.
Fertility declines for black women are three times greater
than for whites (12 percent compared with 4 percent). Given that
homicide rates of black youths are roughly nine tim es high er than
those of white youths, racial differences in the fertility effects of
abortion are likely to translate into greater hom icide reductions.
Under the assum ption that those black and white births elimi-
nated by legalized abortio n would have experienced the average
criminal propensities of their respective races, then the predicted
reduction in homicide i s 8.9 percent. In other words, taking into
account differential abortion rates by race raises the predicted
impact of abortion legalization on homicide from 5.4 percent to
8.9 percent.
15
Teenagers and unwed women experienc e reductions in fer-
tility of 13 and 7 percent, respectively, well above that for non-
teenage, married women. Rasanen et al. [1999] nd, after con-
trolling for other characteristics, that having a teenage mother
roughly double s a child’s propensity to commit crime, as does
growing up with a single parent.
16
Accounting for these two
factors raises the estimated impact of abortion on homicide from
8.9 percent to 12.5 percent.
Adjusting for unwantedness, which more than doubles an
individual’s likelihood of crime based on the estimates of Rasanen
et al. [1999], raises the estimates from 12.5 percent to 18.5 per-
cent. The impact of unwantedness is large because abo rtion rates
of unwanted pregnancies are very high, whereas wanted pr eg-
nancies are by denition not aborted.
Thus, using past estimates in the literature, we crudely
estimate that crime should fall by 18.5 percent in coho rts that
15. For other crimes, the impact o f race is much lower because rates of
offending and victimi zation are much more similar across races.
16. Comanor and Phillips [1999], using the Na tional Longitudinal Survey of
Youth, nd that adolescents in househol ds with absent fathers are 2.2 times more
likely to be charged with a crime as a juve nile, controlli ng for other observable
factors. That estimate is very close to the Rasanen et al. [1999] nding for Finnish
males that we use in our calculations.
390 QUARTERLY JOURNAL OF ECONOMICS
have access to legalized abortion. As of 1997, roughly 60 percent
of crimes we re committed by individuals bo rn afte r legalized
abortion, implying that (thus far) the hypothetical impact of
abortion on crime is only 60 percent of the impact on affected
cohorts, or about an 11 percent reduction. To the extent that other
factors are correlated with both criminal propensities and abor-
tion likelihoods (e.g. , poverty, maternal education, religiosity),
this rough estimate is likely to understate the true impact.
17
Given th at the observed declines in crime in the 1990s are 30 – 40
percent, abortion may be an important factor in explaining the
crime drop. In the next section we present empirical estimates of
the impact of abortion on crime that are roughly consistent wit h
these hypothetical c alculations.
IV. E
MPIRICAL
E
VIDENCE OF
L
EGALIZED
A
BORTION
A
FFECTING
C
RIME
R
ATES
We begin our em pirical analysis by establishing a relation-
ship between crime changes in the 1990s and legalized abortion
in the early 1970s. We consider three diffe rent sources of varia-
tion: the national time ser ies of crime and abortion, differential
crime patterns across early legalizers and other states, and the
impact of state abortion rates (properly lagged) on state crime
rates. In Section V we focus on arrest rates, which allows us to
decompose the effect of abortion by the age o f o ffenders.
A. National Time Series
Figure II presents per capita crime rates for the United
States for violent crime, property crime, and murder for the
period 1973–1999, as measured in the Uniform Crime Reports
compiled by the Federal Bure au of Inv estigation.
18
Between 1973
17. These estimates will understate the true impact of abortion on crim e if
there are other factors beyond the four we explicitly considered that positively
covary with abortion and crime, such as religiosity, poverty, or low maternal
education. Indeed, this last factor was found by Rasanen et al. [1999] to be the
single most powerful factor leadi ng to criminality by the children. Moreover, to the
extent that abortion reduces crime committed by other family members as a result
of the benecial effects of a reduction in family size (since larger family size
increases the likelihood of criminality), this effect would also be missed. On the
other hand, a countervailing force is that a reduction in the supply of criminals
will induce higher returns to entry into the criminal occupations thereby offset-
ting through recruitment the initial dampening effect on crime. One would sus-
pect this effect to be limited to crimes involving active markets for illega l sub-
stances (drugs) or services (prostitution).
18. Uniform Crime Reports compile the number of crimes reported to the
391LEGALIZED ABORTION AND CRIME
and 1991, violent crime nearly doubled, property crime increased
almost 40 percent, and murder was roughly unchanged (despite
substantial uctuations in the intervening yea rs). The ye ar 1991
represents a local maximum for all three of the crime measure s.
Since that time, each of these crime categories has steadily fallen.
Murder has fallen by 40 percent and the other two categories are
down mor e than 30 percent.
The National Crime Victimization Surv ey (NCVS), which
gathers informati on on self-reported crime victimi zations, offers
another perspective on national crime patterns in Figure III.
According to vi ctimization surveys, violent crime fell through the
early 1980s, increased fro m that point until 1993, and fell sharply
thereafter. Property cr ime fell throughout the period 19 73 to
1991, and began to fall e ven more quickly thereafter. Th e crime
declines in the 1990s are even greater using victimizatio n data
than the repo rted crime statistics. It is notable that the longer
time-series pattern s of UCR and victimization data do not match
police in various cri me categories each year. Whil e the potential shortcomings of
these data are we ll recognized (e.g., O’Brien [1985]), they remain the only source
of geographically disaggregated crime data available in the United States.
F
IGURE
II
Crime Rates from the Uniform Crime Reports, 1973–1999
Data are national aggregate per capita reported violent crime, property crime,
and murder, i ndexed to equal 100 in the year 1973. All data are from the FBI’s
Uniform Crime Reports, published annually.
392 QUARTERLY JOURNAL OF ECONOMICS
closely, yet both demon strate a distinct break from trend in the
1990s.
The timing of the break in the national crime rate is consis-
tent with a legalized abortion story. In 1991 the rst cohort
affected by
Roe v. Wade
would have been roughly sevente en years
old, just beginning to enter the highest crime adolescent years.
19
In the early-legalizing states (in which slightly more than 20
percent of all Americans reside), the rst cohort affected by le-
galized abort ion would have been twenty years of age, roughly the
19. The Supreme Court handed down the decision in Roe v. Wade on January
22, 1973. Typicall y, there is a six-to-seven-month lag between the time that an
abortion would be performed and the time that the birth would have occurred.
Thus, the rst births affected would be those born in late 1973.
If women who already had children in 1973 used aborti on to prevent increases
in family size, then abortion may indirectly lower criminality for the remaining
children who will receive greater per child contributions of parental resources
[Becker 1981; Barber, Axinn, and Thornton 1999]. Sampson and Laub [1993, p.
81] and Rasanen et al. [1999] nd that family size signicantly increases delin-
quency. Note that this family size effect suggests that criminality could be reduced
for chil dren who were born a number of years in advance of any abortion that
prevents further increases in family size, and thus would allow the effect of
abortion on crime to be observed prior to the time that the direct effect of abortion
would b e observed.
F
IGURE
III
National Crime Victimization Survey, 1973–1998
Data are national aggregate per capita violent crime and property crime vic-
timizations, indexed to equal 100 in the year 1973. All data are based on the
National Crime Survey, conducted annually. Data have been adjusted to correct
for a one-time shift associated with the redesign of the survey in the e arly 1990s.
393LEGALIZED ABORTION AND CRIME
peak of the age-crime prole [Blumstein et al. 1986; Cook and
Laub 1998].
The continual decrease in crime between 1991 and 1999 is
also consistent with the hypothesi zed effects of abortion. With
each pa ssing year, the fraction of the criminal population that
was born postlegalization increases. T hus, the impact of abortion
will be felt only gradually. To formalize this idea, we dene an
index that is designed to reect the effect o f all previous abortions
on crime in a particular year
t
. Obviously, recent abortions will
not have any direct impact on crime today since infants commit
little crime. As the postlegalization cohorts age, however, we can
estimate the effect of abortion by seeing how much crime (proxied
by the percentage of arrests committed by those of that age) is
committed by the particular cohort. Thus, we dene the “effective
legalized abortion rate” relevant to crime in year
t
as the
weighted average legalized abortion rate across all cohorts of
arrestees, i.e.,
(1)
Effective
_
Abortion
t
5
O
a
Abortion
*
t2 a
(
Arrests
a
/
Arrests
to tal
)
,
where
t
indexes years and
a
indexes the age of a cohort.
Abortion
is the number of abortio ns per live birth, and the ratio of arrests
inside the parentheses is the fractio n of arrests for a given crime
involving members of cohort
a
. In a steady state with all cohorts
subjected to the sam e abortion rate, the effective abor tion rate is
equal to the actua l abortion rate. For many years foll owing the
introduction of legalized abortion, the effective abortion rate will
be below th e actual abortion rate since many active criminal
cohorts are too old to have been affected by legalized abortion. For
instance, fo llowing
Roe v. Wade,
the actual abortion rate (per
1000 live births) rose to a steady state of about 400. Yet we
estimate that the effective abortion rate in 1991 was only about
33 for homicide, 63 for vio lent crime, and 126 for property crime.
Because prope rty crime is disproportionately done by the young,
the effect of abortion legalization is felt earlier.
20
The effective
rates grew steadily, risin g to 142, 180, and 252, re spectively, by
1997. If legalized abortion reduces crime, then cr ime should con-
tinue to fall (all else equal) as long as the effective abortion rate
20. Details of this calculation are avail able from the authors. T his effective
abortion rate includes legal abortion exposure prior to 1973 in the ve states that
legalized in 1970.
394 QUARTERLY JOURNAL OF ECONOMICS
is rising, precisely the pattern observed in actual crime data in
Figures II and III.
21
B. Comparing Crime Trends in Early-Legalizing States versus
the Rest of the United States
As noted earlier in the paper, ve states (Alaska, California,
Hawaii, New York, and Washington) legalized or quasi-legalized
abortion around 1970; in the remaining states, abortion did not
become legal until 1973. The staggered timing of the introduction
of legalize d abortion provides a potential avenue for assessing its
impact.
22
Using this source of variation to explore the co nse-
quences of abortion legalization, Levine et al. [1996] analyze the
fertility effects; Angrist and Evans [1996] study the impact on
female labor supply ; and Gruber , Levine, and Staiger [1999]
examine the effect on a variety of measures of child welfare.
For the purposes of analyzing crime, the comparison of early
legalizers to all other states is less than ideal. First, criminal
involvement does not jum p or fall abruptly with age, but rather
steadily increases through the teenage years before eventually
declining. Early-legalizing states only have a three-year head
start. Thus, it may be difcult to identify an impact on overall
crime rates since even in the peak crime ages three c ohorts
account for less than 20 percent of overall arrests. Second, states
that legalized abortion in 1970 continue d to have higher abortion
rates even after
Roe v. Wade.
For instance, in 1976, three years
after
Roe v. Wade
was handed down, the early-legalizing states
21. It is worth noting one ostensible inconsistency between our predictions
and the disaggregated time-series data. As noted by Cook and Laub [19 98] and
Blumstein and Rosenfeld [1998], there was a sharp spike in youth hom icide rates
in the late 1980s and early 1990s, especially among Afri can-Americans. These
cohorts were born after legalized abortion. Importantly, this nding is not incon-
sistent with the central claim that abortion legalization contributed to lower crime
rates, but merely shows that this dampening effect on crime can be outweighed in
the short term by factors that stimulate crime. Elevated youth homicide rates in
this period appear to be clearly linked to the rise of crack and the easy availability
of guns. That abortion is only one factor inuencing crime in the late 1980s points
out the caution required in drawing any conclusions regarding an abortion-cri me
link based on time-series evidence alone.
22. Evidence in Levine et al. [1996] suggests that there was a substantial
amount of border cro ssing in orde r to ob tain legal abortions prior to 1973. To the
extent that is true, the observed differences in crime between early-legalizing
states and all others will be muted. It appears, however, that the more afuent
tended to travel for abortions, which probably diminishes the importance of such
activity for assess ments about crime. Some evidence of this is se en in the fact that
abortions performed in New York on white women were cut in half in the wake of
the decision in Roe v. Wade, but the re was a far smaller drop in the number of
abortions performed in New York on black women.
395LEGALIZED ABORTION AND CRIME
had a 1985 population-weighted average rate of 593 abortions per
live births, compared with 308 for all other states. Given that the
impact of abortion on crime happens only gradually, it is difcult
to disentangle the separate impacts of early legalization and
higher steady state abortion rates.
23
Bearing in mind these important caveats, a comparison of
crime trends in early-legalizing and all other states is displayed
in Table I, as well as the difference between those two values. For
each of three crime categories (violent, pr operty, murder), we
present percent changes in crime by six-year periods for the years
1976 –1994, and for the period 1994 –1997. The bottom panel of
the table also presents the effective abortion rate for violent crime
for the two sets of states at the end of each time period, computed
using equation (1).
24
Prior to 1982, legalized abortion should have no impact on
crime since the rst cohort affec ted by abortion is no more than
twelve years old. These years are included as a check on any
preexisting trends in crime rates across the two sets of states. As
Table I shows, these preexisting trends are not statistically dif-
ferent across early-legalizing and all other states, nor is the
relative pattern constant across the three crime categories. Both
property and violent crime w ere increasing at a slower rate in
early legalizing states between 1976 and 1982, whereas murder
was rising faster in early-legalizing states.
As shown in the bottom panel of Table I, by 1988 the effective
abortion rate for violent crime in early-legalizing states was
64.0 compared with 10.4 in the rest of the Un ited States. To
explore whether crime rates began to respond to early abortion
legalization between 1982 and 1988, look at the rows labeled
“Difference” in the 1982–1988 colum n. A negative sign for th is
difference suggests that crime fell faster in the states that le-
galized abortion earlier (consistent with the theory of this pa-
per), while a positive sign suggests the opposite. Here we see
the evidence of the impact of early legalization for the 1982–
23. From the broader perspective of determining whether crime rates re-
spond to abortion, this distinction may be irrelevant. However, the inability to
distinguish the two channels of impact lessens the extent to which a comparison
of early legalizers to other states re presents a distinct source of variation from the
regression analysis using abo rtion rates across states after 1973.
24. The effective abortion rate for violent crime falls between the correspond-
ing measure s for property crime and homicide. The pattern of di fferences is
similar for the other crime categories, except that the gap rises more (less) quickly
for property crime (homicide).
396 QUARTERLY JOURNAL OF ECONOMICS
1988 period is mixed. Property cr ime fell signicantly in early-
legalizing states relative to the rest of the United States (
2
9.8
percentage points), and the difference is more than twice as large
as the preexisting trend in the rst column. There is no apparent
impact on violent crime or murder by 1988. Nonetheless, the
earlier impact on property crime is consistent with the fact that
offenses comm itted by the very young are dispropor tionately con-
centrated in property crime. For instance, in 1995 those under
age eighte en accounted for over one-third of all property crim e
arrests, but less than 20 percent of violent crime and murder
arrests.
TABLE I
C
RIME
T
RENDS FOR
S
TATES
L
EGALIZING
A
BORTION
E
ARLY VERSUS
THE
R
EST OF THE
U
NITED
S
TATES
Crime category
Percent change in crime rate over the period
Cumulative,
1982–19971976–1982 1982–1985 1988–1994 1994–1997
Violent crime
Early legalizers 16.6 11.1 1.9 2 25.8 2 12.8
Rest of U. S. 20.9 13.2 15.4 2 11.0 17.6
Difference 2 4.3 2 2.1 2 13.4 2 14.8 2 30.4
(5.5) (5.4) (4.4) (3.3) (8.1)
Property crime
Early legalizers 1.7 2 8.3 2 14.3 2 21.5 2 44.1
Rest of U. S. 6.0 1.5 2 5.9 2 4.3 2 8.8
Difference 2 4.3 2 9.8 2 8.4 2 17.2 2 35.3
(2.9) (4.0) (4.2) (2.4) (5.8)
Murder
Early legalizers 6.3 0.5 2.7 2 44.0 2 40.8
Rest of U. S. 1.7 2 8.8 5.2 2 21.1 2 24.6
Difference 4.6 9.3 2 2.5 2 22.9 2 16.2
(7.4) (6.8) (8.6) (6.8) (10.7)
Effective abortion rate
at end of period
Early legalizers 0.0 64.0 238.6 327.0 327.0
Rest of U. S. 0.0 10.4 87.7 141.0 141.0
Difference 0.0 53.6 150.9 186.0 186.0
Early legalizing states are Alaska, California, Hawaii, New York, and Washington. These ve states
legalized abortion in late 1969 or 1970. In the remaining states, abortion became legal in 1973 after Roe v.
Wade. Percent change in crime rate is calculated by subtracting the xed 1985 population-weighted average
of the natural log of the crime rate at the beginning of the period from th e xed 1985 population-weighted
average of the natural log of the crime rate at the end of the period. The rows labeled “Difference” are the
difference between early legalizers and the rest of the United States (standard errors are reporte d in
parentheses). The bottom panel of the table presents the effective abortion rate for violent crime, as
calculated using equation (1) in the tex t, based on the observed age distribution of national arrests for violent
crime in 1985. Entries in the table are xed 1985 population-weighted averages of the states. Abortion data
are from the Alan Guttmacher Institute; crime data are from Uniform Crime Reports. Because of missing
crime d ata for 1976, the 1976–1982 calculations omit the District of Columbia. Precise data sources are
provided in the Data Appendix.
397LEGALIZED ABORTION AND CRIME
By 1994, the gap in the “effective abortion rate” be tween
early-legalizing states and all othe rs had grown to 150.9. The
early-legalizing states experienced declines in crime relative to
the rest of the United States in all three crime categories. The
trend accelerates between 1994 and 1997, with double-digit (and
highly statistically signicant) differ ences for each of the crimes.
The last column of Table I shows that the cumulative decrease in
F
IGURE
IVa
Changes in Violent Crime and Abortion Rates, 1985–1997
F
IGURE
IVb
Changes in Property Crime and Abortion Rates, 19 85–199 7
398 QUARTERLY JOURNAL OF ECONOMICS
crime between 1982–1997 for early-legalizing states compared
with the rest of the nation is 16.2 percent great er for murder, 30.4
percent greater for violent crime, and 35.3 percent gre ater for
property crime. Realistically, these crime decreases are too large
to be attributed to the three-year head st art in the early-legaliz-
ing states. Put another way, the observed differences in the
“effective abortion rate” documented in the bottom of Table I
reect not only the head start on abortion, but also higher steady
state rates. Thus, the so urce of variation exploited in Table I is
not entirely distinct from that used in the state-level panel re-
gressions below.
C. State-Level Changes in Crime as a Function of
Postlegalization Abortion Rates
The preceding discussio n provides sug gestive evidence of an
impact of abortion on crime. In what follows, we explore this
relationship more systematically by using a panel data analysis
F
IGURE
IVc
Changes in Murder and Abortion Rates, 1985–1997
The vertical axis in Figures IVa–IVc corresponds to the log change in the named
crime category between 1985 and 1997. The horizontal axis is the change in the
effective abortion rate corresponding to the crime category between 1985 and
1997. The effective abortion rate is the estimated average abortion rate per 1000
live births for criminals in the state, as calculated using equation (1) in the text.
Washington, DC, which is an extreme outlier with respect to abortion rates, i s
omitted from the gures, but is included in all other statistical analyses.
399LEGALIZED ABORTION AND CRIME
to relate state abortion rates after
Roe v. Wade
to state-level
changes in crime over the period from 1985 through 1997.
Before presenting regression results, Figures IVa– IVc show
simple plots of log-changes in crime rates between 1985 and 1997
against the change in the s tate-level effective abortion rate over
that same time period.
25
The three gures co rrespond to violent
crime, property crime, and murder, respect ively. In each case,
there is a clear negative relationship between crime changes over
the period 1985–1997 and abortion rates in the years imm edi-
ately following
Roe v. Wade.
The tted population-we ighted re-
gression lines are also included in the gures. The
R
2
from these
simple regressions range from .12 (murder) to .45 (property
crime), as reected in the relatively tighter t of the regression
line for the latter crime category.
The raw relationship between abortion rates in the 1970s
and falling crime in the 1990s emerges even more clearly in Table
II. State s are ranked based on effective abortion rates in 1997 and
25. The gures plot the scatter diagrams for all 50 states. The District of
Columbia is dropped from the graph, as it is an extreme outlier that does not
accurately reect the abortion rates of D.C. residents, as indicated in footnote 27,
below. All states had effective abortion rates close to zero in 1985, so the change
in the effective abortion rate between 1985 and 1997 is almost identical to the
effective abortion rate in 1997.
TABLE II
C
RIME
C
HANGES
1985–1997
AS A
F
UNCTION OF
A
BORTION
R
ATES
1973–1976
Abortion
frequency
(Ranked by
effective
abortion rate
in 1997)
Effective
abortions
per 1000
live births,
1997
% Change in crime rate,
1973–1985
% Change in crime ra te,
1985–1997
Violent
crime
Property
crime Murder
Violent
crime
Property
crime Murder
Lowest 67.5
+
31.8
+
29.8 2 21.1
+
29.2
+
9.3
+
4.1
Medium 135.0
+
28.8
+
31.1 2 19.7
+
18.0
+
2.2 2 12.6
Highest 257.1
+
32.2
+
15.2 2 9.7 2 2.4 2 23.1 2 25.9
States are ranked by effective abortion rates for violent crime in 1997, with the seventeen states with
lowest abortion rates classied as “lowest,” the next seventeen states classied as “ medium,” and the highest
seventeen states (including District of Columbia) classied as “highest.” The effe ctive abortion rate is the
estimated average abortion rate per 1000 live births for criminals in the state, as calculated using e quation
(1) in the text, based on the observed age distribution of national arrests for violent crime in 1985. All values
in the table are weighted averages using 1985 state populations as weights. Percent change in crime per
capita is calculated by su btracting the xed 1985 population-weighted average of the natural log of the crime
rate at the beginning of the period from the xed 1985 population-weightedaverage of the natural log of the
crime rate at the end of the period. Because crime rates are extremely low until the midteenage years,
legalized abortion is not predicted to have had a substantial impact on crime ove r the period 1973–1985, but
would be predicted to affect crime in the period 1985–1997. Abortion data are from the Alan Guttmacher
Institute; crime data are from Uniform Crime Reports. Precise data sou rces are provided in the Data
Appendix.
400 QUARTERLY JOURNAL OF ECONOMICS
divided into th ree categories: low, medium, and high. Mean ef-
fective abortion rates, and percent changes in murder, vio lent
crime, and property crime for the periods 1973–1985 and 1985–
1997 are shown in the table for the three sets of states. Crime
data for the period 1973–1985 are included as a check on the
validity of the results. There should be no effect of abort ion on
crime between 1973–1985. To the extent that h igh and low abor-
tion states systematically differ in the earlier period, questions
about the exogeneity of the abortion rate are raise d. It is reas-
suring that the data reveal no clear differences in crime rates
across states between 1973 and 1985 as a functio n of the abortion
rate. In some instance s crime was rising more quickly in high
abortion states; in other cases the opposite is true. F or the period
1985–1997, however, the results change dramatically. For each
crime category, the high abortion states fell relative to the low
abortion states by at least 30 percentage points. In every in-
stance, the medium abortion states had intermediate outcomes
with respect to crime.
The panel data regressions that we report are similar in
spirit to Figure IV and Table II, but utilize no t only the endpoints
of the sample, but also information from the intervening years, as
well as including a range of contro ls:
(2) ln
(
CRIME
st
)
5 b
1
ABORT
st
1 X
st
Q 1 g
s
1 l
t
1 e
st
,
where
s
indexes states and
t
reects time. The left-hand-side
variable is the relevant logged crime rate per capita. Our measure
of abortion is the effective abortion rate (dened earlier) for a
given state, year, and crime cate gory.
26
X
is a vector of state-level
controls that includes prisoners and police per capita, a range o f
variables capturing state economic conditions, lagged state we l-
fare gene rosity, the presence of co ncealed handgun laws, and per
capita beer consumption.
g
s
and
l
t
represent state and ye ar xed
effects. All regressions are weig hted least squares with weights
based on state populations. All of the estimates we present are
adjusted for s erial correlation in panel data using th e method of
Bhargava et al. [1982].
27
26. The weights used in compu ting the effective abortion rates are the per-
centage of arrests by age fo r a given crime category in the United States in 1985.
In other words, abortion rates are state-specic, but the same weigh ting function
is us ed for all states.
27. Blank, George, and London [1996] suggest that the ofcial abortion rate
in Washington, DC is articially elevated beca use women from Maryland and
Virginia frequently travel there to receive abortions. The CDC e stimates that
401LEGALIZED ABORTION AND CRIME
Summary statistics for the sample are provided in Table III.
The summary statistics on abortion correspond to the effective
abortion rate, which is well below the actual a bortion rate
throughout the sample because much of the criminal po pulation
was born prior to legalized abortion. Act ual national abortion
rates in the years imm ediately after
Roe v. Wade
were roughly
300 abortions per 1000 live births, but with considerable varia-
tion across state s. For example, over the period from 1973–1976,
West Virginia had the lowest abortion rate (10 per 1000 live
births), while New York (763) and Washington, D.C. (1793) had
the highest rates. There is a great deal of variation in crimes per
1000 residents, both across states and within states over time .
The same is true for arrest rates.
An important limitation of the data is that state abortion
rates are very highly serially correlated. The correlation between
state abortion rates in years
t
and
t
+
1 is .98. The ve-year and
ten-year correlations are .95 and .91, respectively. One implica-
tion of these high correlatio ns is that it is very difcult using the
data alone to distinguish the impact o f 1970s abortions on current
crime rates from the impact of 1990s abortions on current crime
rates; if one includes both lagged and current abortion rates in
the same specication, standard errors explode due to multicol-
linearity. Con sequently, it must be rec ognized that our interpre-
tation of the results relies on the assumption that there will be a
fteen-to-twenty year lag before abortion materially affects
crime. This lag between the act of abortion and its impact on
crime diffe rentiates it from many other social phenomena like
divorce and poverty which may have both lagged and contempo-
raneous effects, making it very difcult to separately identify any
lagged effects.
Regression results are shown in Table IV. Fo r each of the
three crime categories, two different specications are reported.
The odd-num bered columns present results without control vari-
ables (other than the state- and year-xed effects); the even
columns add the full set of controls.
The top row of the table presents the coefcients on the
abortion variable across specication s. In all six cases, the coef-
cient is negative, implying that higher abortion rates are asso-
about half of all abortions performed in the District of Columbia are on nonresi-
dents (which is the highest percentage for any state); the comparable percentage
in New Jersey is 2 percent [Dye and Presser 1999, p. 143].
402 QUARTERLY JOURNAL OF ECONOMICS
ciated with declining crime. These estimated effects of abortion
are highly statistically signicant—more so than any other vari-
able included in the analysis. The real-world magnitude implied
TABLE III
S
UMMARY
S
TATISTICS
Variable Mean
Standard
deviation
(overall)
Standard
deviation
(within state)
Violent crime per 1000 residents 6.73 2.81 .88
Property crime per 1000 residents 48.04 11.46 4.60
Murder per 1000 resid ents 0.09 0.04 0.02
“Effective” abortion rate per 1000 live
births by crime:
Violent crime 77.11 83.18 66.13
Property crime 132.26 116.46 86.89
Murder 51.00 66.57 55.39
Prisoners pe r 1000 residents 2.83 1.26 0.86
Police per 1000 residents 2.85 0.64 0.27
State personal income per capita
($1997) 23207 3408 1361
AFDC generosity per recipient family
(t–15) 7242 2905 1364
State unemployment rate (percent
unemployed) 6.15 1.55 1.21
Beer consumption per capita (gallons) 23.03 3.32 1.24
Poverty rate (percent below poverty
level) 13.80 3.51 1.6 4
Violent crime arrests per 1000 , under
age 25 3.18 1.46 0.49
Property crime arrests per 1000,
under age 25 12.36 3.76 1.44
Murder arrests per 1000, under age
25 0.11 0.06 0.03
Violent crime arrests per 1000 , age
25 and ove r 2.04 1.06 0.34
Property crime arrests per 1000, age
25 and ove r 4.82 1.58 0.65
Murder arrests per 1000, age 25 and
over 0.06 0.03 0.01
All v alues reported are means of annual, state-level observations for the period 1985–1997 with the
following exceptions. Arrest data cover the years 1985–1996, and AFDC generosity data are for the years
1985–1998. The police and prisons data are once-lagged, and thus correspond to the years 1984–1996. The
values reported in the table are population weighted averages. The effective abortion rate is a weighted
average of the abortion rates for each cohort born in a state, with weights determined by the percentage of
arrests by age for a given crime categ ory in the United States in 1985 as shown in equation (1). All summary
statistics are based on 663 observations, except where otherwise noted. Because of missing data, arrest
statistics are based on 574 observations, compared with a theoretical maximum of 612. AFDC statistics are
based on 714 observations. See Data Appendix for further details.
403LEGALIZED ABORTION AND CRIME
by the coefcients on abortion is substantial. An increase in the
effective abortion rate of 100 per 1000 live births (the mean
effective abortion rate in 1997 for violen t crime is 180 with a
standard deviation of 96 acro ss states) is associated with a reduc-
tion of 12 percent in murder, 13 percent in violent crime, and 9
percent in property crime. In Table II, c omparing the states in the
top third with respect to abortions to the states in the bottom
third, our parameter estim ates imply that crime fel l an additional
16 –25 percent in the former states by 1997 due to greater usage
TABLE IV
P
ANEL
-
DATA
E
STIMATES OF THE
R
ELATIONSHIP BETWEEN
A
BORTION
R
ATES AND
C
RIME
Variable
ln(Violent
crime per
capita)
ln(Property
crime per
capita)
ln(Murder per
capita)
(1) (2) (3) (4) (5) (6)
“Effective” abortion rate
(3 100)
2 .137 2 .129 2 .0 95 2 .091 2 .108 2 .121
(.023) (.024) (.018) (.018) (.036) (.047)
ln(prisoners per capita)
(t 2 1)
— 2 .027 — 2 .159 — 2 .231
(.044) (.036) (.080)
ln(police per capita)
(t 2 1)
— 2 .028 — 2 .049 — 2 .300
(.045) (.045) (.109)
State unemployment rate
(percent unemployed)
— .069 — 1.310 — .968
(.505) (.389) (.794)
ln(state income per
capita)
— .049 — .084 — 2 .098
(.213) (.162) (.465)
Poverty rate (percent
below poverty line)
— 2 .000 — 2 .001 — 2 .005
(.002) (.001) (.004)
AFDC generosity (t 2
15) (3 1000)
— .008 — .002 — 2 .000
(.005) (.004) (.000)
Shall-issue concealed
weapons law
— 2 .004 — .039 — 2 .015
(.012) (.011) (.032)
Beer consumption per
capita (gallons)
— .004 — .004 — .00 6
(.003) (.003) (.008)
R
2
.938 .942 .990 .992 .914 .918
The dependent variable is the log in the per capita crime rate named at the top of each pair of columns.
The rst column in each pair presents results from specications in which the only additional covariates ar e
state- and year-xed effects. The second column presents results using the full specication. The data set is
comprised of annual state-level observations (including the District of Columbia) for the period 1985–1997.
The number of observations is equal to 663 in all columns. State- and year-xed effects are included in all
specications. The prison and police variables are once-lagged to minimize endogeneity. Estimation is
performed using a two-step procedure. In the rst step, weighted least squares estimates are obtained, with
weights determined by state population. In the second step, a panel data generalization of the Prais-Winsten
correction for serial correlation developed by Bhargava et al. [1982] is implemented. Standard errors are in
parentheses. Data source s for all variables are described in the Data Appendix.
404 QUARTERLY JOURNAL OF ECONOMICS
of abortion. One additional abortion is associated with a reduction
of 0.23 property crimes, 0.04 vio lent crim es, and 0.004 murders
annually when a cohort is at its peak crime age. Comparing these
estimates to average crimin al propensities among 18 –24 year
olds, those on the mar gin for being aborted are roughly fou r times
more criminal. These estimates are roughly consistent with, but
somewhat larger than, the b ack-of-the-envelope predictions in
Section III.
The other coefcients in the model appe ar plausibly esti-
mated. The elasticities of incarceration and police with respect to
crime all carry the expected sign, with priso n associated with
signicant reductions in property crime and murder, and police
associated with signi cant reductions in murder.
28
A higher state
unemployment rate is asso ciated with signicant increases in
property crime, but not violent crime, consistent with previous
research [Freeman 1995]. The three other measures of state
economic conditions—per capita income, the poverty rate, and
AFDC generosity (lagged fteen years to roughly correspond with
the early years of life of the current teenagers) do not systemat-
ically affect crime. Shall-issue concealed carry laws appear to
signicantly increase the amount of property crime, but have no
effect on violent crime or murder. Finally, bee r consumption is
weakly l inked with higher crime rates, but n ever signic antly so.
Table V investigates the sensitivity of the abortion coef-
cients to a range of alternative specications. We take the spec-
ications with the full set of controls in Table IV as a baseline.
The abortion coefcients from those regressions are r eported in
the top row of Table V. Each row of the table represents a
different specication. The sensitivity of the results to lar ge
states (since the regressions are population weighted) and states
with very high o r low abortion rates is examined rst. Removing
New York reduces the estimates for violent crime and murder,
while eliminating California increases the abortion coefcient for
those two crime categories. Dropping Washington, DC, which is
an extreme outlier (with an abortio n rate over four times the
national average) increases the estimated impact of abortion.
28. The estimate d effects of incarceration are consistent with previous cor-
relational pan el-data studies (e.g., Marvell and Moody [1994]). The prison coef-
cients obtained here are approximately the same magnitude as Levitt [1996] n ds
when correcting for the endogene ity of the prison population using prison o ver-
crowding litigation as an instrument. Levitt [1997] nds a negative impact of
police on crime using electoral cycles in large cities as an instrument for the size
of the police force.
405LEGALIZED ABORTION AND CRIME
Dropping all three of those high abortion states leads to higher
estimates across the board, suggesting that the crime-reducing
impact of abortion may have decreas ing retur ns.
Omitted variables may also be a concern in the regressions
given the relatively lim ited set of covariates available. One crude
way of addressing this question is to include regio n-year interac-
tion terms in an attempt to absorb geographically correlated
TABLE V
S
ENSITIVITY OF
A
BORTION
C
OEFFICIENTS TO
A
LTERNATIVE
S
PECIFICATIONS
Specication
Coefcient on the “effective” abortion rate
variable when the dependent variable is
ln (Violent
crime per
capita)
ln (Property
crime per
capita)
ln (Murder
per capita)
Baseline 2 .129 (.024) 2 .091 (.018) 2 .121 (.047)
Exclude New York 2 .097 (.030) 2 .097 (.021) 2 .063 (.045)
Exclude Califo rnia 2 .145 (.025) 2 .080 (.018) 2 .151 (.054)
Exclude District of Columbia 2 .149 (.025) 2 .112 (.019) 2 .159 (.053)
Exclude New York, California,
and District o f Colu mbia 2 .175 (.035) 2 .125 (.017) 2 .273 (.052)
Adjust “effective” abortio n rate
for cross-state mobility 2 .148 (.027) 2 .099 (.020) 2 .140 (.055)
Include control for ow of
immigrants 2 .115 (.024) 2 .063 (.018) 2 .103 (.047)
Include state-specic trends 2 .078 (.080) .143 (.033) 2 .379 (.105)
Include region-year interactions 2 .142 (.033) 2 .084 (.023) 2 .123 (.053)
Unweighted 2 .046 (.029) 2 .022 (.023) .040 (.054)
Unweighted, exclude District of
Columbia 2 .149 (.029) 2 .107 (.015) 2 .140 (.055)
Unweighted, exclude District of
Columbia, California, and
New York 2 .157 (.037) 2 .110 (.017) 2 .166 (.075)
Include control for overall
fertility rate (t 2 20) 2 .127 (.025) 2 .09 3 (.019) 2 .123 (.047)
Long differe nce estimates using
only data from 1985 and 1997 2 .109 (.054) 2 .077 (.034) 2 .089 (.077)
Results in this table are variations on the specications reported in columns (2), (4), and (6) of Table IV.
The top row of the current table is the baselin e specication that is presented in Table IV. Except where
noted, all specications are estimated using an annual, state-level panel of data for the years 1985–1997.
Standard errors (in parentheses) are corrected fo r serial correlation using the Bhargava et al. [1982] two-step
procedure for panel d ata. The specication that corre cts for cross-state mobility does so by using an effective
abortion rate that is a weighted average of the abortion rates in the state of birth for fteen year-olds residing
in a state in the PUMS 5 percent sample of the 1990 census. Controls for the ow of immigrants are derived
from changes in the foreign-born population, based on the decennial censuses and 1997 estimates, linearly
interpolated. Region-year interactions are for the nine census regions.
406 QUARTERLY JOURNAL OF ECONOMICS
shocks. The abortion coefc ients are not substantially affected by
this approach.
Since we are measuring the effect of abortions in a state on
crime in that state up to a quarter century later, the issue of
cross-state mobility should be considered. Theoretically, the pres-
ence of such cross-state movements will tend to systematically
bias the abortion coefcient toward zero since the true effective
abortion rate is mea sured with error by our proxy that ignores
mobility. In order to adjust for migration, w e determined the
state of birth and state of residence for all fteen year-olds in the
1990 PUMS 5 percent sample. Using this inform ation, we recal-
culated effe ctive abortion rates as weighted average abortion
rates by the actual state of bir th of fteen year-olds residing in a
state. For all three crime categories the estimated impact of
abortion increases with the migration correction, although the
changes are not large.
We perform a range of other sensitivity checks. Controlling
for the ow of immigrants to a state somewhat reduces the
estimated effect of abortion on crime (particularly for property
crime), but it do es not change their signicance. When we include
state-specic time trends, the estimates change somew hat errat-
ically, and the standar d errors double for murder and property
crime and triple for violent crime. Unweighted panel data regres-
sions (as opposed to population weighted) yie ld sharply smaller
coefc ients, but this is exclusively due to Washington, DC as an
outlier (o wing i n all likelihood to mismeasurement in the DC
abortion rate). Excluding District of Columbia alone, or District of
Columbia in combination with California and New York , leads to
coefc ients from the unwei ghted regressions that are gre ater
than the baseline estimates.
Including c ontrols for lagged changes in overall fertility rates
for the same era as our abortion measures has almost no impact
on our estimated coefcients. Regressions using only the 1985
and 1997 endpoints of our sample (“long-differences”) yield coef-
cients similar to, although somewhat smaller than, the baseline
coefc ients for the overall panel.
V. T
HE
I
MPACT OF
A
BORTION ON
A
RRESTS BY
A
GE OF
O
FFENDER
The preceding section highlighted a strong empirical corre-
lation between abortion rates after
Roe v. Wade
and crime
changes in recent ye ars. In this section we explore the extent to
407LEGALIZED ABORTION AND CRIME
which arrest patterns substanti ate a possible causal interpreta-
tion of these results. In particular, if legalized abortion is the
reason for the decline in crime, then one would expect that de-
creases in crime should be concentrated among those cohorts born
after abortion is legalized.
29
Testing that hypothesis is complicated by the fact that the
age of criminals is not directly observable. The age of arrestees,
however, is reported.
30
Thus, we can analyze whether arrests by
cohort are a function of the abortion rate.
The basic specications used to explain state arrest rates by
age category are identical to the crime regressions in the preced-
ing section, except that the dependent variable is the (natural log
of the) arrest rate per capita for those under age 25 rather than
the overall crime rate for all ages, and 1997 is ex cluded from the
sample because the necessary arrest data are not yet available.
31
Results from the estimation are reported in columns 1–3 of Table
VI. Two specications per crime category are presented: the top
row of results just includes the effective abortion variable and
year- and state-xed effects, while the bottom row adds to these
the remainin g covariates that were used in Table IV above. Be-
cause the dependent variable is denominated by the population
under age 25, the abortion coefcients only reect changes in
arrest rates per person. If the impact of abortion was solely
through changes in coho rt size, then the per capita specication s
we run would yield zero coefcients on the abortion variable. In
all six cases, lagged abortion rates are associate d with decreases
in arrests per capita by those under the age of 25, with estimates
29. It is possible that crime by older cohorts may be affected indirectly by
abortion. For instance, if there are fewer criminals in younger cohorts, this may
increase additional criminal opportunities for older individuals (particularly in
activities such as drug distribution where there may be easy substitutability). On
the other hand, to the extent that lower crime by the young increases the criminal
justice resources available per older criminal [Sah 1991], crime among older
cohorts may also fall. Moreover, as noted above, if abortion results in smaller
family sizes and a concomitant increase in parental resources per child, the effect
of legalization could be observed in crime reductions for older sibl ings. All of these
effects are likely to be of second-order magnitude, however.
30. Arrest data may not a ccurately reect criminal activity for a number of
reasons. Greenwood [1995] argues that juvenile crime is more likely to be com-
mitted in groups so that the arrest frequency of juveniles overstates the true
fraction of crime the y commit. Also, if there are differences across criminals in
avoiding detection, arres ts will be skewed toward the less procient criminals.
31. We use an age cutoff of 25 because it is approxima tely the age of the
oldest coh orts affected by legalized abortion. Arrest data are available by single
year of age up to age 24, but only in ve-year groupings thereafter. The results
presented are not sensiti ve to small perturbations of the age groupings.
408 QUARTERLY JOURNAL OF ECONOMICS
TABLE VI
T
HE
I
MPACT OF
A
BORTION
R
ATES ON
A
RRESTS BY
A
GE
(A
LL
V
ALUES IN THE
T
ABLE
A
RE
C
OEFFICIENTS ON THE
E
FFECTIVE
A
BORTION
R
ATE
(3 100), O
THER
C
OEFFICIENTS
A
RE
N
OT
R
EPORTED
)
Specic ation
ln (arre st per person, under
age 25)
ln (arrests per person, ag e
25
+
)
ln (arrests per person, under
age 25) minus ln (arrests per
person, age 25
+
)
Violent
crime
Property
crime Murder
Violent
crime
Property
crime Murder
Violent
crime
Property
crime Murder
Effec tive abortion rate (3
100) only, no covariates
included 2 .095 2 .085 2 .214 .022 2 .019 2 .034 2 .116 2 .066 2 .180
(.029) (.023) (.051) (.054) (.037) (.037) (.042) (.023) (.034)
Effec tive abortion rate (3
100), including full set of
covariates 2 .044 2 .054 2 .180 .033 .008 2 .036 2 .062 2 .063 2 .137
(.030) (.023) (.062) (.046) (.031) (.050) (.034) (.019) (.046)
Regressions are identical to those in Table IV, except that the dependent variables are arrest rates broken down by age cat egory instead of overall crime rates. The top row of
the table presents results from specications in which the only additional covariates are state- and year-xed effects. The bottom row of the table pres ents results using the full
specication. Covariates included in the bottom row are once-lagged police and prisoners per capita in logs, state unemployment rate, logged state income per capita, the poverty rate,
lagged AFDC generosity, shall-issue conce aled weapons law, a nd beer consumption per c apita. The regressions use annual state-level data for the period 1985–1996 (1997 arrest data
by age are not yet available). Because of missing data, the number of observations varies across columns between 555 and 557, compared with a theoretical maximum of 612. State-
and ye ar-xed effects are included in all specications. The prison and police variables are once-lagged to minimize endogeneity. Estimation is performed using a two-step procedure.
In the rst step, weighted least squares estimates are obtained, with weights determined by state population. In the second step, a panel data generalization of the Prais-Winsten
correction for serial correlation developed by Bhargava et al. [1982] is implemented. Standard errors are in parentheses.
409LEGALIZED ABORTION AND CRIME
ranging between
2
.044 and
2
.214. The abortion coefcient is
statistically signicant in ve out of six spec i cations.
If the arrest data are measur ed without error and there are
no spillovers between the crim e of the young and the old, then we
would not expect legalized abortion to affect the crime of those
born prior to the law change. Colum ns 4– 6, which relate arrest
rates of older cohorts to abortion rates, thus provide a natural
specication test for o ur hypothesis. In none of the crime catego-
ries does the abortion rate variable have a statistically signicant
impact on arrests of older coho rts. In three instances the coef-
cient is positive; in the other three cases the coefcient is nega-
tive. All of the estimates are much smaller in magnitude than was
the case for arrests of those under the age of 25. The last three
columns of the table show “difference in differences” estim ates of
the impact of abortion on cohorts born after legalization relative
to those born befo re. In all cases, the coefcients are sim ilar to
those in the rst three columns of the table. This result strength-
ens the causal interpre tation of the abortion coefcients on the
arrest patter ns of the young.
The implied magnitude of the abortion effects on arrests is
smaller than the parallel estim ates presented in the pr eceding
section analyzing crime rates, but is of the same order of magni-
tude. On average, about half of those arrest ed are unde r the age
of 25.
32
Thus, to generate the crime reduction in Table IV requires
coefc ients on young arrests that are twice as large as the coef-
cients o n overall crime. With the exception of m urder, the arrest
coefc ients are ac tually smaller than the crime coefcients. Part
of this discrepancy may be attributable to the fact that the arrest
regressions reect only reductions in per capita crime by the
young, no t smaller youthful cohorts, but this can explain only a
portion of the gap. It remains an open question as to whether this
discrepancy represents a partially spurious relationship in the
crime regressions, measu rement error in the arrest data, or a
relationship between crime and arrests that is not proportional.
It is important to stress, however, that while the magnitude of the
effects differ s between the crime and arrest regressions, the basic
story with respect to abortion is present in both cases.
33
32. Over the sample period, those under the age of 25 accounted for an
average of 49 percent of violent arrests, 62 percent of property arrests, and 48
percent of murder arrests.
33. We replicated the sensitivity tests that were presented in Table V for the
baseline Ta ble IV regressions using Table VI as the baseline estimates. These
410 QUARTERLY JOURNAL OF ECONOMICS
As a further test of our hypothesis, we analyze arrest rates by
state by single year of age. These data are available for the ages
15 and 24 covering the period 1985 through 1996. If abortion
legalization reduces crime, then we should see the reduction
begin with, say, fteen year-olds abou t sixteen years after legal-
ization, then extend t o sixteen year-olds a year later, and so on.
Because we observe many cohorts in a given state and year, we
are able to include controls for state-year variation. Thus, unlike
the preceding table, where state-year variation was our source of
identication, in the analysis that follows our estimates are based
on differences in abortion rates and crime rates across cohorts
within a given state and year. T he regression we run takes the
following form:
(3) ln
(
ARRESTS
stb
)
5 b
1
ABORT
sb
1 g
s
1 l
tb
1 u
st
1 e
stb
,
where
s
,
t
, and
b
index state, year, and birt h cohort, respectively.
The variable
ARRESTS
is the raw number of arrests for a given
crime. Un like previous tables, we do not divide arrests by popu-
lation to create per capita rates because of the absence of reliable
measures of state population by single year of age. As our mea-
sure of the abortion rate for a particular cohort, we use the
abortion rate in the current state o f residence in the calendar year
most likely to have precede d the arrestees birth.
34
Cross-state
migration will not be captured by this measure, but the results in
earlier sections suggest that the impact of migration on the esti-
mates is small (and that any migration correction would, if an y-
thing, strengthen our results). Because the unit of observation in
the analysis is a state-birth cohort and cohorts are obser ved
repeatedly over time, we will include controls for age, nation al
year-cohort interactions, state-year interactions, and (in some
cases) state-age interactions. We cannot, however, include state-
regressions again reveal ed the robustness of the coefcient estimates, exhibiting
patterns similar to the sensitivity analysis for the full sample. These results are
available from th e authors on request.
34. For example, we use the abortion rate in 1980 to reect the abortion
exposure of fteen year-olds arrested in 1996. Because the arrest data cover a
calendar yea r, there is a possible 730-day window into which an arrestee’s date of
birth may fall (i.e., an arrest is made on January 1 of someone who is 16 years and
364 days old versus an arrest is made on December 31 of someo ne who is 16 years
and 1 day old). With a six-to-seven-month lag from likely time of abortion to time
of birth, this 730-day window is centered on the calendar year that we use to
capture abortion exposure. More complicated attempts to measure abortion expo-
sure yield estimates similar to the ones we present.
411LEGALIZED ABORTION AND CRIME
birth cohort interactions witho ut absorbing all of the variation in
the abortion exposure of a state-birth cohort.
Table VII presents th e results of this analysis for violent
crime and property crime. There are too few murder arrests per
single age category per state to enable us to provide similar
estimates for murder. We present estimates restricting the im-
pact of abortion to be constant over the entire age range (odd
columns) and allowing the impact of abortion to vary by age (even
columns). Some of the regressio ns include state-age interactions,
others just have state-xed effe cts. All of the specications in-
clude year-age interac tions to control for national-level uctua-
tions in the age-crime prole.
35
In all cases, standard errors have
been corrected to reect correlation over time in a given birth
cohort’s observations.
The top row of Table VII presents estimates restricting the
abortion coefcient to be constant across the ages 15–24. In all
instances, the coefcient is stron gly signicantly negative, imply-
ing that higher abortion rates around the time a cohort is born are
associated with lower arrest rates in their teens and twenties.
When the abortion coefcient is allowed to vary by age, 38 of the
40 parameter estimates are negative; m ore than two-thirds of
these estimates are statistically signicant at the .05 level. The
greatest impact of abortion appears to occur in the age range
18 –22. The effects are generally weakest for the youngest ages in
the sample.
The coefcients in this table are not directly comparable to
those in the preceding tables. Because we are analyzing arrests
by single year of age in this table, we are able to use actual
abortion rates as opposed to the effective abortion rates that
average over many cohorts. Comparing states in the top third and
bottom third with respect to abortion frequency, the gap between
those sets of states in actual abortion rates was about 350 per
1000 births. Given th e estimates in the top row of Table VII, this
implies that arrest rates of 15–24 year-olds in the high abortion
states are estimated to have fallen between 5 and 14 percent
relative to the low abortion states.
35. For instance, the arrival of crack appears to have temporarily raised the
violent crime prop ensities, particularly among yo uths.
412 QUARTERLY JOURNAL OF ECONOMICS
TABLE VII
T
HE
R
ELATIONSHIP BETWEEN
A
BORTION
R
ATES AND
A
RREST
R
ATES
,
BY
S
INGLE
Y
EAR OF
A
GE
ln (Violent arrests) ln (Property arrests)
Abortion rate (3 100) 2 .015 — 2 .028 — 2 .040 — 2 .025 —
(.003) (.004) (.004) (.003)
Abortion rate (3 100) interacted with
Age
=
15 — .018 — 2 .008 — 2 .037 — 2 .005
(.008) (.010) (.007) (.008)
Age
=
16 — .008 — 2 .007 — 2 .043 — 2 .011
(.007) (.008) (.006) (.006)
Age
=
17 — 2 .010 — 2 .021 — 2 .042 — 2 .013
(.006) (.007) (.006) (.005)
Age
=
18 — 2 .035 — 2 .039 — 2 .053 — 2 .023
(.004) (.007) (.005) (.005)
Age
=
19 — 2 .040 — 2 .043 — 2 .050 — 2 .036
(.005) (.007) (.005) (.006)
Age
=
20 — 2 .043 — 2 .043 — 2 .038 — 2 .035
(.006) (.007) (.006) (.006)
Age
=
21 — 2 .039 — 2 .039 — 2 .028 — 2 .037
(.009) (.008) (.006) (.006)
Age
=
22 — 2 .028 — 2 .024 — 2 .020 — 2 .032
(.013) (.009) (.008) (.009)
Age
=
23 — 2 .031 — 2 .026 — 2 .015 — 2 .030
(.023) (.013) (.011) (.013)
Age
=
24 — 2 .027 — 2 .016 — 2 .024 — 2 .047
(.040) (.020) (.019) (.018)
R
2
.972 .972 .985 .985 .967 .968 .98 4 .984
Number of observations 5,737 5,737 5,737 5,737 5,740 5,740 5,740 5,740
State-xed effects or State-age
interactions? State-xed State-xed
State p Age
interactions
Stat e p Age
interactions State-xed State-xed
State p Age
interactions
State p Age
interactions
Results in the table are coefcients from estimation of equation (3). The unit of observation in the regression is annual arrests by state by single year of age. The sample covers the period
1985–1996 for ages 15–24. The abortion rate for a cohort of age a in state s in year y is the number of abortions per 1000 live births in state s in year y 2 a 2 1. Note that this is the actual
abortion rate, rather than the “effective” abortion rate used in preceding tables. Therefore, the coefcients in this table are not directly comparable to those of earlier tables. If data were
available for all states, years, and ages, the total number of observations would be 6120. Due to missing arrest data and occasional zero values for arrests, the actual number of observations
is somewhat smaller. A complete set of year-birth cohort interactions are included in all specications to capture national changes in the shape of the age-crime prole over time. State-year
interactions are also included. Some specications include state-xed effects; in other specications, a complete set of state-age interactions is included. Estimation is weighted least squares,
with weights determined by total state population. Standard errors have been corrected to account for correlation over time within a given birth cohort in a particular state. Such a correction
is necessary because the abortion rate for any given cohort is xed over time, but multiple observations corresponding to different years of age are included in the regression. Results for
murder are not included in the table because murder is infrequent, leading to many zeros when a nalyzed at the level of state and single year of age.
413LEGALIZED ABORTION AND CRIME
VI. C
ONCLUSION
We know that te enagers, unmarried wo men, an d poor women
are most likely to deem a pregnancy to be either mistimed or
unwanted, and that a large proportion of these unintended preg-
nancies will be terminated through abortion.
36
According to a
recent National Academy report, there appears to be “a causal
and adverse effect of ear ly childbearing on the health and social
and economic well-being of children; this e ffect is over and above
the important effects of background disadvantages” [Institute of
Medicine 1995, p. 58]. Moreov er, unintended pregnancies are
associated with poorer prenatal care, greater smoking and drink-
ing during pregnancy, and lower birthwe ights. Conseque ntly, the
life chances of children who are born only because their mothers
could not have an abortio n are considerably dampened relative to
babies who were wa nted at the time of conception. The drop in the
proportion of unwanted births during the 1970s and early 1980s
appears to be the result of the increasing availability and resort
to abortion.
The evidence we present is consiste nt with legalized abortion
reducing crime rates with a twenty-year lag. Our results suggest
that an in crease of 100 abortions per 1000 live births reduc es a
cohort’s crime by roughly 10 percent. Extrapolating our results
out of sample to a counterfactual in which abortion remained
illegal and the number of illegal abortions performed remained
steady at th e 1960s level, we estimate that (with average national
effective abo rtion rate s in 1997 for all three crimes ranging from
between 142 and 252) crime was almost 15–25 percent lower in
1997 than it would have been absent legalized abortion.
These estimates suggest that legalized abortion is a primary
explanation for the large drops in murder, prope rty crime, and
violent crime that our nation has exper ienced over the last de-
cade. Indeed, legalized abortion may account for as much as
one-half of the overall crime reduction. Assuming that this claim
is correct, existing estimates of the costs of c rime (e.g., Miller,
Cohen, and Rossman [1993] suggest that the social benet to
reduced crime as a result of abortion may be o n the order of $30
billion dollars annua lly. Increased imprisonm ent between 1991
36. Roughly 75 percent of never-married women who unintentionally become
pregnant will opt for abortion. Overall, almost exactly half of all unintended
pregnancies—whether mistimed or unwanted—will be terminated by abortion
[Institute of Medicine 1995, pp . 41–47].
414 QUARTERLY JOURNAL OF ECONOMICS
and 1997 (the prison population rose about 50 percent over this
period) lowered crime 10 percent based on an elast icity of
2
.20.
Thus, together a bortion and prison growth explain much, if not
all, of the decrease in crime.
37
Roughly half of the crimes committed in the United States
are done by individuals born prior to the legalization of abortion.
As these older cohorts age out of criminality and are replaced by
younger offenders born after abortion became legal, we would
predict that crime rates will continue to fall. When a steady state
is reached ro ughly twenty years from now, the impact of abortion
will be roughly twice as great as the impact felt so far. Our results
suggest that all else equal, legalized abortion will account for
persistent declines of 1 percent a year in c rime over the next two
decades. To the extent that the Hyde Amendment effectively
restricted access to abortion, however, this prediction might be
overly optimistic.
While falling crime rates are no doubt a positive develop-
ment, our drawing a link between falling crime and legalized
abortion should not be m isinterpreted as either an endorsem ent
of abortion or a call for intervention by the state in the fertility
decisions of women. Furthermore, equivalent reduction s in crime
could in principle be obtained throug h alternative s for abortion,
such as more effective birth control, or pr oviding be tter environ-
ments for those children at greatest risk for future crime.
D
ATA
A
PPENDIX
Crime and Police
All crime and police data used in the analysis are from
Federal Bureau of Investigation
Crime in the United States
[an-
nual], exc ept the victimization data in Figure II, which are s um-
marized annually in Bureau of Ju stice Statistics
Sourcebook of
Criminal Justice Statistics
[annual].
Abortion
All abortion data are from Bureau o f the Census
United
States Statistical Abstract
[annual]. The primary source for the
37. This is not to say that other factors did not also contribute to the decli ne
in crime. To the extent that there were other forces pushi ng crime higher, such as
crack, then the set of factors leading to reduce d crime wi ll explain more than 100
percent of the observed decrease in crime.
415LEGALIZED ABORTION AND CRIME
abortion data is an annual survey conducted by the Alan Gutt-
macher Institute.
Prisoners
Data on number of prisoners are from
Correctional Popula-
tions in the United States,
published annually by the Bureau of
Justice Statistics.
Population by Age
These data are from
Estimates for the U nited States, Regions,
Divisions, and States by 5 Year Age Groups and Sex: Annu al Time
Series Estimates,
U. S. Census Bureau [annual].
Poverty
Persons Below Poverty Level, by State, taken from Bureau of
the Census
United States Statistical Abstract
[annual].
Unemployment
Figures used represent the percent unemployed among civil-
ian noninstitutional population sixteen years and older, with
total unemployment estimates based on the Current Population
Survey, taken from Bureau of the Census,
United States Statis-
tical Abstract
[annual].
Fertility
The number of live births per 1000 population, taken from
Bureau of the Census,
United States Statist ical Abstract
[annual].
Income
Per capita state personal income, converted to 1997 dollars
using the Consumer P rice Index, from Bureau o f the Census,
United States Statistical Abstract
[annual].
AFDC Generosity
Public Assistance Payments to Families with Dependent
Children, from Bureau of the Census,
United States Statistical
Abstract
[annual]. The data reported in the Statistical Abstract
are the average mo nthly paym ent per family receiving aid. That
number is multiplied by twelve to obtain a yearly average, and
then converted into 1997 dollar s using the Consumer Price Index.
416 QUARTERLY JOURNAL OF ECONOMICS
Nondiscretionary Concealed Handgun Law
Indicates the year in which the state enacted a law requiring
local law enforcement authorities to grant concealed weapons
permits to anyone meeting certain preestablished criteria. Data
come from Lott and Mustard [1997].
Beer Consumption
Consumption of Malt Beverages from the Beer Institute’s
Brewer’s Almanac
[1995, 1998]. In gallons consum ed per capita.
Cross-State Migration
The corre ctions for cross-state m igration are based on a com-
parison of the state of birth and c urrent state of residence of
fteen year-olds in the 1990 Census Pu blic Use Microdata 5
percent sample.
Foreign-Born Population
Prior to 1994, the de cennial census was the only source of
data on the nu mber of foreign-born individuals living in the
United States. Data from the three Census years and 1997 were
used to in terpolate intervening years. All data are from Bureau of
the Census
United States Statistical Abstract
[annual].
S
TANFORD
L
AW
S
CHOOL
U
NIVERSITY OF
C
HICAGO AND
A
MERICAN
B
AR
F
OUNDATION
R
EFERENCES
Alan Guttmacher Institute, Abortion Fact Book, 1992 Edition, Stanley Henshaw
and Jennifer Van Vort, eds. (New York, NY: Alan Guttmacher Institute,
1992).
——, “An Analysis of Pre-1973 State Laws o n Abortion,” mimeo, March 1989.
Angrist, Joshua, and William Evans, “Schooling and Labor Market Consequences
of the 1970 State Abortion Reforms,” National Bureau of Economic Res earch,
Working Paper No. 5406, January 1996.
Barber, Jennifer, William Axinn, and Arland Thornton, “Unwanted Childbearing,
Health, and Mother-Child Re lationships,” Journal of Health and Social Be-
havior, XL (September 1999), 231–257.
Beck, Allen, et al., “Survey of State Prison Inmates, 1991,” Bureau of Justice
Statistics Bulletin, NCJ 136949, March 1993.
Becker, Gary, A Treatise on the Family (Chicago, IL: Unive rsity of Chicago Press,
1981).
Beer Institute, Brewers’ Almanac (New York, NY: United States Brewers Foun-
dation, 1995 and 1998 editions).
Bhargava, A., L. Franzini, et al., “Serial Correlatio n and the Fixed Effects Model,”
Review of Economic Stu dies, XL IX (1982), 533–549.
Blank, Rebecca, C. C. George, and R. A. London, “State Abortion Rates: The
Impact of Policies, Providers, Politics, Demographics, and Economic Environ-
ment,” Journal of Health Economics, XV (1996), 513–55 3.
417LEGALIZED ABORTION AND CRIME
Blumstein, Alfred, Jacqueline Cohen, Jeffrey Roth, and Christy Visher, eds .,
Criminal Careers and “Career Criminals” (Washington, DC: National Acad-
emy of Sciences, 1986).
Blumstein, Alfred, and Richard Rosenfeld, “Explaining Recent Trends in U. S.
Homicide Rates,” Journal of Criminal Law and Criminology, LXXXVIII
(1998), 1175–1216.
Blumstein, Alfred, and Joel Wallman, eds., The Crime Drop in America (New
York: Cambridge University Press, 2000).
Bouza, Anthony V., The Police Mystique: An Insider’s Look at Cops, Crime, and the
Criminal Justice System (New York, NY: Plenum Press, 1990).
Bureau of the Census, “Popul ation Estimates for the U. S., Regions, Divisions, and
States by 5-year Age Groups and Sex,” Population Estimates Program
(annual).
——, Statistical Abstract of the United States (Washington, DC: U. S. Government
Printing Ofce, annual).
Bureau of Justice Statistics, Correctional Populations in the U. S. (Washington,
DC: U. S. Gove rnment Printing Ofce, annual).
——, The Sourceb ook of Criminal Justice Statistics, Timothy Flanagan and Ed-
mund McGarrell, eds. (Washington, DC: U. S. Government Printing Ofce,
1985 and annual).
Buttereld, Fox, “Number of Victims of Crime Fell Again in ’96, Study Says
Lowest Level Since Reports Began in 1973,” The New Yo rk Times (November
16, 1997a), 10.
——, “Drop in Homicide Rate Linke d to Crack’s Decline,” The New York Times
(October 27, 1997b), A10.
——, “Crime Fell 7 Percent in ’98, Continuing a 7-Year Trend,” The New York
Times (May 17, 1999), A12.
Centers for Disease Control , “Abortion Surveillance,” Morbidity and Mortality
Weekly Report, XLIII (1994), 930–939.
Comanor, William S., and Llad Phillips, “The Impact of Income and Family
Structure on Delinquency,” unpublished manuscript, University of Califor-
nia, Santa Barbara, 1999.
Commission on Population Growth and the American Future, Population and the
American Future (Washington, DC: U. S. Government Printing Ofce, 1972 ).
Cook, Phi lip, and John Laub, “The Social Ecology of Youth Violence,” Youth
Violence, Michael Tonry and Mark H. Moore, eds. (Chicago: University of
Chicago Press, 1998).
Corman, Hope, and Michael Grossman, “Determinants of Neonatal Mortality
Rates in the U. S.: A Reduced Form Model,” Journal of Health Economics, IV
(1985), 213–236.
Dagg, P. K., “The Psychological Sequelae of Therapeu tic Abortion—Denied and
Completed,” American Journal of Psychiatry, CXLVIII (19 91), 578–585.
David, Henry, Zdenek Dytrych, et al., Born Unwanted: Developmental Effects of
Denied Abortion (New York, NY: Springer, 1988).
Donohue, John J., “Understanding the Time Path of Crime ,” Journal of Criminal
Law and Criminol ogy, LXXX VIII (1998), 1423–1452.
Dye, Jane L., and Harriet B. Presser, “The State Bonus to Reward a Decrease in
‘Illegitimacy:’ Flawed Methods and Questi onable Effects,” Family Planning
Perspectives, XXXI (1999), 142–147.
Federal Bureau of Investigation, Uniform Crime Reports for the United States
(Washington, DC: U. S. Government Printing Ofce , annual).
Fletcher, Michael A., “The Crime Conundrum,” The Washington Po st (January 16,
2000), Fl.
Freeman, Richard, “The Labor Ma rket,” Crime, James Q. Wilson and Joan Peter-
silia, eds. (San Francisco, CA: ICS Press, 1995).
Goldin, Claudia, and Lawrence Katz, “The Power of the Pill: Oral Contraceptives,
a Women’s Career and Marriage Decisions,” unpublished manuscript, Har-
vard University, 2000.
Greenwood, Peter, “Juvenile Crime and Juvenile Justice,” Crime, James Q. Wil-
son and Joan Petersilia, eds. (San Francisco, CA: ICS Press, 1995).
Grossman, Micha el, and Steven Jacobowitz, “Variations in Infant Mortality Rates
418 QUARTERLY JOURNAL OF ECONOMICS
among Counties of the United States: The Roles of Public Poli cies and Pro-
grams,” Demography, XVIII (1981), 695–713.
Grossman, Michae l, and Theodore Joyce, “Unobservables, Pregnancy Resolutions,
and Birth Weight Production Functions in New York City,” Journal of Polit-
ical Economy, XCVIII (1 990), 983–1007.
Gruber, Jonathan, Phillip Levine, and Douglas Staiger, “Abortion Legalization
and Child Living Circumstances: Who Is the ‘Marginal Child?’ ” Quarterly
Journal of Economics, CXIV (1999), 263–291.
Institute of Medicine, The Best Intentions: Unintended Pregnancy and the Well-
Being of Children and Families, Sarah S. Brown and Leon Eisenberg, eds.
(Washington, DC: National Academy Press, 1995).
Joyce, Theodore, “The Impact of Induced Abortion on Black and White Birth
Outcomes in the United States,” Demography, XXIV (1987), 229–244.
Joyce, Theodore, and Robert Kaestner, “The Effect of Expansions in Medicaid
Income Eligibility on Abortion,” Demography, X XXIII (1996), 181–192 .
Kane, Thomas, and Douglas Staiger, “Teen Motherhood and Abortion Access,”
Quarterly Journal of Economics, CXI (1996), 467–506.
Kaplan, John, “Abortion a s a Vice Crime: A ‘What if’ Story,” Law and Contempo-
rary Problems, LI (1988), 151–179.
Kelling, George L., and William J. Bratton, “Declining Crime Rates: Insiders’
Views of the New York City Story,” Journal of Criminal Law and Criminol-
ogy, LXXXVIII (1998), 1217–1232.
Levine, P. B., D. Staiger, T. J. Kane, and D. J. Zimmerman, “Roe v. Wade and
American Fertility,” National Bureau of Economic Research Working Paper
No. 5615, June 1996.
Levitt, Steven, “The Effect of Prison Population Size on Crime Rates: Evidence
from Prison Overcrowding Litigation,” Quarterly Journal of Economics, CXI
(1996), 319 –352.
——, “Using Electoral Cycles in Police Hiring to Estimate the Effect of Police on
Crime,” American Economic Review, LXXXVII (1997), 270–290.
Loeber, Rolf, and Magda Stouthamer-Loeber, “Fami ly Factors as Correlates and
Predictors of Juvenile Conduct Problems and Delinquency,” Crime and Jus-
tice, VII, Michael Tonry and Norval Morris, eds. (Chicago, IL: University of
Chicago Press, 1986).
Lott, John R., Jr., and David B. Mustard, “Cri me, Deterrence, and Right-to-Carry
Concealed Handguns,” Journal of Legal Studies, XXVI (1997), 1–68.
Marvell, Thomas, and Carlisle Moody, “Prison Population Growth and Crime
Reduction,” Journal of Quantitative Criminology, X (1994), 109 –140.
Matthews, Stephen, D Ribar, and M Wilhelm, “The Effects of Economic Co ndi-
tions and Access to Reproductive Health Service on Sta te Abortion Rates and
Birthrates,” Family Planning P erspectives, XXIX (1997), 52–60.
Merz, J. F., C. A. Jackson, and J. A. Klerman, “A Review of Abortion Policy:
Legality, Medicaid Funding, and Parental Involvement, 1967–1994,” Wom-
en’s Rights Law Reporter, XVII (1995), 1– 61.
Michael, Robert, “Abortion Decisions in the U. S.,” unpublished manuscript,
University of Chicago, 1999.
Miller, Ted, Mark Cohen , and Shelli Rossman, “Victim Costs of Violent Crime and
Resulting Injuries,” Health Affairs, XII (1993), 186–197.
O’Brien, Robert, Crime and Victimization Data (Beverly Hills, CA: Sage, 1985).
Paulsen, Michael S., “Accusing Justice: Some Variations on the Themes of Robert
M. Cover’s Justice Accused,” Journal of Law and Religion, VII (1989), 33–97.
Posner, Richard, Sex and Reason (Cambridge, MA: Harvard University Press,
1992).
Raine, Adrian, Patricia Brennan, and Sarnoff Mednick, “Birth Complications
Combined with Early Maternal Rejection at Age 1 Year Predispose to Violent
Crime at Age 18 Years,” Archives of General Psychiatry, CI (1994), 984 –988.
Raine, Adrian, Patricia Brennan, Birgitte Mednick, and Sarnoff Mednick, “High
Rates of Violence, Cri me, Academic Problems, and Beh avioral Problems in
Males with Both Early Neuromotor Decits and Unstable Family Environ-
ments,” Archives of General Psychiatry, CIII (1996), 544 –54 9.
Rasanen, Pijkko, et al., “Maternal Smoking during Pregnancy and Risk of Crim-
419LEGALIZED ABORTION AND CRIME
inal Behavior among Adult Male Offspring in the Northern Finland 1966
Birth Cohort,” American Journal of Psychiatry, CLVI (1999), 857–862.
Ruhm, Christopher, “Are Recessions Good for yo ur Health?” Quarterly Journal of
Economics, CXV (2000), 617–650.
Sah, Raj, “Social Osmosis and Patterns of Crime,” Jo urnal of Political Economy,
XCIX (1991), 1272–1295.
Sampson, Robert, and John Laub, Crime in the Making: P athways and Turning
Points through Life (Cambridge, MA: Harvard University Press, 1993).
Short, Allen, “Hatch Calls Bouza Hypocrite on Crime,” Star Tribune, Minneapolis,
MN (August 5, 199 4), 1B.
Tracy, Paul, Marvin Wolfgang, and Robert Figlio, Delinquency Careers in Two
Birth Cohorts (Washington, DC: U. S. Department of Justice, Ofce of Juve-
nile Justice and Delinquency Prevention , Nation al Institute for Juvenile
Justice and Delinquency Prevention, 1990).
Wolfgang, Marvin, Robert F. Figlio, and Thorste n Sellin, Delinquency in a Birth
Cohort (Chicago, IL: University of Chicago Pre ss, 1972.)
420 QUARTERLY JOURNAL OF ECONOMICS
