The Impact of Legalized Abortion on Crime
Total Page:16
File Type:pdf, Size:1020Kb
THE QUARTERLY JOURNAL OFECONOMICS Vol. CXVI May 2001 Issue 2 THEIMPACT OFLEGALIZED ABORTION ONCRIME* JOHN J. DONOHUE III AND STEVEN D. LEVITT Weoffer evidence that legalized abortion has contributed signi cantly to recentcrime reductions. Crime began to fall roughly eighteen years after abortion legalization.The vestates that allowed abortion in 1970 experienced declines earlierthan the rest of the nation, which legalized in 1973 with Roe v. Wade. Stateswith high abortion rates in the 1970s and 1980s experienced greater crime reductionsin the 1990s. In high abortion states, only arrests of those born after abortionlegalization fall relative to low abortion states. Legalized abortion ap- pearsto account for as muchas 50 percent of therecent drop incrime. I. INTRODUCTION Since1991, theUnited States has experiencedthe sharpest drop inmurderrates since the end of Prohibitionin 1933. Homi- ciderates have fallen morethan 40 percent.Violent crime and propertycrime have each declined more than 30 percent.Hun- dreds ofarticles discussing this changehave appeared in the academicliterature and popular press. 1 Theyhave offered an array ofexplanations:the increasing use of incarceration,growth *Wewouldlike to thank Ian Ayres, Gary Becker, CarlBell, Alfred Blumstein, JonathanCaulkins, Richard Craswell,George Fisher, Richard Freeman,James Heckman,Christine Jolls, Theodore Joyce, Louis Kaplow, Lawrence Katz, John Kennan,John Monahan, Casey Mulligan, Derek Neal,Eric Posner,Richard Posner,Sherwin Rosen, Steve Sailer, Jose ´ Scheinkman,Peter Siegelman, Kenji Yoshino,and seminar participants too numerous to mention for helpful comments anddiscussions. Craig Estesand Rose Francis provided exceptionally valuable researchassistance. Correspondence can be addressed to either John Donohue, Crown Quadrangle,Stanford Law School,Stanford, CA 94305,or StevenLevitt, Departmentof Economics,University of Chicago, 1126 E. 59thStreet, Chicago, IL 60637.Email: [email protected]; [email protected]. 1.For a samplingof the academic literature, see Blumstein and Wallman [2000]and the articles appearing in the 1998 Summer issue (Volume 88) of the Journalof Criminal Law andCriminology, especiallyBlumstein and Rosenfeld [1998],Kelling and Bratton [1998], and Donohue [1998]. See Butter eld [1997a, © 2001by thePresident and Fellowsof Harvard Collegeand theMassachusetts Institute of Technology. The Quarterly Journal ofEconomics, May 2001 379 380 QUARTERLY JOURNALOF ECONOMICS in thenumber of police, improved policing strategiessuch as thoseadopted in NewYork, declines in thecrack cocaine trade, thestrong economy, and increasedexpenditures on victim pre- cautionssuch as securityguards and alarms. Noneof these factors, however, can providean entirelysat- isfactoryexplanation forthe large, widespread, and persistent drop in crimein the1990s. Someof these trends, such as the increasingscale of imprisonment,the rise in police,and expendi- tureson victim precaution, have been ongoing for over two de- cades,and thus cannotplausibly explain therecent abrupt im- provementin crime.Moreover, the widespread natureof the crimedrop arguesagainst explanations suchas improvedpolicing techniquessince many cities that havenot improved their police forces(e.g., Los Angeles) have nonetheless seen enormous crime declines.A similarargument holds forcrack cocaine. Many areas ofthecountry that havenever had apronouncedcrack trade (for instance,suburban and ruralareas) havenonetheless experi- encedsubstantial decreasesin crime.Finally, although a strong economyis supercially consistentwith thedrop in crimesince 1991, previousresearch has established onlya weaklink between economicperformance and violentcrime [Freeman 1995] and in onecase even suggested that murderrates might vary procycli- cally [Ruhm 2000]. Whileacknowledging that all ofthesefactors may have also servedto dampen crime,we consider a novelexplanation forthe sudden crimedrop ofthe1990s: thedecision to legalize abortion overa quartercentury ago. 2 TheSupreme Court’ s 1973 decision in Roe v. Wade legalizing abortionnationwide potentially ts the criteriafor explaining alarge,abrupt, and continuingdecrease in crime.The sheer magnitude of thenumber of abortionsperformed satises the rstcriterion that any shockunderlying the recent drop in crimemust be substantial. Sevenyears after Roe v. Wade, over1.6 millionabortions were being performed annually— al- mostone abortion for every two live births. Moreover, the legal- 1997b,1999] for a selectionof articles appearing in The New YorkTimes and Fletcher[2000] for a recentarticle in The Washington Post. 2.We areunaware of anyscholarly article that has examined this effect. We haverecently learned, however, that the former police chief of Minneapolis has writtenthat abortion is “ arguablythe only effective crime-prevention device adoptedin thisnation since the late 1960s” [Bouza 1990]. In hissubsequent 1994 gubernatorialcampaign, Bouza was attacked forthis opinion [Short 1994].Im- mediatelyafter Bouza’ s viewwas publicized just prior to the election, Bouza fell sharplyin the polls. LEGALIZED ABORTION AND CRIME 381 ization ofabortionin vestates in 1970, and thenfor the nation as awholein 1973, wereabrupt legaldevelopments that might plausibly havea similarlyabrupt inuence 15– 20 yearslater whenthe cohorts born in thewake of liberalized abortionwould start reachingtheir high-crime years. Finally, any inuence of a changein abortionwould impact crimecumulatively as succes- siveaffected cohortsentered into their high-crime late adolescent years,providing areasonwhy crimehas continuedto fall year after year. Legalizedabortion may lead toreducedcrime either through reductionsin cohortsizes or through lower per capita offending ratesfor affected cohorts.The smaller cohort that resultsfrom abortionlegalization meansthat whenthat cohortreaches the lateteens and twenties,there will befewer young males in their highest-crimeyears, and thus lesscrime. More interesting and importantis thepossibility that childrenborn after abortion legalization mayon averagehave lower subsequent rates of crim- inality foreither of tworeasons. First, women who have abortions arethose most at riskto give birth tochildren who would engage in criminalactivity. Teenagers, unmarried women, and theeco- nomicallydisadvantaged areall substantially morelikely to seek abortions[Levine et al.1996]. Recentstudies havefound children bornto these mothers to beat higherrisk for committing crime in adolescence[Comanor and Phillips 1999]. Gruber,Levine, and Staiger[1999], in thepaper mostsimilar to ours, document that theearly life circumstancesof those children on the margin of abortionare dif cult along many dimensions: infant mortality, growingup in asingle-parent family,and experiencingpoverty. Second,women may use abortion to optimize the timing of child- bearing.A givenwoman’ s ability toprovidea nurturingenviron- mentto a child can uctuateover time depending onthewoman’ s age,education, and income,as wellas thepresence of afatherin thechild’ s life,whether the pregnancy is wanted,and any drug or alcoholabuse both in uteroand afterthe birth. Consequently, legalized abortionprovides a womanthe opportunity to delay childbearing if thecurrent conditions are suboptimal. Even if lifetimefertility remainsconstant for all women,children are borninto better environments, and futurecriminality is likelyto be reduced. Anumberof anecdotalempirical facts support theexistence and magnitudeof thecrime-reducing impact ofabortion.First, we seea broad consistencywith thetiming of legalization ofabortion 382 QUARTERLY JOURNALOF ECONOMICS and thesubsequent drop in crime.For example, the peak ages for violentcrime are roughly 18 –24, and crimestarts turningdown around1992, roughlythe time at whichthe rstcohort born following Roe v. Wade would hit its criminalprime. Second, as we laterdemonstrate, the vestates that legalized abortionin 1970 saw drops in crimebefore the other 45 statesand theDistrict of Columbia,which did notallow abortionsuntil theSupreme Court decisionin 1973. Third,our more formal analysis showsthat higherrates of abortionin astatein the1970s and early1980s arestrongly linkedto lower crime over the period from 1985 to1997. This nding is trueafter controlling for a varietyof factorsthat inu- encecrime, such as thelevel of incarceration, the number of police,and measuresof the state’ s economicwell-being (the un- employmentrate, income per capita, and povertyrate). The esti- matedmagnitude of the impact oflegalized abortionon crime is large.According to our estimates, as shownon Table II, states with high ratesof abortionhave experienced roughly a 30 percent drop in crimerelative to low-abortion regions since 1985. While onemust be cautious in extrapolatingour results out of sample, theestimates suggest that legalized abortioncan accountfor abouthalf theobserved decline in crimein theUnited States between1991 and 1997. Anumberof factorslead ustobelieve that thelink between abortionand crimeis causal.First, there is norelationship be- tweenabortion rates in themid-1970s and crimechanges be- tween1972 and 1985 (prior tothe point whenthe abortion- affected cohortshave reached the age of signi cant criminal involvement).Second, virtually all ofthe abortion-related crime decreasecan beattributed toreductions in crimeamong the cohortsborn after abortion legalization. There is littlechange in crimeamong older cohorts. Weshould emphasizethat ourgoal is tounderstand why crimehas fallen sharply in the1990s, and toexplore the contri-