The Impacts of Reduced Access to Abortion and Family Planning Services on Abortion, Births, and Contraceptive Purchases ∗ Stefanie Fischer Heather Royer Corey White Cal Poly State University, University of California, Cal Poly State University, San Luis Obispo Santa Barbara, NBER, San Luis Obispo IZA sjfi[email protected] [email protected] [email protected] May 13, 2018 Abstract Between 2011 and 2014, Texas enacted three pieces of legislation that significantly reduced funding for family planning services and increased restrictions on abortion clinic operations. Together this legislation creates cross-county variation in access to abortion and family plan- ning services, which we leverage to understand the impact of family planning and abortion clinic access on abortions, births, and contraceptive purchases. In response to these policies, abortionstoTexasresidentsfell16.7%andbirthsrose1.3%incountiesthatnolongerhadan abortionproviderwithin50miles. Changesinthefamilyplanningmarketinduceda1.2%in- creaseinbirthsforcountiesthatnolongerhadapubliclyfundedfamilyplanningclinicwithin 25miles. Meanwhile,responsesofretailpurchasesofcondomsandemergencycontraceptives tobothabortionandfamilyplanningservicechangeswereminimal. JELCodes: J13,I18,J08,J18,I38 Keywords: familyplanning,abortion,birth,contraception,reproductive,health ∗WethankrepresentativesfromFundTexasChoiceandJaclynBaysfromWomen’sHealthandFamilyPlanning AssociationofTexasforprovidingdata. ThispaperhasbenefitedfromhelpfulcommentsfromMarthaBailey, Tom Chang,ChloeEast,MireilleJacobson,CaitlinMyers,AnalisaPackham,MayaRossin-SlaterandJennaStearns. We thankMacBrownforassistancewithgeocoding. Wealsothankparticipantsatthefollowingconferences: IZAWork- shoponGenderandFamilyEconomics,SocietyofLaborEconomicsAnnualMeetings,SouthernCaliforniaConfer- enceforAppliedMicroeconomics,CSWEPSessiononFamilyandGenderatWEAIAnnualConference,International HealthEconomicsAssociationBiennialWorldCongress, aswellasseminarparticipantsatUCSB’sLaborSeminar, Cal Poly SLO, San Diego State, University of Hawaii, University of California Riverside, and Cal State Fullerton. CertaincalculationsinthispaperarebasedondatafromtheNielsenCompany(U.S.),LLCandmarketingdatabases providedbytheKiltsCenterforMarketingDataCenteratTheUniversityofChicagoBoothSchoolofBusiness. The conclusionsdrawnfromtheNielsendataarethoseoftheresearchersanddonotreflecttheviewsofNielsen. Nielsen isnotresponsiblefor,hadnorolein,andwasnotinvolvedinanalyzingandpreparingtheresultsreportedherein. 2 1 Introduction Access to abortion and family planning services has declined precipitously over the past decade. Between2008and2014,thenumberoffacilitiesprovidingabortionsintheUnitedStatesfell6.8%, continuingalongdeclinesincetheearly1980s. Insomestates,includingTexas,thisdrophasbeen even more dramatic: the number of abortion-providing clinics shrunk by at least 25% in 10 states over the 2008 to 2014 period (Jones and Jerman, 2014, 2017).1 Coinciding with this, the abortion rateisatitslowestlevelsincetheadoptionofRoev. Wade.2 In parallel, the funding of family planning services, which primarily include the dispensary of contraceptives, pregnancy testing, sexually transmitted infections (STIs) testing and treatment, primary care, cancer screenings, and preconception and prenatal care, has similarly decreased (Zolna and Frost, 2016). Per capita funding levels of Title X, the federal program devoted solely to the provision of family planning services and targeted to low-income women, hit their peak in 2010 and have fallen subsequently.3 At its apex of funding, one in four women (and nearly half of poor women) who received contraceptive services did so at a publicly funded clinic.4 Funding cuts to family planning services, including Title X, are likely to continue given the current health care discussions. In 2017, President Trump signed legislation allowing states to withhold Title X fundsfromfamilyplanningclinicsthatareaffiliatedwithabortionproviders.5 In this study, we exploit three recent policy changes in Texas to separately understand the ef- fects of reductions in access to abortion and family planning services. Over the 2011 to 2014 period, the Texas legislature implemented legislation that both limited the ability of non-abortion family planning providers to receive government funding and placed more stringent requirements on the operation of abortion clinics. In the aftermath of these policies, over half of abortion clin- ics closed by 2015, family planning providers experienced funding cuts of 66% and one-quarter 1Note, while the change in Texas is large, it is not an outlier. There are seven states with at least as large of a declineinabortion-providingclinicsoverthistimeperiod. 2Seehttp://www.latimes.com/nation/la-na-abortion-rate-2017-story.html. 3Seehttps://www.hhs.gov/opa/title-x-family-planning/about-title-x-grants/funding-history/index.html. 4Source: https://www.guttmacher.org/fact-sheet/publicly-funded-family-planning-services-united-states. 5TitleXfundinghasneverbeenavailableforabortionservices. 3 of publicly funded family planning clinics closed (White et al., 2015). The first two pieces of legislationconcernedfamilyplanningfundingandthelastimpactedabortionclinics. As access to these services may affect fertility decisions on multiple margins, we focus on three sets of outcomes to better understand how they affect fertility behavior: abortions, births, and contraceptive purchases. Our analysis leverages spatial and temporal variation in access to reproductive services across counties in Texas using a difference-in-difference design with county fixedeffects. Usingdataonthelocationofabortionprovidersandpubliclyfundedfamilyplanning clinics over time, we operationalize the changes in access by focusing on changes in distance to the nearest abortion or publicly funded family planning provider. We define a publicly funded family planning clinic as one that receives state or federal funding. For abortion providers, our measureofaccessexploitsclosureswhereasforfamilyplanningclinics,itleveragesbothclosures and changes in the source of funding (e.g., from public funding to non-public funding). Overall, due to the reduced funding, the number of family planning clinics fell and, for many of those that remained open, so did their ability to serve their customer base. As the impacts of distance are unlikelytobelinear,ourmeasuresofaccessaredichotomous-whetherornotthereisanabortion or family planning clinic within a pre-specified driving distance. For abortion access, much of the action operates on whether or not there is an abortion provider within 50 or 100 miles. For family planning, not surprisingly, the most impactful distance is shorter: 25 miles. In 2015, 24% of the Texas population had no abortion clinic within 50 miles and 11% had no publicly funded family planningcliniccloserthan25miles. How might the reductions in abortion and family planning access impact fertility outcomes such as abortions and births? A priori, the effects are ambiguous. Reduced access to abortion clinics could cause a woman to have a child when she otherwise would not have, leading to fewer abortions and an increase in births. Alternatively, forward-looking individuals may practice safer sex or abstain, resulting in fewer abortions and potentially lower fertility rates.6 If the increased distance is not prohibitive, one might expect no alteration in either births or abortions. Similarly, 6ThisisthebasicfindingofKaneandStaiger(1996)forteenagersintheresponsetotheclosingofabortionclinics anddeclinesinMedicaidfunding. 4 the effect of reduced access to family planning services may also be ambiguous. Reduced access may lessen the frequency of contraceptive use, such as IUDs and condoms, which are often dis- pensed for free or reduced cost at such clinics. As a result, the incidence of unintended pregnancy may rise, possibly leading to either increased abortions, increased births, or both. The impact of family planning services may also operate through sexual education and family planning practice knowledge. In this case, it would be reasonable to expect fertility rates to increase with more restrictedaccesstofamilyplanningservices. Several features make Texas an interesting and useful setting for studying access to abortion and family planning. First, the policies examined here are reflective of those currently on the policy agenda nationwide. Second, estimated effects in Texas are likely more informative about theeffectsofnationwidepolicychangescomparedtotheanalysisofotherstates. BecauseofTexas’ size,travelacrossstatelinestootherstatesislessfeasibleformostresidents. Third,unlikeinmost other states, by law, family planning services are administered separately from abortion services, and thus, we can separately estimate effects of changes in access to abortion and family planning services.7 Fourth,Texasmaintainsaconsistentandhigh-qualitysetofdataonabortionsbycounty and age. National abortion data are limited and the quality (i.e., completeness) of state-level data varysignificantly(JacobsonandRoyer,2011). At first glance, the effects of this legislation look dramatic as seen in Figure 1. This figure dis- plays the time-series patterns of births in Texas alongside a synthetic control for Texas. The three vertical bars represent the three pieces of legislation we exploit - first, the Texas Department of State Health Services (TDSHS) cuts in 2011 reduced funding for family planning clinics by 67%; second, the Women’s Health Program (WHP) effectively eliminated Medicaid fee-for-service re- imbursementoffamilyplanningservicesforPlannedParenthoodaffiliatesinearly2013;andthird, later that year, House Bill 2 (HB2) imposed significant regulations on the operation of abortion providers. The fertility rates for Texas and its synthetic control begin to diverge slightly after the enactmentoftheTDSHScutsandthepaceofseparationaccelerateswiththeWHPlegislationand 7TheotherstateswithsimilarpoliciesincludeArizona,Arkansas,Colorado,Indiana,Ohio,andWisconsin.Source: https://www.guttmacher.org/state-policy/explore/state-family-planning-funding-restrictions. 5 HB2.8 Ourprimaryidentificationstrategyexploitsquasi-experimentalvariationinaccessacrosscoun- ties within Texas rather than statewide variation as in Figure 1. First, we find that having no abor- tion provider within 50 miles reduces the observed number of abortions by 16.7%. This estimate may not capture the true effect on abortions as women could travel to other states not covered by ourdatatoreceiveanabortionorcouldself-administeranabortion.9 Forthisreason,theimpactof the reduction in abortion access on births, a 1.3% increase, is more informative of the total effect on fertility-related behaviors. The effect ofreduced family planning access onbirths, as measured by whether or not there is a funded clinic within 25 miles, is similar. Overall, not having a funded clinic within 25 miles increases births by 1.2%. The effects are heterogenous across different de- mographic groups, and the groups most impacted by reduced access to family planning services are distinct from those most affected by reduced abortion access which highlights the importance ofseparatelyestimatingtheeffectofaccesstoeachtypeofclinic. While it is standard in the abortion and family planning literature to focus on the outcomes of abortions and births, such analyses miss impacts on precautionary behaviors (e.g., contraceptive use). This is mainly due to data limitations rather than a lack of interest. Most utilized data (e.g., National Survey of Youth Women as used in Goldin and Katz (2002) or the National Survey of FamilyGrowth)areretrospectiveandmeasuredatlowfrequencies. EarlierpapersbyAkerlofetal. (1996) and Kane and Staiger (1996) develop theoretical models showing that fertility-impacting policies could influence the use of contraceptives. We use a new source of data on contracep- tivesbasedonweeklyretailpurchasesofcondomsandemergencycontraceptivesfromtheNielsen 8TheTDSHScutsimpactbirthswithadelay. Inourlateranalysis, theeffectsofthechangesinfamilyplanning servicesactwitha1-yeardelay. Therearetwopossibleexplanationsforthis. First,giventhelengthoftimebetween conceptionandbirthof40weeks,thereisadelaybetweenthepolicy’senactmentandtheobservedeffectofthepolicy. Second, one of the most common services of family planning providers is the insertion of intrauterine birth control devices(IUDs), whichhavelifespansofseveralyears. Thus, whileareducedabilitytoprovideIUDswillaffectthe flowofwomenreceivingIUDs,theeffectonthestockofwomenwithIUDs,therelevantat-riskgroup,takeslonger tomanifest. 9TheTexasPolicyEvaluationProjectatTheUniversityofTexasatAustinestimatesthatatleast100,000women in Texas have attempted a self-induced abortion. This statistic is likely higher in Texas than in other states due to the close proximity of Texas with Mexico where misoprostol, an abortion-inducing drug, is available at pharmacies withoutaprescription. Seehttp://liberalarts.utexas.edu/txpep/news/article.php?id=10043. 6 Retail Scanner database. Contraceptive purchases exhibit little response to the changes in repro- ductiveservices. We provide several specification checks to ensure that we are identifying the effect of access ratherthanpotentiallycoincidingfactors. Ourexploitedquasi-experimentalvariationoccursonthe cusp of the Great Recession when fertility rates were falling. To understand whether our results are biased by trends caused by the Great Recession or other factors, we conduct several tests such as limiting the time period of our analysis or including region-specific time trends, and we obtain similar results. We also attempt to predict changes in clinic access using pre-policy changes in fertility rates and find no statistically significant relationship, further suggesting that differential pre-trends in the outcome are unlikely to be biasing our results.10 One caveat to our work is that we use cross-county and cross-time variation within Texas, effectively contrasting more affected with less affected counties in Texas. This contrast, of course, will miss the overall effect of the legislativechangesonreproductiveservicesinTexas. Toascertainhowthisaffectstheconclusions of our analysis, we compare the time trends in Texas with a synthetic control as shown earlier in Figure 1. While one might argue that such analysis is not as credible as those produced from ourmainidentificationapproach,theestimatedeffectsonbirthsareofsimilarmagnitude-a2.8% increase using the synthetic control approach, compared to 1.3% and 1.2% increases for abortion andfamilyplanningaccess,respectively(acombinedeffectof2.5%). Thisstudycomplementstheextensivepreviousworkonfamilyplanningandabortionservices in three important ways. First, we focus on a substantial and significant contraction in family planningandabortionservices. Muchoftheexistingliteraturefocusesonearlyexpansionsinfam- ily planning and abortion access (e.g., Roe v. Wade and the adoption of the birth control pill).11 Exceptionsincludetheimplementationofparentalconsentandnotificationlaws(SabiaandAnder- 10Thisindirecttestforparalleltrendsinthepre-policyperiodissimilartoaprocedureusedinLahey(2014). 11For example, many studies have examined the expansion of oral contraceptives and show that it led to delayed childbearing,reducedfertility,increasedcareerinvestmentforwomen,andbetterchildoutcomes(Myers,2017;Bai- ley, 2013; Ananat and Hungerman, 2012; Bailey, 2012, 2010; Kearney and Levine, 2009; Bailey, 2006; Goldin and Katz,2002). Usingmorerecentvariationincontraceptives,Grossetal.(2014)showthatexpandingaccesstoemer- gency contraceptives has little effect on birth or abortion rates, while Lindo and Packham (2015) conclude that ex- panded access to long-acting reversible contraceptives through the Colorado’s Family Planning Initiative reduced teenagefertilityrates. 7 son,2016;Colmanetal.,2013;Guldi,2008;Joyceetal.,2006;Levine,2003;Averettetal.,2002; Levine, 2001; Blank et al., 1996; Joyce and Kaestner, 1996) which only affect teenagers, waiting periods (Bitler and Zavodny, 2001) and acts of violence at abortion clinics (Jacobson and Royer, 2011). Given the current policy environment and the fact that contractions in coverage may incur differentimpactsthanexpansionsincoverage,ourstudyisrelevantforunderstandingtheeffectsof policies under debate today. Second, we focus on contemporary variation in access (i.e., changes withinthelastdecade). Duringthe2000’s,contraceptivetechnologies(e.g.,IUDs,hormonalpatch, vaginal ring, and female condom) improved in terms of their effectiveness and safety, and emer- gency contraceptives entered the market.12 Moreover, over time, female labor force participation has markedly increased, making a woman’s decision to bear a child more complicated. Thus, fertility-related policies today might affect behaviors differently from the past. Third, the distinct and separate quasi-experimental variation in abortion clinic access and family planning services (the correlation in the variation is 0.16) combined with the legislative environment in Texas (i.e., fundedfamilyplanningclinicsareprohibitedfromprovidingabortionservices)allowsustoisolate theimpactofchangesinfamilyplanningservicesfromtheeffectofchangestoabortionservices. More directly, our paper adds to the literature evaluating the impacts of recent changes to reproductive services in Texas. Several findings from recent and concurrent work emerge: 1) within-county changes in distance to the nearest abortion provider strongly correlate with within- county changes in abortion rates – varying from 10 to 50% depending on the change in distance (Quast et al., 2017; Grossman et al., 2017; Cunningham et al., 2018), 2) by the start of 2013, the closureofclinicswithinonelargenetworkoffamilyplanningprovidersledtoanincreaseofbirth rates of 1.2% for every 100 mile increase in the distance to the nearest facility (Lu and Slusky, 2017), 3) teenagers were particularly susceptible to the 2011 family planning cuts with birth rates increasing3.4%asaresult(Packham,2017),and4)theexclusionofPlannedParenthoodaffiliates fromMedicaidledto30%declinesinlong-actingreversiblecontraceptives(LARCs)andinjectable contraceptives among Medicaid recipients in areas with Planned Parenthood affiliates (Stevenson 12Seehttp://www.ourbodiesourselves.org/health-info/a-brief-history-of-birth-control/. 8 et al., 2016). While this literature confirms that the policies impacted reproductive decisions and outcomes, it is difficult to characterize the complete effect of the Texas legislation because of the priorliterature’sfocusonsubpopulations,somebutnotallofthelegislativechanges,andalimited setofoutcomes. The goal of our paper is to provide a more comprehensive look at the impact of the Texas legislative changes - examining both the effects of abortion and family planning access. Using data through 2015 (and data on births through 2016), we study a broader set of fertility outcomes in an attempt to gain a better understanding of the different ways in which Texas women are changing their fertility behavior. These outcomes include abortions, births and over-the-counter contraceptive purchases. Impacts of abortion access on births have largely been overlooked with the exception of the concurrent working paper Cunningham et al. (2018). Cunningham et al. (2018) examine the effects of abortion access on abortions and births; while they find similar impacts of abortion access on abortions, they conclude that there are no detectable birth effects. This difference with our findings is discussed in more detail in Section 5.2. Effects on retail contraceptive purchases are unknown even though the results of Stevenson et al. (2016) open up thepossibilityofcompensatorybehavior. 2 Background In Subsection 2.1 we describe the policy setting in the U.S. and Texas. Subsection 2.2 includes a detaileddiscussionofthethreepiecesofTexaslegislationleveragedinthisstudy. 2.1 Policy Setting In2014,therewere1,795,160womeninneedoffreeorsubsidizedreproductivecareinTexas. The Guttmacher Institute characterizes a woman in need if she is sexually active, is able to conceive, wishes not to become pregnant, and is an adult with a family income below 250% of the federal poverty level or is younger than 20 years of age (regardless of income). Publicly funded family 9 planning clinics serve these women and they encompass a diverse set of health providers includ- ing public health departments, Federally Qualified Health Centers (FQHC), Planned Parenthood affiliates, hospital outpatient clinics, and other independent non-profit health centers.13 Services providedbypubliclyfundedclinicsincludefreeorsubsidizedcontraceptives,screeningsforSTIs, Pap tests, vaccination for human papilloma virus (HPV), other key preventive care services, and sexualeducation. Publicly funded clinics may receive a variety of federal and state grants. Title X, one of the main funding sources for family planning clinics, is a federal program dedicated to family plan- ning.14 Congress introduced Title X in 1970 as part of the Public Health Service Act. The goal of this legislation was to make family planning services available to women who wanted them but could not afford them. Today, Title X clinics still play a critical role in ensuring all women have access to family planning services. It remains the only federally funded program dedicated solely toprovidingreproductivecaretolowincomeanduninsuredindividuals. ReceiptofTitleXfundsis alsotiedtootherfederalprograms. ClinicsreceivingTitleXfundsareeligibleforthefederal340B DrugPricingProgram,whichprovidesdiscountsonpharmaceuticals(includingcontraceptives)of up to 50%. Clinics receiving Title X funds are exempt from state-level parental consent laws (in- cluding Texas) that require teenage women to gain parental consent before obtaining prescription contraceptives.15 Despitethislargedemandforsubsidizedreproductiveservices,publicfundingfortheseclinics remainscontroversialamongpolicymakersandthegeneralpublic. Thosethatseektolimitfunding oftenviewtheseclinicsascloselytiedtoabortionclinics. Federallawdoespermitpubliclyfunded family planning clinics to provide abortions, but it is unlawful for federal dollars to fund such procedures. CertainstatesincludingTexas,however,goastepfurtheranddisallowfamilyplanning 13TherealsoexistsintheU.S.asmallnumberofnon-profitfamilyplanningclinicsthatarefundedexclusivelyby privatecontributionsandreceivenotaxdollars,thoughinTexasthesearequiterare. 14Publicly funded family planning clinics are also funded by Medicaid (the largest source of funding), Title V (maternalandchildhealth),andTitleXX(socialservices). 15Note that laws requiring parental consent for prescription contraceptives are distinct from those that require parental consent for abortion; only Texas and Utah have laws requiring parental consent for prescription contracep- tives. 10 clinics that receive any public funding from providing abortions. This law has been in effect in Texassince2003.16 Proponentsofthelawbelievethatintheabsenceofsuchrestrictions,abortion services are indirectly funded as clinics can use public funds for eligible services, thereby freeing upnon-publicfundsforabortions. Because of these federal and state laws, in Texas there are two distinct types of clinics that are relevant to our study: stand-alone abortion clinics that exclusively provide abortion services and areprivatelyfunded,andpubliclyfundedfamilyplanningclinicswhichprovidecontraceptivecare andotherservices,butnot abortions.17 2.2 Legislative Background We leverage three state-level reproductive policies that create large unanticipated shocks to the supplyofabortionandnon-abortionfamilyplanningservicesinTexas: (1)the2011cuttoTDSHS forpubliclyfundedfamilyplanningservices,(2)the2011changeintheWomen’sHealthProgram which was rolled out in 2013 and disallowed certain clinics from receiving Medicaid reimburse- ments, and (3) House Bill 2 which took effect at the end of 2013 and the beginning of 2014 and greatlyreducedaccesstoabortionclinics. 2.2.1 CutstoTDSHSFunding In2011,theTexasgovernmentenactedtwopiecesoflegislationwhichdrasticallycutfundstofam- ilyplanningclinicsinthestate,andinparticularreducedfundingtoPlannedParenthoodaffiliates. ThefirstfundingcutreducedtheTDSHSbudgetforfamilyplanningservices. Previously,TDSHS funded clinics through federal and state grants including Title V (maternal and child health), Title X(familyplanning)andTitleXX(socialservices). The budget cut to TDSHS reduced funding by about 67% – a cut from $111 million per bi- 16SeeSeventy-eighthLegislature,RegularSessionGeneralAppropriationsAct. 17Womenmayalsoobtainabortionsfromahospitalorageneralpractitioner,butinTexasin2012–accordingtoa reportbyTDSHS–only0.3%ofallabortionstookplaceinthesetypesoffacilities,wheremostoftheminvolvedan extenuatingcircumstance(i.e., ectopicpregnancy). Byandlarge, Texaswomenobtaintheseservicesinstand-alone clinics.