i

Sex Ratio at Birth in IndiaRecent Trends and Patterns

Purushottam M. Kulkarni

(A report prepared for the United Nations Population Fund)

Sex Ratio at Birth in India- Recent Trends and Patterns

© UNFPA 2019

Published in 2020

Published by: United Nations Population Fund (UNFPA) 55 Lodi Estate, New Delhi 110003 India Tel: +91-11-4632333 E-mail:india.office@unfpa.org

Citation Advice: United Nations Population Fund 2020. ‘Sex Ratio at Birth in India: RecentTrendsandPatterns’2020.UNFPA,NewDelhi,India

DESIGN AspireDesign,NewDelhi

DISCLAIMER All rights reserved. The contents, analysis, opinions and recommendationsexpressedinthisreportaresolelytheviewsoftheauthoranddonotnecessarilyrepresent the views of the United Nations Population Fund – UNFPA. Thedocumentmaybequoted,reproducedortranslated,eitherinpartorfull,withdueacknowledgements.UNFPAwillnot,inanyway,beliablefortheuse,orany consequencesarisingout of theuseof anyother information from thisreport.Thisreportisforinformingandeducatingthepublicatlargeandisnottobesoldorusedforcommercialpurposes.

ii

iii

Contents

Foreword v

Acknowledgements vii

List of Abbreviations ix

Executive Summary xi

1. Introduction 1

2. Estimates of Sex Ratio at Birth from Different Sources of Data 3

2.1CivilRegistrationSystem 3

2.2SampleRegistrationSystem(SRS) 3

2.3 Census 4

2.4NationalFamilyHealthSurveys(NFHS) 6

2.5HealthManagementInformationSystem(HMIS) 7

2.6Regionalvariations 8

3. Correspondence among Various Estimates 9

4. Estimates of Numbers of Missing Girls and Missing Female Births 13

5. Sex Ratio at Birth by Birth Order 18

6. Conditional Sex Ratio at Birth 20

7. Differentials in Sex Ratio at Birth 23

7.1PriorworkforIndia 23

7.2AnalysisbasedonNFHS-4 24

8. Why Sex Selection? 33

8.1 Evidenceonvaluesofsonsvis-à-visdaughtersinIndia 33

9. Principal Findings 42

References 45

Annexures 50

iv

v

Foreword

Sonpreferenceisadeep-rootedculturalphenomenoninmanycountriesincludingIndia,whichresultsindiscriminatoryandharmfulpracticesagainstwomenandgirls.Itmanifestsitselfthroughprenatalandpost-nataldiscriminationagainstgirlsintheformofgender-biasedsexselection.TheSustainableDevelopmentGoalsby2030 includea target (5.1)onendingharmfulpracticesagainstwomenandgirls.For years, as part of itsmandate, the United Nations Population Fund (UNFPA)has focused on this issue, guided by the International Conference on PopulationandDevelopment (ICPD) Programme ofAction of 1994 in Cairo. Thiswas furtherreinforcedattheNairobisummitonICPD25lastyearwhenheadsofgovernments,civil society and grassroots organizations committed to eliminate all forms ofharmfulpractices.

Gender-biasedsexselectionismeasuredthroughsexratioatbirth,acomparisonofthenumberofgirlsversusthenumberofboysborninagivenyear.Thisrequiresaccurateandreliabledataonsexratioatbirthtojudgetheextentofimbalances.InIndia,severalsourcesprovidethisindicator,however,mostoftentheseestimatesdonotagreetoeachother.Thisreportanalysesvariousestimatesofsexratioatbirthandattemptstoarriveatthemostplausiblelevel,afterapplyingacorrectionfactor.

Further, this report elaborates on the severity of gender-biased sex selection byexamining the pre and post-natal discrimination against girls, the estimation ofnumberofgirlsmissingatbirth,thepracticeofsex-selectionbybirthorderandthesexcompositionoftheexistingnumberofchildren,basedonbackgroundcharacteristicsandbygeographicalregions.Thereportalsoexploresvariousfactorsassociatedwiththisphenomenonandthereasonsbehindsuchdiscriminatorypractices.

TheGovernmentofIndiahasenactedseverallawstobanpre-natalsexdetectiontocurbthepracticeofgender-biasedsexselectionandinitiatedseveralprogrammestoenhancethevalueofthegirlchild.Civilsocietyorganisationshavelongbeenactivelyengagedinlarge-scalecampaignstoaddresssonpreference.Althoughtrendsofthesexratioatbirthsuggestthatthereisacontinuedpreferenceforsonsinthecountry.Oneofthecriticalaspectsofthisanalysisisthepost-natalneglectofthegirlchild,whichhasbeenacknowledgedinpublicdiscourse,butstillneedsappropriatemeasurestoaddressit.IwishtothankProf.P.M.Kulkarniforhiseffortsinbringingoutthisreport.

I hope that the analysis and insights from this report, looking specifically at thisharmfulpractice inIndia,willcomplementUNFPA’s StateoftheWorldPopulation2020Reportthatfocuseson19harmfulpracticesthatdiscriminateagainstwomenandgirls.Further,Ihopethisreportwillbeusefulforpolicymakers,programmeplanners,academiaandcivilsocietyorganisationstodeviseevidencebasedstrategiestoaddressthesexratioimbalanceinthecountry.

Argentina Matavel PiccinUNFPA Representative India and Country Director Bhutan

vi

vii

Acknowledgements

TheriseinmasculinityinIndia’ssexratioatbirthhasbeenamatterofdeepconcernforquitesometime.Inviewofthis,theUnitedNationsPopulationFund(UNFPA) commissioned this study to examine the current situation regardingsexratioatbirthinIndia,obtainestimatesofmissingfemalebirthsandmissinggirls, and analyse influences of various factors on pre-natal and post–nataldiscrimination.TheinitiativeforthisresearchcamefromMs.ArgentinaMatavelPiccin,Representative,UNFPA,India.Thereporthasbenefitedonaccountofthesupport of anumber of officers fromUNFPA, India.Ms.EnaSingh,AssistantRepresentative, UNFPA, India, took keen interest in the study and providedvaluable suggestions. Sanjay Kumar has been associated with the researchthroughouttheentirecourseoftheproject.Heexaminedthemethodology,drewattentiontoimportantrelevantsourcesofdata,andgavefeedbackonanalysisand presentation. Sanjay’s contribution to the study has been invaluable.InteractionswithDhanashriBrahmeandShobhanaBoyleduringtheconductofthestudyandtheircommentsonearlierdraftshavesignificantlystrengthenedthe report. Hemant Bajaj and Laetitia Jones Mukhim provided efficientadministrativesupport.

The present work has benefited immensely from discussions with ProfessorsChristopheGuilmoto,CEPED,Paris,RavinderKaur,IIT,DelhiandMaryJohn,CentreforWomen‘sDevelopmentStudies,NewDelhi,eminentresearchersinthefield.

TheanalysisinthereporthasusedtheunitleveldatafromvariousroundsoftheNationalFamilyHealthSurvey(NFHS)andtheIndiaHumanDevelopmentSurvey (IHDS). Access provided to the data files by the survey organisations,the International Institute for Population Sciences (IIPS), Mumbai and theUniversityofMarylandandtheNationalCouncilofAppliedEconomicResearch(NCAER),NewDelhi,isgratefullyacknowledged.

Purushottom M. Kulkarni

viii

ix

List of Abbreviations

BLY Birthslastyear

CEB Childreneverborn

CRS CivilRegistrationSystem

EFCM Excessfemalechildhoodmortality

GBSS Genderbiasedsexselection

HMIS HealthManagementInformationSystem

IHDS TheIndiaHumanDevelopmentSurvey

IIPS InternationalInstituteforPopulationSciences

MA MovingAverage

MCA MultipleClassificationAnalysis

NFHS NationalFamilyHealthSurvey

OBC OtherBackwardCaste

ORGI OfficeoftheRegistrarGeneral&CensusCommissioner,India

PCPNDT Pre-ConceptionandPre-NatalDiagnosticTechniques

PNDT Pre-NatalDiagnosticTechniques

PES Postenumerationsurvey

SC ScheduledCaste

SFMS SpecialFertilityandMortalitySurvey

SRB Sexratioatbirth

SRS SampleRegistrationSystem

ST ScheduledTribe

UNFPA United Nations Population Fund

x

xi

Thesexratioatbirth(SRB)inIndiahasbecomemore masculine in the recent decades. Theimbalance in sex ratios stems from strong

son preference combined with declining fertility,and the availability of and access to sonographicscanningduringpregnancy.ThepracticeofgenderbiasedsexselectioncontinueseventhoughIndiahasenactedlawsbanningtheuseofpre-nataldiagnostictechniquesforsexdetection.Theinstancesofgenderbiasedsexselectionareobviouslynotrecordedbutthe numbers of cases can be estimated indirectlybasedonthedeviationoftheobservedSRBfromthenaturallevel.Tothisend,thisstudyfirstexamineddataonIndia’sSRBfromvarioussources,identifiedthe most plausible estimates, and then used these to estimate thenumbers ofmissing female births.Further,thestudyestimatesthenumberofmissinggirls based on the 2011 census enumeration andpresentsthedecompositionofthemissingnumbersby two factors, pre-natal discrimination (sexselection at birth) and post-natal discrimination(excess female childhood mortality). For the five-year period beyond the 2011 census, the studyestimates gender biased sex selection and excessdeathsofgirlsbelowagefive.Thestudygoesastepfurther and presents variations in the SRB by thestageof familybuilding, that is,atdifferentbirthordersandbythesexcompositionofpreviousbirths,up to the third order. In its analysis, the studyexamines, socioeconomic and spatial differentialsintheSRBatvariousstagesoffamilybuildingandassessesthenetinfluencesofvariousfactorsontheprobabilityofamalebirth.Finally,thestudylooksatrecentevidenceonreasonsforsonpreferenceand,inparticular,onthevalueaccordedtosonsvis-à-visdaughters.Themainresultsarepresentedbelow.

The SRB in India is clearly more masculine thanthe natural level. In the absence of sex selectionthe SRB is around 105male births per 100 femalebirths or around 950 female births per 1000malebirthswhereasinIndiathenumberoffemalebirthsper 1000 male births ratio has been much below950intherecentdecades.EstimatesoftheSRBareavailable from various sources, and an assessment

of these revealed that the census based indirectestimate obtained by reverse survival is the mostplausible one. At the national level, this was 923female births per 1000male births for the period2004-2011. The sample registration system (SRS)estimateoftheSRBforthisperiodis903andseemsto be an underestimate (whenmeasured in termsoffemalesper1000males)byabouttwopercentatthe national level and needs to be corrected; thecorrection factor varies somewhat for states. TheSRB has been fluctuating in the range 900 to 930female births per 1000male births since 2000 forIndiawithnocleartrend.

TheregionalpatternintheSRBiswellrecognized.States in the northern-western region showmuchmoremasculineSRBthanintheotherregions;somestatesinthecentralregionalsoshowlowratiosbutnot to the levels of the northern-western regions.The eastern, northeastern, and southern regionsgenerally show ratios near natural. In Punjab,Jammu and Kashmir, and Himachal Pradesh theSRBseemstohaverisenbutisstill lowerthanthenaturallevel.

It is estimated that close to 400 thousand femalebirthsaremissed in Indiaannuallyasaresultofgender biased sex selection, amounting to aboutthreepercentoffemalebirths.Thedegree(numberoffemalebirthsmissedaspercentoffemalebirthsoccurred) is high in most states in the northernand western regions, moderate in Uttar Pradesh,HimachalPradeshandMadhyaPradesh,andlowornegligibleinmoststatesintheeasternandsouthernregions.

Atthe2011censusenumeration,aboutfourmilliongirls of ages 0-6 may be considered to have beenmissing;2.5milliononaccountofsexselection(pre-nataldiscrimination)and1.5millionduetoexcessfemale mortality (post-natal discrimination). Thissituationhaspersistedbeyond2011aswell.Further,while pre-natal discrimination is concentratedin the northern and western regions, post-natal- discrimination is common across the country;

Executive Summary

xii

thesouthernregionanda fewotherstates showrelatively low levels but the regionaldifferencesinpost-nataldiscriminationarenotaswideasinpre-nataldiscrimination.

At higher birth orders and among those whohave no son, the SRB is very highly masculinein the northern, western, and central regions.Sex selection at the third birth following twodaughters seems to be verywidely prevalent. Inthenorthernregion,theSRBatthefirstorderisalsomoremasculine thannatural implying thatthereissomesexselectionatthefirstbirthitselfindicating that some couples desire to avoid thebirthofevenonedaughter.

Some differences in the SRB by socioeconomicbackground are seen especially at the secondand thirdbirths.For the secondbirthafterfirstdaughter, the SRB is generally more masculinethanaverageinthehighesteducationandwealthclasses.Atthethirdbirthfollowingtwodaughters,theSRBishighlymasculine;thisismoresointhemost recent period of 2010-14. Further, the SRBis highly masculine at the highest wealth andeducation levels, in the northern and westernregions.Highlymasculine SRB is also associatedwithhighmediaexposure.

Evidence on perceived values of sons vis-à-visdaughtersshowsthatsonsarevaluedforoldagesupport, financial as well as for residence; suchreliance is relatively higher in the northernand western regions compared to other regions.Though some changes in attitudes are seen in

recentinvestigations,thesearenotlargeenoughand parents by and large continue to expectsuch support primarily from sons rather thanfrom daughters. Besides, in spite of the legalentitlementsandprovisions,itisnotcommonfordaughterstoinheritparentalproperty.

The analysis shows that in spite of effortsmadeby enactment of laws and campaigns by thegovernment and civil society organisations, sexselection has continued. Though some changehasbeenseeninPunjab,Haryana,andHimachalPradesh, the SRB is yet to return to thenaturallevel in these states. Besides, in recent years,the SRB in some states outside the northern-westernregionhasalsobecomemoremasculine.Given that sonpreference iswidelyprevalent inIndia,thereisapossibilityofthepracticeofsexselectionspreadingtoareaswhichhavehithertonotshownitonalargescale,oncetheavailabilityofsonographicscanfacilitiesandaffordabilityoftheservicesrise.

Itmustalsoberecognisedthatalargenumberofgirlsare‘missing’duetopost-nataldiscrimination,reflected in higher childhood mortality amongfemales than amongmales.While thematter ofgender biased sex selection has been receivingmediaandpolicyattentioninIndia,andrightlyso,post-nataldiscriminationrarelyfiguresinpublicdiscussions.Itisimperativethatcivilsocietyandpolicymakersaccorddueattentiontothisconcernaswellandadoptappropriatemeasurestoaddressit.

1

ThesexratioofIndia’spopulationhasbeeninfavourofmalesincontrastto the situation inmost of theworldwherewomenoutnumbermen inthe population. In his seminalwork on India’s sex ratio, Visaria (1968)

examinedthedatauptothecensusof1961andidentifiedhighermortalityamongfemalesascomparedtomalesastheprincipalfactorresponsiblefortheratiotobeinfavourofmalesinIndia’spopulation.Ontheotherhand,forthecountryasawhole,thesexratioatbirthwasnearthenaturallevel;theratioisusuallycloseto105malebirthsper100femalebirths,generallyintherange104to106,oraround952femalebirthsper1000malebirths,intheabsenceofanydistortion.However,datasincethe1990shaverevealedariseinmasculinityinIndia’ssexratioatbirth (SRB)and this issuehasbeenexamined inanumberof studies(Premi,2001;Bhat,2002;Arnoldetal.,2002;BhatandZavier,2007;GuilmotoandAttene,2007;Guilmoto,2008,2009).Moreover,ithasemergedthatgenderbiasedsex selection has been practised especially since the 1990s on a large scale inmanypartsofthecountrycausingtheSRBtobecomemoremasculine(Arnoldetal.,2002;Jhaetal.,2006;Kulkarni,2007;Visaria,2007;Kulkarni,2012;BongaartsandGuilmoto,2015).Thisoutcomestemsfromstrongsonpreferencecombinedwithdecliningfertilityandenabledbytheavailabilityofandaccesstopre-natalsexdetectiontechnologies,especiallytheuseofsonographicscans.

Inpopulationswithsonpreference,stoppingstrategies(havechildrenuntilthedesirednumberofsonsisbornandthenstopchildbearing)areoftenadoptedbutitiswellrecognisedthatsuchstrategiesdonotaltertheSRB(Goodman,1961).However, sex selection through sex detection and gender biased sex selectiondoesinfluencetheSRB.Suchaplanwouldbeadoptedbycoupleswhowanttohaveacertainnumberofsons(oracertainsexcompositionofchildren)butatthesametimelimitthetotalnumberofchildren.Thismayalsobedoneincaseofaversiontochildrenofaparticularsex(‘daughteravoidance’hasoftenbeenmentionedinliterature).Inthepast,technologyforpre-natalsexdetectionwasnotavailableandhenceresortingtogenderbiasedsexselectionwasnotanissue.FemaleinfanticidewaspracticedtosomeextentinafewpopulationsandthereisevidenceofthisforpartsofIndia(Visaria,1968;Georgeetal.,1992),andwhilethispracticeaffectedthechildsexratio itdidnotaffect theSRBpersesinceinfanticideoccurspost-birth.However,sincethe1980s,pre-natalsexdetectionhasbecomeeasilyaccessibleinmanypartsoftheworldandwithadvancementin technology, pre-natal sex selection (also referred to as gender biased sexselection)hasengineeredariseinmasculinityatbirth.

India is not unique to this phenomenon; South Korea, China, Vietnam, andcountriesaroundtheCaucasushavealsoseenthispracticeonafairlylargescalethough this has been phased out in SouthKorea (Guilmoto andAttane, 2007;Guilmoto,2015;Guilmotoetal.2018).There isalsoevidenceofsuchapracticeamongpersons of Indianorigin in theUnitedKingdom (DubucandColeman,2007).Inordertoeliminategenderbiasedsexselection,Indiahasenactedlawstobanpre-natalsexdetection;thePre-NatalDiagnosticTechniques(Regulation

Introduction 1

The sex ratio at birth was near the natural level; the ratio is usually close to 105 male births per 100 female births, generally in the range 104 to 106, or around 952 female births per 1000 male births, in the absence of any distortion. However, data since the 1990s have revealed a rise in masculinity in India’s sex ratio at birth (SRB)

2

and Prevention ofMisuse) Act, 1994 (PNDT Act), thatwas amended in 2003 asthe Pre-Conception and Pre-Natal Diagnostic Techniques (Prohibition of SexSelection)Act(PCPNDTAct).

Availableevidenceshowsthat,inspiteoftheAct,thepracticeofgenderbiasedsexselectionpersistsinIndia.Sincethepracticeisillegal,itisdifficulttohaveanydocumenteddataonthenumbersofsuchcases.However,anestimateofthenumberofinstancesofgenderbiasedsexselection,thatis,thenumberofmissingfemalebirths, maybedrawn indirectly fromtheSRB.ThoughdataonSRB inIndiaareavailable fromvarious sources, theestimatesobtained fromdifferentsourcesdonotalwaysagreewiththeresultthattherearediverseinferencesonthelevelsandchangesintheratio.Therefore,thisstudyfirstliststhesourcesofdataontheSRBandpresentsestimatesfromthesesince1991;forestimatesfromvarioussourcesforearlierperiodsseeKulkarni(2007;2009).

ThestudythendiscussestheacceptabilityofvariousestimatesoftheSRBtoarriveatthemostplausibleestimates.ThesevaluesoftheSRBhavethenbeenused,inconjunctionwithestimatesofthenumbersofbirths,toestimatethenumbersofmissingfemalebirths..

The practice of gender biased sex selection, resulting from ‘pre-nataldiscrimination’againstfemales,isonlyonefactorcausingfemaledeficit.Higherthanexpectedfemalechildhoodmortality,duetoneglectofthegirlchild,called‘post-nataldiscrimination’,istheotherfactorthatcausesfemaledeficit;infact,thepioneeringworkonestimationofmissingwomenbySen(1990)andCoale(1991)wasinthecontextofthisfactor.Hence,thestudyestimatesthenumberofmissinggirlsat the2011censusenumerationandpresents itsdecompositionbythetwofactors, pre-natal discrimination and post-natal discrimination. This has beendoneforgirlsofages0-6,sincethechildsexratioforthisagegroupiscommonlyused in India indiscussionson femaledeficit.Further, for thefive-yearperiodbeyondthe2011census,estimatesofmissingfemalebirthsandexcessdeathsofgirlsbelowagefivehavebeenobtained.

ThestudyalsoexamineshowtheSRBvariesbythestageoffamilybuilding,thatis,atdifferentbirthordersandbythesexcompositionofpreviousbirthsuptothethirdorder.Thisaidsanunderstandingofthelocationofsexselectionwithinthereproductivespan.Thisisfollowedbyanexaminationofsocioeconomicandspatialdifferentialsinthesexratioatbirthatvariousstagesoffamilybuildingandanassessmentofthenetinfluencesofvariousfactorsontheprobabilityofamalebirthatordersuptothethird.Finally,thestudylooksatrecentevidenceonreasonsforsonpreferenceand,inparticular,onvaluesofsonsvis-à-visdaughters.

Sex selection, in addition to being intrinsically undesirable, also has adverseimplicationsforthesocietycausingseximbalancesinthesociety.TheimpactofsuchseximbalanceonmarriagesqueezeinIndiahasbeenexaminedinpaperbyKaur(2004),Guilmoto(2012),andKauretal.(2016)andisnotthesubjectmatterofthepresentstudy.

In international convention, the SRB is expressed in terms ofnumber ofmalebirthsper100femalebirths.However,inIndiaSRBistraditionallypresentedasnumberoffemalebirthsper1000malebirthsandhenceratiosinthisconventionarepresentedinthetablesinthisstudy.Inthisconvention,alowerSRBmeanshighermasculinityatbirth.TheconversionfromSRBinoneconventiontoanotherisstraightforward:

SRB (malebirthsper 100 femalebirths)= 100000/SRB (femalebirthsper 1000malebirths).

The study estimates the number of missing girls at the 2011 census enumeration and presents its decomposition by the two factors, pre-natal discrimination and post-natal discrimination

3

Estimates of Sex Ratio at Birth from Different Sources of Data

A number of independent data sets allow estimation of the sex ratio atbirth. These include the Civil Registration System (CRS), the SampleRegistration System (SRS), the decennial censuses, and the National

FamilyHealthSurveys(NFHS).Besides,someothersurveys,andadministrativerecordsincludingtheHealthManagementInformationSystem(HMIS)alsogiverelevantdata.

2.1 Civil Registration SystemTheCRShasbeeninoperationinIndiaforalongtime.Initially,registrationofbirthsanddeathswasvoluntarybutaftertheenactmentoftheRegistrationofBirthsandDeathsActin1969itismandatorytoregisterallbirthsanddeaths.TheORGIpublishesannualreportsthatprovidedataonregistrationandthesegive SRB for India and states andunion territories.However, for someyears,tabulationsofbirthsbysexarenotavailableforafewstates.Thoughthelevelofcompletenessofcoverageofcivilregistrationofbirthshasshownanimpressiverisefrom56percentin2000to86-88percentduring2014-2016,thecoverageisfarfromcomplete(RegistrarGeneral,2019a).Sinceregistrationcanbesexselective,theestimatesofSRBarelikelytobebiased(newbornboysaremorelikelytoberegisteredthangirls,asnotedbyVisaria,1968)andthisisalimitationthatneedstobenoted.TrendsintheSRBaccordingtotheCRSsince1991areshowninTable1.Atthenationallevel,theSRBhasfluctuatedbetween857and909;thelowestvaluewasseenintheyear2010,however,thisappearstobeanoutlierasthisismuchdifferentfromthevaluesfor2009(898)and2011(909).Overall,theSRBwashighlymasculinethroughouttheperiod;thelevelwassomewhathigherduring2007to2013(excepttheyear2010whenitwasunusuallylow)buttheredoesnotseemtobeanydiscerniblelong-termtrend.

2.2 Sample Registration System (SRS)

TheSampleRegistrationSystem(SRS)wasintroducedinIndiain1964-65onapilotbasisandin1969onaregularbasissincetheCivilRegistrationSystemdidnothaveagoodcoverageatthetime.TheSRShasbeenprovidingestimatesoffertility andmortality on a regular basis since 1970 (Registrar General, 2019).The SRS is a dual record systemwith continuous registration andhalf-yearly

2

Since registration can be sex selective, the estimates of SRB are likely to be biased

4

retrospectivesurveysthatarematchedandcorrected.In2017,thesampleoftheSRScovered8850units(4961ruraland3889urban)whichhadatotalpopulationof7.9million.FortheSRB,estimatesforIndiaandlargestatesareavailableasthree-yearmovingaveragessince1998-2000(someestimatespriortothisperiodareavailableforIndiaasawhole).SincetheSRS,asthenameshows,isbasedon registration in a sample of geographic units, sampling errors can be largeand the organisation gives three-year averages of the SRB rather than singleyearestimates.Theestimatesforthethree-years 1990-92totheyears2015-2017arepresentedinTable1.AccordingtotheSRSestimates,theSRBhasbeenquitelow,below910,wellbelowthenaturallevel,throughouttheperiod.TherewasanapparentriseinthemasculinityintheSRBduringthefirstfewyearsofthecenturywiththeSRBfallingto880andamildrecoveryseenafter2003-2005butasmalldeclineafter2012-14.SincetheSRSchangedthesampleunitsin2004andagainin2014(theSRSnormallychangesthesampleunitsevery10years,usingthelatestcensusastheframe),someoftheseturnaroundsmaybeattributabletothesechanges.Notwithstandingthepossibleeffectofchangesinthesample,theSRSestimatesshowahighlymasculineSRBinrecentdecades.

2.3 Census

TheIndiancensusisdecennialandhasbeenorganizedregularlyforoveracentury.Since1981,thecensushasquestionson‘birthslastyear’tomarriedwomenandon‘thenumberofchildreneverborn’aswellas‘thenumberofsurvivingchildren’toevermarriedwomen.Thisinformationistabulatedbysexofchildrenandageofwomenwhichallowsustocomputethesexratioofbirthslastyearandofallbirthstowomen.TheSRBbasedonbirthslastyear(BLY)referstotheone–yearperiodbeforethecensusenumeration.OnecannotspecifyareferenceperiodfortheSRBbasedonallbirths(childreneverborn-CEB)sincethesebirthswouldhaveoccurredoveralongperiod;fortheyoungerwomenthebirthswouldberecentbutforolderwomen,manyofthesebirthswouldhaveoccurredsometimeago.HencewecomputetheSRBbasedonbirthstowomenintheagerange20-29,labeledCEB(20-29),asalmostallofthesebirthswouldhaveoccurredduringaperiodof 10-15yearsbeforethecensusandmostduringthe10-yearperiodbeforethecensus.Thoughno specific ‘referenceperiod’ as such canbe given for this SRB, itdoesrefertoarecentperiod.TheBLYandCEB(20-29)estimatesbasedon2001and2011censusesaregiveninthelastcolumnofTable1.

Further,thecensusesprovideage-sexdistribution.Inpublicdiscoursesandthemedia,thechild-sexratiohasoftenbeenusedtocommentonthelevelofSRBandby implicationon sex selection.The child sex ratio for the age group0-6iscommonlyusedinIndiabecausethisratioisavailablesoonafterthecensuswhereastabulationsoncompleteage-sexdistributiontakelongertime.Strictlyspeaking,childsexratioisnotidenticaltoSRBsincechildsexratioisinfluencedbysexdifferentialsinearlychildhoodmortalityinadditiontotheSRB.Butifinformationonchildmortalitybysexisavailable,onecanestimatetheSRBfromthechildsexratioindirectlybyapplyingthetechniqueofreversesurvival.Forthispurpose,agegroupssuchas0-4or5-9maybeusedinordertoestimaterecentSRB.However,sexselectivemisreportingofageandsexselectiveomissioncandistort this ratio. It has been seen that such an effect isminimal in the agerange0-6(Bhat,2002)andhenceitispreferabletoestimatetheSRBbasedonthechildsexratiofortheages0-6.KumarandSathyanarayana(2012)haveobtainedsuch estimates from the 2001 and 2011 census data. These refer to the seven-

The study estimates the number of missing girls at the 2011 census enumeration and presents its decomposition by the two factors, pre-natal discrimination and post-natal discrimination

5

Table 1:Table 1: Estimates of sex ratio at birth from various sources, India, 1991-2017(femalebirthsper1000malebirths)

Year/mid-year of period

Source

CRS SRS @ NFHS-3Annual MA $

NFHS-4Annual MA $

HMIS Census based

1991 865 900 870 947 859 850

1992 863 894 957 925 865 858

1993 863 885 937 915 840 868

1994 862 879 913 933 894 876 Based on

1995 870 883 897 928 877 882 2001 Census

1996 869 891 969 926 897 894 CEB20-29 939

1997 881 901 929 930 895 901

1998 883 928 938 907 899 Indirect 935

1999 895 898 933 928 926 905

2000 886 894 936 933 871 919 BLY 906

2001 875 892 914 927 929 921

2002 872 883 956 919 963 917

2003 868 882 894 917 925

2004 872 880 894 908 923 Based on

2005 876 892 907 916 2011 Census

2006 891 901 920 916 CEB20-29 928

2007 903 904 928 919

2008 904 906 920 912 900 Indirect 923

2009 898 905 919 909 927

2010 857 906 875 909 913 BLY 899

2011 909 908 902 913 917

2012 908 909 931 911 915

2013 898 906 942 918

2014 887 900 905 918

2015 881 898 923

2016 877 896 926

2017 929

@:Three-yearmovingaverage;$:MA:Five-yearmovingaverage.CEB(20-29):Childreneverborntowomenofages20-29atthecensus.BLY:Birthslastyear.Indirect:Indirectestimatecomputedbyapplyingreversesurvivaltochildsexratio(ages0-6).Sources:CRS(CivilRegistrationSystem):RegistrarGeneral(2013a,2018a);SRS(SampleRegistrationSystem):RegistrarGeneral(variousyears,2001-2018);NFHS-3and4(NationalFamilyHealthSurvey-3,-4):ComputedfromNFHS-3,-4datafiles;HMIS(HealthManagementInformationSystem):HMIS(2018);Censusbased:CEBandBLYestimatescomputedfrom2001and2011Censusfertilitytables;Censusbased:IndirectestimatesfromKumarandSathyanarayana(2012).

6

year periods before the census, that is,March 1994-February 2001 andMarch2004-February2011orroughly1994-2000and2004-2010basedonthe2001and2011censusesrespectively.Theseestimates,labeledindirectestimates,arealsoshownin Table 1.

ThethreeestimatesoftheSRB,fromBLY,CEB(20-29),andindirectfromchildsexratio,dovarysomewhat.TheratiobasedonBLYislowerthanthatbasedonCEB(20-29).TheestimatebasedonBLYreferstothelastyearbeforethecensuswhereasthatbasedonCEB(20-29)referstoalongerperiodbeforethecensusandhighervaluesforthelattermayindicatethattheSRBhasbeenfallingovertheyearspriortothecensus.However, it isseenthatwhilethenumberofbirthsreportedasBLY in the2001 censuswas 19.9million, theestimatednumberofbirthsbyapplyingtheSRSbirthratetothethenpopulationwouldbecloseto26.3million. Similarfigures for the 2011 census are: reported births last year20.9millionandestimatedbirths26.5million.Thus,alargenumberofbirthsthatoccurredinthepreviousyearwerenotreportedinthecensusesasbirthslastyear.Inviewofthis,aninferenceontrendsisnotwarrantedmerelyfromacomparisonoftheestimates.Incidentally,theindirectestimatesfromthecensuschildsexratiosareclosetothosebasedonCEB(20-29)forboththecensuses.

2.4 National Family Health Surveys (NFHS)

TheNationalFamilyHealthSurvey(NFHS)istheDemographicandHealthSurvey(DHS)forIndiaandfourroundsoftheNHFShavebeencarriedout inIndiasofar;thelatestround(NFHS-4)wasin2015-16andtheprevious(NFHS-3)in2005-06.ThesurveyshavealargesamplesizeandthiswasparticularlysoinNFHS-4;6,01,509householdsand6,99,686womenofages15-49wereinterviewedinNFHS-4and1,09,041householdsand1,24,385womenofages15-49interviewedinNFHS-3(InternationalInstituteforPopulationSciences(IIPS)andMacroInternational,2007;InternationalInstituteforPopulationSciences (IIPS)andICF,2017).SincetheNFHSobtainedcompletefertilityhistories,itispossibletocomputeSRBforvarioustimeperiods;theNFHSdatafilesalsoallowanexaminationofSRBbythesexcompositionofpreviousbirthsinordertoseehowtheSRBis influencedbysexpreference,especiallysonpreference,anissuethatwillbeaddressedinalatersection.EstimatesofSRBforIndiaforcalendaryearsupto2004forNFHS-3andupto2014forNFHS-4,thelastcompleteyearscoveredinthetwosurveysrespectively,areobtainedfromtheNFHS-3andNFHS-4datafiles;sampleweightsprovidedinNFHSdatafileshavebeenapplied.

Theestimatesarepresented inTable 1and inFig.1.Therewasagapof 10yearsbetween the NFHS-3 and the NFHS-4 and hence for the years prior to 2005,estimatesareavailablefromboththesurveys.ItisseenthattheNFHS-4estimatesfor theyears2000to2002areslightly lower (moremasculine) thanthosefromthe NFHS-3, and for earlier years, the NFHS-4 showsmuch lower values. It islikelythatforchildrenbornlongago,morethan10yearsbeforethesurvey,thereisselectiveomissionofgirlsinreporting.Oneplausiblereasonisthat,theNFHSbeingaretrospectiveenquiry,birthsofdaughterswhoweremarriedbythesurveydatemaynothavebeenreportedbysomewomeninthesurvey.Regardlessofthereasonsforsuchomission,itisnotadvisabletodrawinferencesbasedontheNFHSestimatesreferringtoperiodsmuchbeyond10yearsbeforethedateofthesurvey.Further,sincetheNFHSaresamplesurveys,theestimateshavesamplingerrorsasaresultofwhichfluctuationsareseenintheannualseries(Fig.1).Hence,five-yearaverageshavealsobeencomputedandpresentedinFig.1andinTable1.

The NFHS data files allow an examination of SRB by the sex composition of previous births in order to see how the SRB is influenced by sex preference, especially son preference

7

Fig.1:Fig.1: Trends in SRB based on NFHS-3 and NFHS-4, India, annual estimates and five-year moving averages (female births per 1000 male births)

975

950

925

900

875

850

825

8001991 1996 2001 2006 2011 2016

1000

NFHS-4-MA NFHS-4 NFHS-3-MA NFHS-3

Note:MA:Five-yearMovingAverage;thesevaluesareshownagainstthemid-yearinthegraph.Source:Table1.

From theNFHS-3 series, it is seen that the SRB (five-year average)was fairlystablefrom1992to2002(thelastmiddleyearforwhichfive-yearaveragecouldbecomputedfromtheNFHS-3),fluctuatingmildlyaround925femalebirthsper1000malebirths.TheNFHS-4seriessince2000showsaslightlylowerSRB,fluctuatingaround915upto2012(thelastmiddleyearforwhichfive-yearaveragecouldbecomputedfromtheNFHS-4).Thus,thetwosurveysshowlowerthannaturalSRBinIndiafluctuatingintherange910to935femalebirthsper1000malebirthsovertheperiod1992to2012(ignoringtheNFHS-4estimatesforyearsbefore2000forreasonsnotedabove).

TheIndiaHumanDevelopmentSurvey(IHDS)isanotherseriesoflargesurveyswhichalso collectdataon fertility; this surveywasconductedduring2004-05andagain in 2011-12 (IHDS, 2018). Thefirst roundof the IHDS (IHDS-1)nearlycoincideswiththeNFHS-3andthereportshowedthatatthenationallevel52percentofbirthswereboys(basedonbirthsduringthe10-yearperiodbeforethesurvey).ThisisequivalenttoanSRBof923femalebirthsper1000malebirthswhichisclosetotheNFHS-3estimatefortheperiodwhichis930for1995-99and919for2000-2004.

2.5 Health Management Information System (HMIS)

The HMIS is a relatively new system introduced by the Ministry of HealthandFamilyWelfareoftheGovernmentofIndiaaspartoftheNationalHealthMissiontocollectinformationonvariousaspectsofhealthservicesinordertomonitor the performance of the programme (HMIS, 2018). The information isprovidedbyhealthworkersand institutionsanddataat thedistrict levelareuploadedtotheHealthStatisticsInformationPortalusingawebbasedHealthManagement InformationSystem(HMIS) interface.Periodic reportsaremadeavailablebythesystem.Theinformationpertainstoservicesbyboththepublicandtheprivatesectors.TheHMISreportssince2008-09(monthly,quarterly,andannual,atnational,state,anddistrictlevel)areavailableonthewebsiteandthesexratioatbirthhasbeenincludedinthereports.ThevaluesofSRBfromthissystemareshowninTable1.

Two rounds of National Family Health Surveys (NFHS) show lower than natural SRB in India fluctuating in the range 910 to 935 female births per 1000 male births over the period 1992 to 2012

8

Exceptforthefirstyearofinformation,2008-09,theSRBforIndiahasbeenintherange913and929.Thus,itismoderatelybelowthenaturallevel.

Itshouldbenotedthatthesystemisgettingestablishedandthecoverageisnotcomplete;thesystemalsogivesthelevelofcoverageofbirthsandthishasbeenhoveringaround75-80percentinrecentyears(HMIS,2018).Thereisapossibilitythatreportingofbirthshassomesexselectivity,andthiscouldbiastheestimatesto that extent.

2.6 Regional variations

The various sources provide estimates for states andunion territories aswellthough the SRS estimates are only for large states; estimates for states/unionterritoriesaregiveninAppendixTables.Geographicvariations intheSRBarevery conspicuousand seenacrossall the sources though theprecisevaluesdovary. States in the northern-western region show very low ratios, notablyPunjab, Haryana, Chandigarh, Delhi, Rajasthan, Himachal Pradesh, Gujarat,Maharashtra,andUttarakhand.Insomeofthesestates,theratioseemstohavebecome less masculine over the period; in particular, Himachal Pradesh andPunjabshowadistinctimprovementthoughtheratioinPunjabcontinuestobemoremasculinethanthenaturallevel.Ontheotherhand,inallthesouthernstates, inWest Bengal, Odisha, Chhattisgarh, and in the northeastern states,the SRB is generally close to the natural level. A few of these did showhighmasculinitybasedontheCRSdatainafewyears,butotherwisehaveSRBwithintherange935to970withnocleartrend.

Kumar and Sathyanarayana (2012) provide district level estimates of the SRBbasedonthecensus0-6sexratioviareversesurvivalandthesearedepictedinamap(Map2intheirpaper).TheseallowonetoseeamoredisaggregatedspatialpatternoftheSRBthanwhatisseenfromestimatesatthestatelevel.

States in the northern-western region show very low ratios, notably Punjab, Haryana, Chandigarh, Delhi, Rajasthan, Himachal Pradesh, Gujarat, Maharashtra, and Uttarakhand

9

AsseeninTable1,theestimatesofSRBforIndiaasobtainedfromvarioussourcesdonotalwaysagree.Thiscallsforacomparativeassessmentoftheestimatesfromtheavailablesources.TheSRSestimatesareforthree-

yearperiodsandhencesimilarthree-yearaveragesareobtainedfromtheannualestimates from the CRS. For comparing NFHS estimates, only the estimatesfortherecentperiodsareusedsince,asnotedearlier, theNFHSestimatesforperiods10yearsbeforethesurveyarehighlymasculinebecauseofalikelihoodofomissionofdaughtersbornwellbeforethesurveyintheretrospectiveenquiry.Moreover,five-yearaveragesof theNFHSareused since theannual estimateshavelargesamplingerrors.Fig.2presentsacomparativeviewoftheestimatesofSRBforvariousyearsandtimeperiods.Itisseenthatthethree-yearaveragesfromtheCRSarelowerthanthecorrespondingSRSestimatesbyabout10points.AlowervaluefromtheCRSshouldnotcomeasasurprisesinceregistrationofsonsismorelikelythanthatofdaughters.Further,theSRSestimatesarelowerthanthosefromtheNFHS-3andNFHS-4thoughtheveryrecentSRSestimatesarequiteclosetotheNFHSonesforthecorrespondingperiods.TherecentHMISestimatesarealsoclosetotheNFHSestimates.

Ofthecensusbasedestimates,thosefromBLY(birthslastyear)areclosetotheSRSestimates for the correspondingyears.However, theestimates impliedbythechildsexratios,obtainedindirectlybyreversesurvival,arehigherthantheSRSestimatesforthecorrespondingperiods(acomparativepicturecanbeseenfromthelowerpanelinTable2).Theestimatebasedonthesexratioforages0-6inthe2001censusreferstotheperiod1994-2000whichis935whereastheSRSestimateforroughlythesameperiodis895.Theestimatefromthe2011censusreferringto2004-10is923whereastheaverageofSRSestimatesforroughlythesameperiodis9031.TheCEB(20-29)estimatesfromthebirthstowomenofages20-29areslightlyhigherthantheonesimpliedSRB,byaboutfivepoints.But,asthesedonothaveaclearlydefinedreferenceperiod,itisnotpossibletocomparethesewithothers.Inadditiontothesourcesnotedintheprevioussection,theUnitedNations PopulationDivision also provides populationdata base for allcountries in the reports onworldpopulationprospects and thevalues of SRBforIndiaforfive-yearperiodsaccordingtothelatestpublication(U.N.,2019)are1.106,1.111,1.112,1.101,1.099,and1.099(expressedasratiosofmalebirthsto100femalebirths)fortheperiods1990-95,1995-2000,2000-2005,2005-2010,2010-2015,

1 Since the SRS does not give annual figures but only three-year averages, and there are some gaps in the SRS series of SRB estimates, it is not possible to get the averages precisely for the seven-year periods 1994 to 2000 and 2004 to 2010 and hence the average of the estimate for 1995-97, centered on 1996 and of 1998-2000 centered on 1999, is used for comparison with the 2001 census based estimate for 1994-2000 and the average of 2005-07 and 2008-10 estimates is used for comparison with the 2011 census based estimate for 2004-10.

Correspondence among Various Estimates

3

The estimate based on the sex ratio for ages 0-6 in the 2001 census refers to the period 1994-2000 which is 935 whereas the SRS estimate for roughly the same period is 895

10

2015-2020respectively;intermsoffemalebirthsper1000malebirths,theseare:904,900,899,908,910,and910.TheseareclosetotheSRSestimatesthoughnotidentical;theU.N.estimatefor2000-05ismuchhigherthanthatbytheSRS.

Fig.2:Fig.2: Comparison of estimates of SRB from various sources, India(femalebirthsper1000malebirths)

Source:Table1;SRS-correctedseriesbasedoncorrectionof0.978toSRSestimates

940

920

900

880

860

840

1991 1996

CRS - 3 year avg

NFHS-4 5-year avg HMIS SRS-corrected

SRS 3 year avg NFHS-3 5-year avg

2001 2006 2011 2016

960

Table 2:Table 2: Comparison of estimates of SRB from various sources for the periods 1994-2000 and 2004-10, India(femalebirthsper1000malebirths)

Period CRS SRS NFHS Indirect from Census Child sex ratio, ages 0-6NFHS-3 NFHS-4

1994-2000EstimateRatioto2001CensusIndirectestimate

878

0.939

895

0.957

929

0.994

935(2001census)

2004-10EstimateRatioto2011censusindirectestimate

886

0.960

903

0.978

911

0.987

923(2011census)

Sourceforestimates:Table1.

Note:theCRS,SRSandNFHSestimatesareaveragesforthespecifiedperiods.

Thus,weseethattheestimatesdodiffer;theCRSestimatesandthecensusbasedestimatesfromBLYarelowerthantheSRSestimateswhich,inturn,arelowerthantheNFHSfive-yearaverages,andthecensuschildsexratiobasedimpliedestimates2. Ifregistrationofbirthsissexselective,withthelikelihoodofregistrationbeinghigherformales,theCRSandSRSwouldunderestimatetheSRB(asmeasuredintermsoffemalebirthsper1000malebirths).Butitisalsoconceivablethatcensusenumerationfavoursfemalesandthisoverestimatesthechildsexratio(femalesper1000males)inthecensusandconsequentlytheSRBcomputedfromit.However,resultsfromthepostenumerationsurvey(PES)ofthe2011censusshowthatthereishardlyanysex-selectiveomissioninearlyages(RegistrarGeneral,2014a).Thenetomissionratesare32.57per1000formalesand32.57forfemalesfortheagegroup0-4and22.92per1000formalesand22.17forfemalesforthe5-9agegroup.

2 For a recent work on comparison of estimates of SRB, see Rajan et al. (2017).

11

Theslightdifferenceinthenetomissionratesinthe5-9agegroupdoesnotimpactthechildsexratiofor0-6agesbyevenonepoint.Thus,theSRBimpliedbythechildsexratiodoesnotappeartohavebeenaffectedbysexselectiveunder-enumeration.

Moreover,asnotedearlier,sexselectiveagemisreportinghasverylittleeffectonthe0-6agegrouptakenasawhole.Further,migrationatveryyoungagesisgenerallynot sex selectiveandhencewouldnot influence thechild sexratio.The computationof SRB from the child sex ratiodoes involve assumptions ofsexdifferentialsinearlychildhoodmortality.However,minordeparturesfromsuchassumptionsdonotaffecttheestimatenotably.Thechildsexratiofromthe2011censusis919femalesper1000malesandtheSRBwouldnotbelowerthanthisvalueunlessfemaleearlychildhoodmortalityislowerthanmalemortalitywhereasallevidencepointstowardshigherfemalethanmalemortalityduringearly childhood in India.According to the SRS report for 2011, theunder-fivemortalityratewas59per thousandforfemalesand51per thousandformales(under-fivemortalityratesbysexareavailableinrecentreportsoftheSRS;seeRegistrar General, various years). Given the evidence on sex differentials inchildmortality(whichshowshigherchildhoodmortalityamongfemalesthanmales)andonomissionrates(whichdonotvarybysex)incensusenumeration,estimatesofSRBshowinglowermasculinitythanthecensuschildsexratiosarenot tenable.

Inviewofthis,theindirectestimatesobtainedfromthechildsexratiosmaybeacceptedasthemostplausible.However,theseareavailableonlyonceintenyearsanddonotgiveatimeseriestoassesschangesovershortperiods.Ontheotherhand,theSRSgiveacontinuousseriesbuttheseareseentobeunderestimatesofthefemalebirthstomalebirthsratio.OnecouldthencorrecttheSRSestimatesonthebasisoftheratioofSRSestimatetotheindirectcensusbasedestimateforthesameperiod.Theseratios,showninTable2,are0.957basedon2001censusand0.978basedon2011census.Fortherecentperiod,thecorrectionof0.978asobtained from the 2011 censusmaybe applied. Thus the SRS estimatemaybecorrectedbydividingitby0.978incaseoffemalebirthstomalebirthsratio;asaroundfigure,thecorrectionamountstoraisingtheSRSestimateby2percent.The estimates so corrected from 2000 onwards are shown in Fig.2 alongwiththedirectestimates.Afterapplyingthecorrection,India’sSRBduring2000-2016fluctuatesintherange899to929(roundedto900to930)femalebirthsper1000malebirthswithnocleartrend.

Inarecentanalysis,Kauretal.(2016)alsonotedamismatchbetweentheSRSestimateandindirectcensusbasedestimatesof theSRBandadjustedtheSRSestimates on the basis of the relationship between SRB and child sex ratiothroughchildmortalityrates(fordetailsseeAppendix–Iofthepapercited).Fortherecentyears,theadjustedvaluesofSRBarehigherthantheSRSestimatesby2-3percent(femalesper1000males)attheall-Indialevel,closetothecorrectionnoted above. For periods before 2000, the correction is higher, by about fourpercent,whichis inlinewiththecorrectionfactorof0.957basedonthe2001censusshowninTable2.

The SRS give a continuous series but these are seen to be underestimates of the female births to male births ratio. One could then correct the SRS estimates on the basis of the ratio of SRS estimate to the indirect census based estimate for the same period

After applying the correction, India’s SRB during 2000-2016 fluctuates in the range 899 to 929 female births per 1000 male births with no clear trend

12

Table 3:Table 3: Comparison of estimates of SRB from various sources for the period 2004-10, India and large states

India/state

Estimates of SRB for the period 2004-10 from various sources

(female births per 1000 male births)

Ratios to indirect census based estimate

CRS2004-2010Average

SRS2005/07&2008/10average

NFHS-42004-2010

Indirectestimatebasedon2011census

CRSest./Censusindirect

SRSest./Censusindirect

NFHS-4est./CensusIndirect

India 886 903 911 923 0.960 0.978 0.987

AndhraPradeshβ 980 917 915 938 1.045 0.978 0.975

Assam 888 933 959 961 0.924 0.971 0.998

Bihar $ 910 930 943 $ 0.966 0.986

Chhattisgarh 916 977 951 971 0.943 1.006 0.979

Delhi 874 877 787 873 1.001 1.005 0.901

Gujarat 885 897 910 895 0.989 1.002 1.016

Haryana 840 845 818 845 0.994 1.001 0.969

HimachalPradesh 894 936 965 916 0.976 1.022 1.053

Jammu&Kashmirα 918 863 933 866 1.060 0.997 1.077

Jharkhand 854 923 946 958 0.891 0.963 0.987

Karnataka 973 934 934 949 1.026 0.985 0.984

Kerala 947 962 961 965 0.981 0.997 0.996

MadhyaPradesh 892 917 921 923 0.966 0.993 0.998

Maharashtra 850 883 856 896 0.949 0.985 0.955

Orissa 925 935 938 942 0.982 0.993 0.996

Punjab 812 834 837 854 0.950 0.977 0.980

Rajasthan 824 871 877 897 0.918 0.971 0.978

TamilNadu 934 935 928 942 0.992 0.993 0.985

UttarPradesh $ 875 909 914 $ 0.958 0.995

WestBengal 913 937 938 954 0.957 0.982 0.984

Sourceforestimates:Appendixtables1-5

β :IncludingTelangana;α;IncludingLadakh;$:Dataforsomeyearsarenotavailable.

A comparison of estimates for large states is shown in Table 3 for the period2004-10.Itisseenthatinamajorityofstates,theratioofSRSestimatetocensuschild sex ratio based estimate is close to 0.98, that is, the SRS estimates arelower(showinggreatermasculinity)thanthechildsexratiobasedestimatesbyabout twopercent. The SRS estimates are lower bymore than twopercent inUttarPradesh,Bihar,Jharkhand,Rajasthan,andAssam.Ontheotherhand,inafewstates,theSRSestimatesareveryclosetothecensusbasedestimateswiththedifferencebeinglessthanonepercent.Thus,theSRSestimatesseemtobeacceptableinthesestatesbutrequireasmallcorrectionatthenationallevelandforotherstates.

13

A lower (more masculine) than natural SRB obviously implies theoccurrence of gender biased sex selection. Such sex selection againstfemale births, indicates pre-natal discrimination, and is one factor

causingfemaledeficitinthepopulation.DepartureoftheSRBfromthenaturallevel,inconjunctionwiththenumberofbirths,givesanestimateofthenumberofmissing female births provided there is no sex bias in reporting of births.The other factor behind relative femaledeficit inpopulation is excess femalemortality. Generally femalemortality is lower thanmalemortality.Maternalmortality does certainly raise femalemortality in childbearing ages butwithsubstantialdecline in this, femaledisadvantagerelative tomales isno longeranissue.However, insomepopulations,femalemortality ishigherthanmalemortality even during childhood and in some, female childhoodmortality islowerthanmalemortalitybutnotaslowasseeninpopulationsatasimilarlevelofmalemortality.Femalemortalityhigherthanthatexpectedattheprevailinglevelofmalemortalityisattributedtofemaleneglectinhealthandnutrition,whichamountstopost-nataldiscrimination.Usingdemographictechniques,onecanestimatethesizeoffemalepopulationatagivenagethatwouldhavebeenpresentatatimepointintheabsenceofpre-andpost-nataldiscriminationandthe difference between this number and the number actually enumerated atthattimepointisthenumberof‘missingwomen’.AnumberofresearcheshaveprovidedestimatesofnumbersofmissingfemalebirthsaswellasestimatesofmissingwomenforanumberofcountriesincludingIndia;forrecentwork,seeBongaartsandGuilmoto(2015)andKashyap(2019).SomeIndiaspecificstudiesarebyArnoldetal.(2002),Bhat(2002),Jhaetal.(2006),Kulkarni(2007),andKauretal.(2017).

Inthisstudy,weestimatethenumberofmissingfemalebirthsaswellas thenumberofmissingfemalesofyoungages,ormissinggirls,onthebasisofIndia’s2011censusenumeration.Agemisreportingcanobviouslyinfluencetheestimatesandhencewechoosetheagegroup0-6,which,asnotedearlier,isleastaffectedbysexselectivemisreporting.Thus,theestimatesofmissingfemalebirthsrefertotheseven-yearperiodprecedingthe2001census,thatis,March2004toFebruary2011sinceMarch1,2011 is thereferencedateforthe2011censusenumeration.TheestimatednumbersofmissinggirlsarereckonedasonMarch 1,2011.TheprocedurehasbeendescribedintheAppendix.

Estimates of Numbers of Missing Girls and Missing Female Births

4

Female mortality higher than that expected at the prevailing level of male mortality is attributed to female neglect in health and nutrition, which amounts to post-natal discrimination

14

TheestimationhasbeendoneforIndiaandforlargestatesforwhichdataonlifetablesareavailableandtheresultsareshowninTable4.Theestimatesshowthattherewere2.6millionmissingfemalebirthsinIndiaduring2004-2011,anannualaverageof378thousandwhichamountsto3.1percentoffemalebirths.EstimatesofannualnumberofmissingfemalebirthsbyBongaartsandGuilmotofortheperiods2000-2005and2005-2010are0.62millionand0.63millionrespectively.TheseestimatesarehigherthanthatinthepresentstudyprimarilybecausetheBongaartsandGuilmotopaperisbasedontheU.N.estimatesofSRBwhicharemoremasculinethanthecensusbasedindirectestimateoftheSRBusedhere.ThenumbersofmissingfemalebirthsperannumarehighinUttarPradeshandMaharashtra(over50thousand)andmoderateinRajasthan,Gujarat,Haryana,MadhyaPradesh,andPunjab(25thousandormore).Inrelativeterms(numberof missing female births as percent of female births), Haryana and Punjabare verypoorlyplaced:missing female births amounted to over 10 percent offemalebirthsinthesestates,followedbyDelhi,JammuandKashmir,Gujarat,Maharashtra,Rajasthan,UttarPradesh,HimachalPradeshandMadhyaPradesh.Ontheotherhand,insomestates,Chhattisgarh,Kerala,Assam,JharkhandandWestBengal,theSRBduring2004-2011wasabove952;clearlygenderbiasedsexselectiondidnottakeplaceonanotablescaleinthesestates.NotethatestimatesforsmallstateshavenotbeenobtainedheresincetheSRSprovidessomeofthenecessarydataonlyforlargestates.However,estimatesoftheSRBbyKumarandSathyanarayana(2012)showthattheSRBduring2004-2011formostofthestatesinthenortheasternregionwashigherthan952andgenderbiasedsexselectionwasobviouslynotcommoninthesestates.

Estimates of the totalnumber ofmissing girls and also thedecomposition bypre-natal and post-natal discrimination are provided in Table 4. It is seenthat nationally about four million girls of ages 0-6 were missing at the 2011census. Of these, 2.5millionweremissing due to gender biased sex selection(pre-nataldiscrimination)and1.5millionduetoexcessfemalemortality(post-natal discrimination).3 The relative contributions of pre-natal and post-nataldiscriminationsareintheroughratioof5:3.

InHaryana,Punjab,JammuandKashmirandDelhithenumberofmissinggirlsofages0-6 is 10percentormoreoftheenumeratedgirlsofthisage.Intermsof volume, Uttar Pradesh, Maharashtra, Rajasthan, Bihar, Gujarat, MadhyaPradesh,Haryanahavelargenumbersofmissinggirls.Therelativesharesofpre-natalandpost-nataldiscriminationvaryacrossstates. InDelhi,Maharashtra,Punjab,JammuandKashmir,andHaryanaover80percentandinGujarat,andHimachalPradeshover70percentofmissinggirlsaremissedonaccountofpre-natal discrimination. As seen above, Chhattisgarh, Kerala, Assam, JharkhandandWestBengaldonotseemtohavenotablyhighprevalenceofgenderbiasedsexselection.However,eveninthesestatesthereisevidenceofsomepost-nataldiscrimination.Inaddition,inKarnataka,post-nataldiscriminationdominatesthecountofmissinggirls.InRajasthan,TamilNadu,AndhraPradesh,MadhyaPradesh, Uttar Pradesh, Odisha, and Bihar both the factors have substantialsharesincontributingtothenumberofmissinggirls.

3 The number of missing girls in the 2011 census due to pre-natal discrimination is slightly less than the number of missing female births during the preceding period of seven years because the missing girls due to pre-natal discrimination are the ‘expected numbers of survivors’ on March 1, 2011 of the missing female births during the preceding period.

It is seen that nationally about four million girls of ages 0-6 were missing at the 2011 census. Of these, 2.5 million were missing due to gender biased sex selection (pre-natal discrimination) and 1.5 million due to excess female mortality (post-natal discrimination)

15

Table 4:Table 4: Estimates of numbers of missing girls of ages 0-6 at 2011 census enumeration and missing female births during 2004-11, India and large states

Number of missing female births during 2004-11 (in thousands)

Number of missing girls 0-6 at 2011 census enumeration (in thousands )

Number of missing girls 0-6 at 2011 census enumeration as percent of enumerated girls

Total Annual as % of female births

Total

Missing due to Total

Missing due to

GBSS EFCM GBSS EFCM

India 2646 378 3.1 4007 2515 1492 5.1 3.2 1.9

AndhraPradeshβ 72 10 1.5 112 69 43 2.5 1.5 1.0

Assam neg neg Neg 32 neg 32 1.4 0.0 1.4

Bihar 103 15 1.0 309 98 211 3.3 1.1 2.3

Chhattisgarh neg neg Neg 30 neg 30 1.7 0.0 1.7

Delhi 88 13 9.1 94 86 8 10.0 9.2 0.9

Gujarat 250 36 6.4 307 238 69 8.4 6.5 1.9

Haryana 211 30 12.7 241 201 40 15.6 13.0 2.6

HimachalPradesh 15 2 3.9 21 15 7 5.8 4.0 1.8

Jammu&Kashmirα 93 13 9.4 107 90 18 11.5 9.6 1.9

Jharkhand neg neg neg 62 neg 62 2.4 0.0 2.4

Karnataka 11 2 0.3 47 11 36 1.3 0.3 1.0

Kerala neg neg neg 6 neg 6 0.3 0.0 0.3

MadhyaPradesh 180 26 3.2 285 167 118 5.5 3.2 2.3

Maharashtra 411 59 6.3 456 400 56 7.2 6.3 0.9

Odisha 29 4 1.0 68 27 41 2.7 1.0 1.6

Punjab 172 25 11.5 193 166 27 13.6 11.8 1.9

Rajasthan 339 48 6.1 452 317 135 9.0 6.3 2.7

TamilNadu 41 6 1.1 59 40 19 1.6 1.1 0.5

UttarPradesh 679 97 4.2 1113 634 480 7.6 4.3 3.3

WestBengal neg neg neg 30 neg 30 0.6 0.0 0.6

GBSS:Genderbiasedsexselection

EFCM:Excessfemalechildhoodmortality

neg:negligible.

β:IncludingTelangana.α:IncludingLadakh.

Source:Computationsbytheauthor.

Suchanestimationofmissinggirlsisnotpossibleforthepost-2011censusperioduntil the next enumeration takes place; this will occur in the 2021 census.However,wedohavedataonSRBfromtheSRSforsomeyearsafter2001andthesecanbeusedtoestimatethenumbersofmissingfemalebirthsforaperiodoffiveyearsaftertheenumerationandexcessfemalechildhoodmortality.TheprocedurehasbeendescribedintheAppendixandtheresultsaregiveninTable5.

16

Table 5:Table 5: Estimates of missing female births during 2011-16 and excess female deaths before age 5 for births of 2011-16, India and large states

State

Missing female birthsduring 2011-16

Excess female deaths before age 5 among births during 2011-16

Total(in thousands)

Annual (in thousands)

as % of female births

Number (in thousands)

as % of female births

INDIA 1941 388 3.0 895 1.4

AndhraPradeshβ 63 13 1.7 37 1.0

Assam 27 5 1.5 31 1.8

Bihar 80 16 1.1 93 1.3

Chhattisgarh neg neg neg 31 2.0

Delhi 65 13 9.5 6 0.8

Gujarat 238 48 7.8 31 1.0

Haryana 156 31 12.2 19 1.5

HimachalPradesh 12 2 4.5 1 0.4

Jammu&Kashmirα 26 5 5.0 5 0.9

Jharkhand 13 3 0.6 31 1.6

Karnataka neg neg neg 26 0.9

Kerala neg neg neg 6 0.5

MadhyaPradesh. 127 25 2.7 93 1.9

Maharashtra 249 50 5.5 27 0.6

Orissa neg neg neg 25 1.2

Punjab 66 13 6.3 12 1.2

Rajasthan 225 45 5.2 108 2.5

TamilNadu 77 15 2.8 15 0.5

UttarPradesh 667 133 4.9 312 2.3

WestBengal neg neg neg 35 0.8

Uttarakhand 37 7 9.6 3 0.9

β:IncludingTelangana;α:IncludingLadakh;neg:Negligible.

Source:Computationsbytheauthor.

ItisestimatedthatthenumberofmissingfemalebirthsinIndiaduring2011-2016was 1,941 thousandor about twomillion, anaverageof 388 thousandperyearandamountstothreepercentoffemalebirths,closetotheestimatefortheperiod2004-2011.Essentially,theSRBhasfluctuatedsince2000butnotshownacleartrendandhencethedegreeofsexselectionhasremainedfairlysteady.TheestimatebyBongaartsandGuilmoto(2015)fortheperiod2010-15is0.63millionper year; this is higher than our estimate of 0.388million because BongaartsandGuilmotohaveusedtheU.N.estimateofSRBwhichismoremasculinethantheadjustedvalueofSRBusedinthepresentstudy.Kashyap(2019)hasgivenanestimateof1536.3thousandfortheperiod2010-15,lowerthanourestimateof1941thousand;thisisdue,inpart,tosomedifferenceinthemethodologyadoptedand

17

inpart,duetotheadjustmentinSRBmadeinthepresentstudy(seeGuilmotoetal.,2020,foradiscussiononthedifferencesinmethodology).

Thegeographicpatternisfairlysimilartothatobservedfortheperiod2004-2011.ThenumberofmissingfemalebirthsisveryhighinUttarPradesh,Maharashtra,Gujarat,Rajasthan,HaryanaandMadhyaPradesh.InChhattisgarh,Karnataka,Kerala,Odisha,andWestBengal,thereisnoevidenceofsuchsexselectiononanotablescale. Inrelativeterms(assessed intermsof thenumberofmissingfemale births as percent of female births), Haryana shows the highest level(12.2 percent) followed by Uttarakhand, Delhi, Gujarat, Punjab,Maharashtra,UttarPradesh,HimachalPradesh,andJammuandKashmir.Inotherstates,therelativelevelislow.PunjabandJammuandKashmirshowanimpressivedeclineintheextentofgenderbiasedsexselectionthoughinbothofthesestates,thelevelcontinuestobewellabovethenationalaverage.

Excessfemaleunder-fivemortalityisoftheorderof1.4percentoffemalebirthsnationallyandintherangeof0.5to2.5percentforthelargestatesforwhichdatahavebeenexamined.ThedegreeisrelativelyhighinRajasthanandUttarPradeshandlowinHimachalPradesh,Kerala,TamilNadu,Maharashtra,WestBengal,Delhi, JammuandKashmir, andUttarakhand.Overall, the inter-statevariation in the level of post-natal discrimination is much less than in thelevelofpre-nataldiscrimination.Thereisnoevidencethatthesetwolevelsarecorrelated;thus,thedegreesofpre-natalandpost-nataldiscriminationsdonotseemtoberelatedempirically.

The number of missing female births is very high in Uttar Pradesh, Maharashtra, Gujarat, Rajasthan, Haryana and Madhya Pradesh

18

Ifthereisgenderbiasedsexselection,onewouldexpectittotakeplaceafterthefirstchildisborn,andthusaffectSRBatbirthsofhigherorders.ToseeifthisisthecaseinIndia,SRBestimateswereobtainedbybirthorder.Itis

possibletodosofromtheNFHSsincethesurveyhadobtainedfertilityhistoriesfromwomeninchildbearingagesandtheentiresequenceofbirthswasrecorded.EstimatesoftheSRBfromtheNFHS-3andNFHS-4werefirstobtainedbyfive-yearperiods.Ashasbeennotedearlier,estimatesforperiodsmuchearlierthanthesurveydatearepossiblyaffectedbyselectiveomissionofdaughtersandhenceonlythetworecentfive-yearperiodsduringwhichallthestateswerecoveredwereused.Theseare:1995-1999and2000-2004fromNFHS-3and2005-2009and2010-2014fromtheNFHS-4.

TheSRBhasbeenestimatedonlyforthefirstthreeorderssinceinthesamplesthenumbersofbirthsathigherordersaresmallandtheestimateswouldhaverelativelylargesamplingerrors.TheNFHS-3datafilescontainedinformationon58,128birthsduring1995-99ofwhich41,903wereofthefirstthreeordersand52,892birthsduring2000-04ofwhich39,870wereofthefirstthreeorders.TheNFHS-4datafilescontainedinformationon2,78,146birthsduring2005-09ofwhich2,17,327wereofthefirstthreeordersand2,63,508birthsduring2010-14ofwhich2,18,834wereof thefirst threeorders.

Further,sincethesamplingerrorsforstatelevelestimatesarelarge,estimatesareobtainedforgroupsofstatesinsixregions.TheregionalizationfollowedintheNFHS-4report(IIPSandICF,2017)hasbeenadoptedhere.Theregionsare:

Northern: JammuandKashmirandLadakh,HimachalPradesh,Uttarakhand,Punjab,Haryana,Delhi,Rajasthan,andChandigarh;

Western: Gujarat,Maharashtra,Goa,Diu and Daman, Dadra and Nagar Haveli;Central: UttarPradesh,Chhattisgarh,MadhyaPradesh;Eastern: Bihar,Jharkhand,Odisha,WestBengal;Northeastern: Sikkim,ArunachalPradesh,Nagaland,Manipur,Mizoram,Tripura,

Meghalaya,Assam;Southern: Andhra Pradesh, Telangana, Karnataka, Kerala, Tamil Nadu,

Lakshdweep, Puducherry, Andaman and Nicobar Islands.

Sixunionterritories,showninitalics,werenotincludedinNFHS-3.Sampleweightshavebeenapplied.

TheresultsarepresentedinTable6.Itisseenthatatthenationallevel,theSRBismoremasculinethanthenaturallevel(thatis,significantlybelow950)atthesecondorderinthreeperiodsandatthethirdorderinalltheperiods.Inthenorthernregion,theSRBishighlymasculineatthesecondorderandveryhighlymasculineat thethirdorder.Besides, theSRBat thefirstorder isalsohighlymasculineinthenorthernregionsince2000indicatingsomeselectionevenatthefirstbirth.TheSRBinthewesternregionisalsoveryhighlymasculineatthe

Sex Ratio at Birth by Birth Order5

The SRB has been estimated only for the first three orders since in the samples the numbers of births at higher orders are small and the estimates would have relatively large sampling errors

19

thirdorder.TheeasternandcentralregionshavemoderatelymasculineSRBatthesecondandthirdordersintherecentyears.Inthesouthernregion,highermasculinityisseenatthethirdorderinsomeperiods.Ontheotherhand,thenortheasternregionshowsnosuchpattern.

Overall,thereisclearevidencethatdisparitiesintheSRBacrossregionsincreasewiththebirthorder:theSRBismoremasculinethanthenatural levelatthethirdand secondorders. This is prominently so in thenorthernandwesternregions and to a lesser extent in the central, southern, and eastern regions.Essentially,theobservationsontheregionalpatternmadeearlierarereinforcedfromthetabulationsbybirthorder.

Table 6:Table 6: Sex Ratio Birth by birth order and time periods, India and Regions, NFHS-3 and NFHS-4(femalebirthsper1000malebirths)

Region

NFHS-3 1995-99 NFHS-3 2000-2004

Birth Order Birth order

1 2 3 All 1 2 3 All

Northern 999 863* 844* 894* 871* 861* 779* 842*

Western 979 901 887 926 907 834* 818 877*

Central 1018 926 982 948 972 945 928 935

Eastern 1017 848* 868 937 934 954 978 941

Northeastern 890 980 1009 952 1062 924 923 975

Southern 948 923 797* 914 969 959 810* 935

INDIA 988 898* 894* 930* 942 921 887* 918*

Region

NFHS-4 2005-2009 NFHS-4 2010-2014

Birth order Birth order

1 2 3 All 1 2 3 All

Northern 864* 866* 828* 859* 888* 882* 743* 863*

Western 946 917 707* 896* 945 902 767* 899*

Central 961 910* 909* 922* 926 894* 873* 906*

Eastern 979 939 894* 937 953 932 907* 936

Northeastern 926 1039 936 966 904* 959 925 917*

Southern 916 961 985 943 916 959 873* 924*

INDIA 937 926* 877* 919* 927* 919* 852* 911*

*:SRBislowerthan952femalebirthsper1000malebirthsatthe1%levelofsignificance.ForcompositionofRegions,seetext.Source:ComputedfromNFHS-3andNFHS-4datafiles.

800

850

900

950

1000

Birth order 3Birth order 2Birth order 1All

911927 919

852

The SRB is more masculine than the natural level at the third and second orders.

Fig. 3:Fig. 3: SRB by birth order, NFHS-4, India, 2010-14(Femalebirthsper1000malebirths)

Source:Table6

20

TheriseinthemasculinityoftheSRBinIndiaisattributedtosonpreferenceandgenderbiasedsexselectioninordertoavoidthebirthofagirl,oftencalled‘daughter avoidance’. In the context of low family size desires, for various

reasonsincludingquality-quantitytrade-offandgovernmentpolicies,sonpreferencecould lead to sex selectionrather thancontinuingchildbearinguntil thedesirednumberofsonsareborn;thiseffectiscalled‘fertilitysqueeze’or‘intensification’(Das Gupta and Bhat, 1997; Guilmoto, 2009; Bhalla et al., 2013). Correspondingly,theoverarchingfactorforsexselectionatthesecondorderbirthisthesexofthesurvivingfirstchildandathigherorderbirths, thesexcompositionof survivingchildren.Inotherwords,thedecisionsonsexselectionarelikelytobesequentialdependingonthesexofchildrenalreadybornandsuchbehaviormaybemanifestafterthefirstbirth.Ifthefirstbirthisagirl,thereis likelihoodofadoptingsexselectionandthiswouldthenresultinfewerdaughtersatthesecondbirthbutsuchatendencywouldbelesslikelyifthefirstchildisason.Similarly,theSRBatthethirdbirthwouldbemoremasculineifthefirsttwochildrenweregirls.

The NFHS had collected detailed fertility histories and from these conditionalSRB, that is, SRBconditionedon the sex compositionofprevious children, canbecomputed.Thishasbeendonefromthedataofthelatesttworounds,NFHS-3andNFHS-4.First, SRBat the secondbirthhasbeencomputed for thosewhosefirstchildwasasonandthosewhosefirstchildwasadaughter.Further,giventhattheSRBatthethirdordershowslargedeviationsfromthenaturalrangeasseenabove,SRBatthethirdbirthhasbeencomputedbysexcompositionofthefirsttwobirths.Itmustbenotedthatchildlosscouldinfluencesuchdecisions.Besides, in case twins are born, there isno sequential decisionmaking at thatstage.Therefore,incomputingSRBs,twinbirthsandthoseafterachildlossarenot taken intoaccount. Sampleweightshavebeenapplied.The tabulationsaremadeforIndiaandregionsbutnotstatessincethenumbersofbirthsbysexofthefirstbirthandbysexcompositionsofpreviouschildrenaresmallatthestatelevel.Further,thedenominatorsformanyofthecategoriesofsexcompositionofpreviouschildrenaresmallintheNFHS-3evenatthelevelofregionandhenceonlytheestimatesatthenationallevelarepresentedfromtheNFHS-3dataset.TheresultsareshowninTable7.

Atthenationallevel,theSRBatthesecondbirthwhenthefirstchildisadaughterismoremasculinethanthatwhenthefirstchildisasonandsignificantlylowerthan952femalebirthsper1000malebirthsinallthefourtimeperiods.Thisisalsoseeninthenorthern,western,andcentralregionsinboththeperiodsofNFHS-4 considered, 2005-09 and 2010-14, and in the southern region during 2005-09.Further,theSRBatthethirdbirthfollowingtwodaughtersissignificantlymoremasculinethanthenaturallevelinalltheperiods1995-99,2000-2004,2005-2009,and2010-14atthenationallevel.Thenorthern,western,andcentralregionsalsoshowsuchapatternduring2005-09andalltheregionsshowthispatternduring2010-14.Inparticular,

Conditional Sex Ratio at Birth6

the SRB at the second birth when the first child is a daughter is more masculine than that when the first child is a son and significantly lower than 952 female births per 1000 male births in all the four time periods

21

Table 7:Table 7: Sex Ratio at Birth by Sex Composition of Previous Births, India and regions(femalebirthsper1000malebirths)

Survey/RegionFirst Birth

Second Birth Third Birth

Sex of first birth Sex composition of first two births

Male Female Any Both sonsOne daughter and one son

Both daughters

Any

NFHS-3(1995-99)INDIA

979 928 863* 895* 970 860* 798* 866*

NFHS-3(2000-04)INDIA

941 1003 834* 913* 967 916 812* 894*

NFHS-4(2005-09)

Region

Northern 865* 1015 728* 863* 1004 882* 633* 812*

Western 946 1033 798* 910 918 735 579* 697*

Central 959 970 840* 904* 1004 909 857* 913

Eastern 981 917 959 939 901 905 883 898*

Northeastern 928 1040 1051 1046 971 899 897 915

Southern 917 1032 877* 953 1108 1089 850 998

INDIA 938 990 858* 922* 977 899* 781* 874*

NFHS-4(2010-14)

Region

Northern 888* 999 782* 880* 905 812* 585* 726*

Western 943 1082 775* 909 916 843 572* 721*

Central 927 929 845* 886* 956 897 766* 863*

Eastern 953 916 917 917 1081 930 760* 895*

Northeastern 902* 1020 904 959 905 981 822* 912

Southern 917 1001 919 960 986 834 788* 845*

INDIA 927* 974 860* 914* 984 884* 708* 832*

Note:Twinsandthosewiththepreviousbirth/birthsnotsurvivingareexcluded.*:SRBislowerthan952femalebirthsper1000malebirthsatthe1%levelofsignificance.ForcompositionofRegions,seetext.Source:ComputedfromNFHS-3andNFHS-4datafiles.

0

200

400

600

800

1000

AnyBoth daughters

1 daughter & 1 son

Both sons

FB - Any

FB - Female

FB - Male

First Birth(FB)

Second Birth Third Birth

927974

860914

984884

708

832

Fig. 4:Fig. 4: Conditional SRB by sex composition of previous births, NFHS-4, India, 2010-14(Femalebirthsper1000malebirths)

Source:Table6

22

TheSRBatthethirdbirthaftertwodaughtersisextremelylowinthenorthernandwesternregions.Moreover,theSRBfollowingonedaughterandonesonisalsomoremasculinethanthenaturallevelatthenationallevelinthreetimeperiods.TheNFHS-4datashowthisphenomenoninthenorthernregionandtoasmallerextentinthewesternregion.Thus,sexselectionatthethirdbirthwhenthefirsttwochildrenaredaughtersseemswidelyprevalentandthisisofahighdegreeinthenorthernandwesternregionsinwhichsexselectionisalsoseentobeprevalentatthesecondbirth.

These results are in broad agreement with the findings based on two othernational surveys, the Special Fertility andMortality Survey of 1998, and theIndiaHumanDevelopmentSurvey(IHDS)-1,aswellasfromanearlieranalysisoftheNFHS(see,Jhaetal.,2006;Desaietal.,2010,andJhaetal.,2011).

23

Guilmoto(2009),drawinguponCoale’spreconditionsforfertilitydecline,states:“Weadapttheseprerequisitestosexselectionbysayingthatparentshavetobeable,willing,andreadytopracticesexselection”(p.526).Various

socioeconomicfactorshaveabearingontheability,willingness,andreadinesstoresorttosexselection.Placeofresidence(rural-urban)hasimplicationsforaccessto the technology since sonographic scan centres aremostly located inurbanareas.Besides,placeofresidencemayalsoplausiblyimpactsexdifferentialintheperceivedvalueofchildren.Educationandincomeorwealthlevelmayalsoinfluenceaccesstotechnology(viaawarenessandaffordability)aswellasvalueofchildren.Stricturesandtenetsofreligionimpactdesirabilityofsonsasalsoacceptabilityofabortion.Socialbackgroundsuchascasteortribemembershipmayalsohaveabearingonrelativevaluesofsonsanddaughtersandthusonsonpreferenceordaughteravoidance.Mediaexposurebrings inawarenessoftechnologyandalsochangeinattitudesongenderissues.Valuesanddisvaluesofsonsvis-à-visdaughtersandsocialpressurestohavesonsmayalsovaryregionally.

Ideally, theeffectof thebackgroundfactorson theprobabilityof resorting togender biased sex selection should be analysed. However, gender biased sexselectionisillegalinmostcountriesandhencenotrecordedandcouplestoomaynotreport these in surveysdue to the illegalityandsocialdisapproval.But iftheSRBismoremasculinethanthenaturallevel,theimplicationisthatgenderbiased sex selectiondoes takeplace. Therefore, instead ofdifferentials in theprobabilityofgenderbiasedsexselection,onecanexaminedifferentialsintheSRB.Suchananalysisispossibleifdataonfertilityhistoriesorbirthsequencesareavailable.

7.1 Prior work for IndiaForIndia,researchershaveexaminedinfluencesofseveralfactorsontheSRBor on theprobability of a birth beingmale on the basis of thedata from theNationalFamilyHealthSurveys(NFHSs)andotherlargesurveys(RetherfordandRoy,2003; Jhaetal., 2006;BhatandZavier,2007;ArokiasamyandGoli,2012).The NFHS collects complete fertility histories and these data can be used toestimateSRBatvariousbirthordersandbysexcompositionofpreviouschildren.RetherfordandRoy (2003)analyseddata fromthe fertilityhistories fromtheNFHS-1 (carriedoutduring 1992-93)andNFHS-2 (1998-99)andobserved that inNFHS-1,theSRBatthethirdorderbirthswashighlymasculineforcoupleswithnolivingsonandinNFHS-2,SRBwashighlymasculineatboththesecondandthethirdordersforcoupleswithnolivingsons.Itwasfoundthattherearenotableregionalvariations,withstatesinthewesternregionshowinghighlymasculineSRBinNFHS-1andNFHS-2,moreprominentlyinNFHS-2.MultivariateanalysisusinglogisticregressionrevealedhardlyanyinfluencesofsocioeconomicfactorsinNFHS-1,exceptapositiveeffectofurbanresidence,aneffectnotseeninNFHS-

Differentials in Sex Ratio at Birth 7

“We adapt these prerequisites to sex selection by saying that parents have to be able, willing, and ready to practise sex selection”

24

2,andnegativeeffectofMuslimreligion.AnalysisoftheNFHS-2datashowedthatmiddleschoolorhighereducation,religionsotherthanHinduandMuslim,andmediaexposurehadaneffect(highermasculinity)ontheSRBatthesecondandhigherorderbirths.Effectsofotherfactorswereinsignificantorunclear.Thepaper carriedout logistic regressionanalysis separately for states in fourregions,North,West,EastandSouthbuthardlyanysignificantneteffectswereseeninthesestatelevelanalyses.

BhatandZavier(2007)alsoanalyseddatafromtheNFHS-1andNFHS-2.Theyfirstexamineddeterminantsoftheuseofpre-nataldiagnostictechnologiesbasedontheNFHS-2whichhadenquiredabout theuse of such technologies for birthsduringthethree-yearperiodbeforethesurvey.Theresultsoflogisticregressionanalysisshowedthatanumberofsocioeconomicfactorsdoinfluencetheuseofsuchtechnologies.However,thoughuseofpre-nataldiagnostictechnologyisaprerequisiteforsexselection,asitcanbeusedtodetectthesexofthefetus,itisnotnecessarilyusedforthatpurpose.Therefore,useofpre-nataldiagnostictechnology as such need not necessarily imply gender biased sex selection.Besides, those using such a technology for sex selectionmay not reveal it insurveys. The paper also examined determinants of the probability of a malebirth,usingdatafromtheNFHS-1andNFHS-2.Thiswasfoundtobesignificantlyhighinthenorth-westernandthenorth-centralregions.Socioeconomicfactorsshowedhardlyanysignificanteffects.However,theprobabilityofamalebirthwassignificantlyhighincaseofnosurvivingmalesiblingbeforethebirth.Theprobabilitywaslowerincaseofahighidealfamilysize.Maternalfactorssuchasage,anemiastatus,useofpre--natalcare,birthattendance,body-massindexwere also used in this analysis but it was also noted that there are issues ofmeasurementinrespectofthesevariables.

Jhaetal.(2006)analyseddatafromtheSpecialFertilityandMortalitySurvey(SFMS), a very large survey with a sample size of 1.1 million households (incontrast toabout 100 thousand in theNFHS-1andNFHS-2)andcovered 133738birthswhichoccurredduring1997.ThistoofoundthattheSRBatthesecondorderbirthswasveryhighlymasculineincasethefirstchildwasfemale.Similarly,theSRBatthethirdorderbirthswashighlymasculineifthefirsttwochildrenwerefemale.Amongsocioeconomicvariables,mother’seducationhadapositiveeffectontheprobabilityofamalebirthbutreligionandsizeofagriculturallandowneddidnotshowasignificanteffect.Atthethirdorderbirthsforthosewithno living sons, urban residence had a significant positive effect. ArokiasamyandGoli(2012)focusedontheruralpopulationandonthebasisoftheNFHS-3data(2012)notedapositiveeffectofthesizeoflandholdingontheprobabilityofamalebirthafteroneortwodaughters.Wealthindexalsoshowedapositiveinfluenceprominentlyatthesecondbirthwhereasmaternaleducationandcastedidnotshowacleareffectintheirstudy.

7.2 Analysis based on NFHS-4Thestudiescitedaboveuseddatareferringto the 1990sandearly2000s.Overtime, therehavebeenmanychanges; the technologyhasbecomemorewidelyavailablewiththespreadofpre-natalscanfacilities(forthepurposeofpregnancycare rather than for sex selection per se), the laws prohibiting sex detectionhavebeen strengthened (theamendedPCPNDTAct) in India, therehavebeencampaigns both by the government and by civil and religious organisationsagainstsexselection,andsocioeconomicchangeshavebeentakingplacewhichcouldplausiblychangefamilysizedesires,relativevaluesofsonsanddaughters,andgenderattitudes.

The SRB at the third order births was highly masculine if the first two children were female

25

Therefore, this studyexamines the influencesof socioeconomic factorson sexselectionindirectlyviatheprobabilityofamalebirthonthebasisofthedatafromthelatestroundoftheNFHS(NFHS-4)whichwasconductedduring2015-16throughoutIndia.ThesamplesizeoftheNFHS-4wasquitelarge,muchlargerthanthesize intheearlierrounds,andthustheNFHS-4providesrecentdatathattoofromalargesample.Theresultsfromthisanalysisarepresentedbelow.

7.2.1 Gross differentialsThe SRB at the first birth is computed by categories of selected backgroundcharacteristics: Place of residence (Rural, Urban), Education of mother (Noeducation,Primary,Secondary,Higher),Wealthindexquintiles(Poorest,Poorer,Middle,Richer,Richest),Religion(Hindu,Muslim,Otherreligions),SocialGroup(Scheduled Caste-SC, Scheduled Tribe-ST, Other Backward Caste-OBC, OtherslabelledasNonSC/ST/OBC),Exposuretomassmedia(Noorlow,Moderate,High)and Region (Central, Southern, Western, Northern, Eastern, Northeastern).TheregionalizationisinaccordancewiththatintheNFHS-4asnotedearlierinsection5.

ThewealthindexhasbeenconstructedintheNFHSonthebasisofhouseholdownership of assets and housing conditions and quintile groups have beenprovidedintheNFHSdatasets.ExposuretomassmediahasbeenascertainedbasedonresponsestoNFHSquestionsonwhethertherespondentreadsnewspapersorwatchestelevision;thosewhodonotreadnewspaper/watchtelevisionatallordosolessthanonceaweekarecategorisedashaving‘Noorlowexposure’,thosewhoreadnewspapers/watchtelevisionat leastweeklybutnotdailyashaving‘Moderateexposure’,and thosewhoreadnewspaper/watch televisiondailyashaving‘Highexposure’.

Theanalysishasbeendoneseparatelyfortwofive-yeartimeperiods,2005-2009and2010-2014.Estimatesforthetwotimeperiodsallowonetoseeifthepatternhas changedover time.Similarly,SRBat the secondbirthhasbeencomputedbythebackgroundvariablesseparatelyforthosewiththefirstbirthasonandthosewiththefirstbirthadaughter.Further,onlythosebirthswherethefirstchildwassurvivingareincludedsinceitisthesexcompositionofthenumberoflivingchildrenthatmattersinsexselection.TheSRBatthethirdbirthhasbeencomputedbythesexcompositionofthefirsttwobirths:twosons,onesonandonedaughter,andtwodaughtersandonlythosethirdbirthswherethefirsttwo children survived have been included. Twins have been excluded in thecomputationoftheSRB.Sincethenumberofbirthsbeyondthethirdorderwassmall,SRBhasbeencomputedonly for thefirst threeorders.Sampleweightshavebeenapplied.

Table8presentstheSRBsatthefirstbirthbybackgroundscharacteristics(thetablealsoshowsresultsoflogisticregressionanalysiswhicharediscussedlater,inthenextsub-section).TheSRBintheurbanpopulationislowerthantheruralpopulationandsignificantlylowerthan952intheperiods2005-2009,and2010-2014.Differentials by education ofmother donot showa pattern. The SRB islowerfortherichestwealthindexgroupcomparedtothepoorest.Differentialsbyreligionandsocialgroupdonotshowaclearpattern.ThenorthernregionshowslowerSRBthanotherregionsinboththeperiods.TheSRBissignificantlylowerthan952athighmediaexposureonlyintherecentperiod.

AsseeninTable7,ifthefirstbirthisason,theSRBatthesecondbirthdoesnotshow lower thannaturalSRB inanyof the regions.Hence, resultsof furtheranalysisinthiscasehavenotbeenpresentedhere.Ifthefirstchildisadaughter,

The SRB is lower for the richest wealth index group compared to the poorest

26

theSRBatthesecondbirthissignificantlylowerthan952inbothruralandurbanareas,amongbirthstowomenwithsecondaryandhighereducation,andfromupperquintilesofwealthindex(Table9).TheSRBforbirthstowomenbelongingtoreligionsexceptMuslim,andthosenotbelongingtoSCandSTcategoriesisalsolow.Thenorthern,western,andcentralregionsalsohavelowSRB;itisverylowinthenorthernregion.TheSRBisalsolowforthosewithhighmediaexposure.SomecategoriesshowalowSRBinoneofthetwoperiods.Overall,thereisclearevidenceofsexselectionatthesecondbirthifthefirstchildisadaughterandthis is seenat thenational level, in someregions,prominently thenorthern,andinsociallyoreconomicallyadvantagedclasses.Ontheotherhand,amongMuslims,thereisabsenceofsexselectionatthisstage.

Table 8:Table 8: Sex Ratio at First Birth and Adjusted Sex Ratios by Background Characteristics, India, NFHS-4(SRBisexpressedasfemalebirthsper1000malebirths)

Characteristic Period 2010-14 2005-09

CategoriesSRB

Adjusted SRB

ODDS RATIO SRB

Adjsuted SRB

ODDS RATIO

Placeofresidence Rural®

Urban

935

910*

931

918

1

1.01

947

919

940

933

1

1.01

Educationof

woman

Noeducation®

PrimarySecondary

Higher

959

901

919*

936

950

893

918

957

1

1.06

1.03

0.99

917

991

939

914

897

981

946

947

1

0.91@

0.95

0.95

Wealthindex

quintile

Poorest®

Poorer

Middle

Richer

Richest

937

971

899*

936

895*

920

971

904

943

896

1

0.95

1.02

0.98

1.03

979

926

945

950

889*

970

921

946

955

898

1

1.05

1.03

1.02

1.08@

Workstatusof

woman

Didnotwork®

Worked

927

939

927

931

1

1.00

941

872

942

869

1

1.08

Religion Hindu®

Muslim

Other

930

917

911

929

919

920

1.00

1.01

1.01

943

931

878*

943

923

897

1

1.02

1.05

Socialgroup NonSC/ST/OBC®

SC

ST

OBC

931

915

992

916*

938

912

985

915

1

1.03

0.95

1.02

947

930

965

930

953

927

958

928

1

1.03

0.99

1.03

Region Central®

Southern

Western

Northern

Eastern

Northeastern

927

917

943

888*

953

903

925

925

943

895

947

885

1

1.00

0.98

1.03

0.98

1.05

959

917

946

865*

981

928

958

924

944

877

970

917

1

1.04

1.02

1.09@

0.99

1.05

Mediaexposure Noorlow®

Moderate

High

941

924

921*

925

919

929

1

1.01

1.00

959

933

928

947

925

935

1

1.02

1.01

All 927* 938

Note:Twinbirthsexcluded.AdjustedSRBobtainedbasedonlogisticregressionanalysis.

*:indicatestheSRBissignificantlylowerthan952at1%levelofsignificance.

®:Referencecategory.

@:indicatestheoddsratio(oddsofabirthbeingmale)isdifferentfromthereferencegroupat1%levelofsignificance.

27

Table 9:Table 9: Sex Ratio at Second Birth after First Daughter and Adjusted Sex Ratios by Background Characteristics, India, NFHS-4 (SRBisexpressedasfemalebirthsper1000malebirths)

Characteristic Period 2010-14 2005-09

Categories SRB

AdjustedSRB

ODDS RATIO SRB

Adjusted SRB

ODDS RATIO

Place ofresidence

Rural®Urban

876*825*

856868

10.99

877*817*

842891

10.95

Education ofwoman

Noeducation®PrimarySecondaryHigher

913902855*719*

878879864777

11.001.021.13@

902879844*714*

838848879836

10.990.951.00

Wealth indexquintiles

Poorest®PoorerMiddleRicherRichest

912921876*854*727*

884905870860770

10.981.021.031.15@

972888892838*694*

938879900848719

11.071.041.111.30@

Workstatusofwoman

Didnotwork®Worked

863*785*

863784

11.10

855*897*

855895

10.96

Religion Hindu®MuslimOther

844*970831*

843968859

10.87@0.98

850*923810*

849912844

10.931.01

Socialgroup Non-SC/ST/OBC®SCSTOBC

824*886875866*

841881871858

10.950.970.98

806*918856*863*

826912818861

10.91@1.010.96

Region Central®SouthernWesternNorthernEasternNortheastern

845*919775*782*917904

837948791805881855

10.88@1.061.040.950.98

840*877798*727*9591050

8249008257639121023

10.92@1.001.080.90@0.81@

Mediaexposure Noorlow®ModerateHighAll

910917827*860*

875911844

10.961.04

966808*814*857*

923796835

11.16@1.11@

Note:SeefootnotestoTable8.

Onlythosebirthswiththefirstchildsurvivingareincluded.

For those with the first two sons, the SRB at the third birth is not highlymasculine(seeTable7).WhileinsomecategoriestheSRBisactuallyhigherthan952,inmostcasesthisisnotsignificantlyso.Moreover,ahighvalueofSRBneednotbetakentoimplysexselectioninfavouroffemalessincesofarnoevidencehas emerged, from research studies or anecdotal, of the prevalence of such apractice.Hence, further analysis forwomenwith thefirst two births as sonsisnotpresentedhere.Forthosewithonedaughterandoneson,theSRBatthethirdbirthislowinsomecategoriesespeciallyintherecentperiod,2010-14,butthereisnoconsistentpatternovertimeandacrosscategoriesofthebackgroundvariablesexaminedhere(Table10).

28

Table 10:Table 10: Sex Ratio at Third Birth after One Daughter and One Son and Adjusted Sex Ratios by Background Characteristics, India, NFHS-4(SRBisexpressedasfemalebirthsper1000malebirths)

Characteristic

Period 2010-14 2005-09

Categories SRB

Adjusted SRB

ODDS RATIO SRB

Adjusted SRB

ODDS RATIO

Placeofresidence

Rural®Urban

921765*

905816

11.11

897*906

894915

10.98

Educationofwoman

Noeducation®PrimarySecondaryHigher

922946791*994

8879408341160

10.941.060.76

930859875762

930874871708

11.061.071.31

Wealthindexquintile

Poorest®PoorerMiddleRicherRichest

981885858813*676*

939872884871722

11.081.061.081.30@

945871852865984

9338678578751011

11.081.091.070.92

Workstatusofwoman

Didnotwork®Worked

885*863

887842

11.05

902*860

901874

11.03

Religion Hindu®MuslimOther

882*909818*

870963796

0.000.901.09

884*966891

886956902

0.000.930.98

Socialgroup NonSC/ST/OBC®SCSTOBC

8529011011858*

865908993853

10.950.871.01

871867926922

880869941913

11.010.940.96

Region Central®SouthernWesternNorthernEasternNortheastern

897835842812*930989

886892889845889945

10.991.001.051.000.94

9101091736*883906891

9051116753885894887

10.81@1.20@1.021.011.02

Mediaexposure Noorlow®ModerateHighAll

945895810*884*

896898866

11.001.04

914902883899*

902923890

10.981.01

Note: See footnotes to Table8.

Onlythosebirthswiththefirsttwochildrensurvivingareincluded.

Forwomenwhosefirsttwochildrenaredaughters,theSRBatthethirdorderishighlymasculineoverallandformostofthecategoriesofbackgroundvariables(Table 11). The SRB is extremely low in the top two wealth and educationcategories,inurbanpopulation,non-Muslims,socialgroupnon-SC/ST/OBC,andthenorthernandwesternregions.Forthemostrecentperiod,2010-14,theSRBissignificantlylowerthan952inallthecategoriesexceptthereligioncategoryMuslimandinthesouthernandnortheasternregions.However,somecategories(loweducation,lowwealth,Muslims,scheduledtribes,scheduledcastes,eastern,northeastern, and southern regions, and low andmoderatemedia exposure),donotshowsignificantlylowSRBduring2005-09.TheSRBatthisstage(thirdorderbirthaftertwodaughters)hasbecomemoremasculineintherecentyearsandmorepervasivethaninthepast.Besides,widedifferentialspersist.Clearly,

29

strongsonpreferencehas ledtoahighdegreeofsexselectionifthefirsttwochildrenaregirlsandthisisparticularlysoforsomesocioeconomicclassesandinthenorthernandwesternregions.

7.2.2 Multivariate analysis: Net differencesThedifferentials inSRBnotedabovearegrossdifferences.Sincemanyof thesocioeconomicfactorsarehighlyassociatedleadingtopossibleconfoundingofeffects, it isnecessary toassessnet influencesofvarious factors.To this end,logisticregressionanalysishasbeencarriedoutwiththesexofthebirth,maleor female, as thedichotomousdependentvariable and the socioeconomicandspatialvariableslistedintheprevioussub-sectionasexplanatoryvariables.Thedichotomousdependentvariablehasavalueof1formaleand0forfemale;thustheanalysisexaminesthenetinfluencesofvariousfactorsontheprobabilityofabirthbeingamale,controllingfortheeffectsofotherexplanatoryvariablesusedintheanalysis.

Therearefourregressions:i)forthefirstbirth,ii)forthesecondbirthforthosewhosefirstbirthwasadaughter, iii) for the thirdbirth for thosewhosefirsttwobirths includedonesonandonedaughter,and iv) for the thirdbirthforthosewhosefirsttwobirthsweredaughters.Since,asnotedabove,theSRBatthesecondbirthfollowingonesonandatthethirdbirthfollowingtwosonswasnotfoundtobehighlymasculine,resultsforthesesituationsarenotpresented.Analysis has been carried out for the time periods 2005-2009 and 2010-2014separately.

For women with two daughters, the SRB at the third birth is significantly lower than 952 in all the categories except the religion category Muslim and in the southern and northeastern regions

30

Table 11:Table 11: Sex Ratio at Third Birth after First Two Daughters and Adjusted Sex Ratios by Background Characteristics, India, NFHS-4(SRBisexpressedasfemalebirthsper1000malebirths)

Characteristic Period 2010-14 2005-09

Categories SRB

Adjusted SRB

ODDS RATIO SRB

Adjusted SRB

ODDS RATIO

Placeof

residence

Rural®

Urban

760*

582*

724

665

1

1.09

818*

680*

759

844

1

0.90

Educationof

woman

Noeducation®

Primary

Secondary

Higher

820*

733*

656*

356*

759

706

702

432

1

1.08

1.08

1.76@

902

834

665*

342*

805

824

760

473

1

0.98

1.06

1.70@

Wealthindex

quintile

Poorest®

Poorer

Middle

Richer

Richest

795*

852

706*

593*

481*

690

814

730

657

606

1

0.85@

0.95

1.05

1.14

950

903

776*

693*

435*

864

878

802

745

507

1

0.98

1.08

1.16

1.70@

Worksstatus

Ofwoman

Didnotwork®

Worked

710*

673*

711

645

1

1.10

792*

623*

793

612

1

1.30@

Religion Hindu®

Muslim

Other

690*

807

700*

680

856

745

1

0.79@

0.91

768*

904

625*

762

925

674

1

0.82@

1.13

Socialgroup NonSC/ST/OBC®

SC

ST

OBC

610*

767*

807*

714*

648

767

780

697

1

0.85@

0.83

0.93

642*

827

966

794*

697

809

945

775

1

0.86

0.74@

0.90

Region Central®

Southern

Western

Northern

Eastern

Northeastern

766*

786

573*

584*

760*

822

751

862

616

628

697

721

1

0.87

1.22@

1.20@

1.08

1.04

857

850

579*

633*

882

888

826

935

621

714

791

813

1

0.88

1.33@

1.16

1.04

1.02

Mediaexposure Noorlow®

Moderate

High

All

841*

705*

618*

708*

784

692

657

1

1.13

1.19@

931

908

654*

780*

837

897

716

1

0.93

1.17@

Note:SeefootnotestoTable8.

Onlythosebirthswiththefirsttwochildrensurvivingareincluded.

Alltheexplanatoryvariablesarecategorizedandthereferencecategoryineachhas beendesignated. The odds ratio for a specific category shows the ratio ofoddsofabirthbeingamaleinthespecificcategorytotheoddsforthereferencecategory. On the basis of the regression coefficients, the predicted value ofthe probability that a birth is male has been computed for each category ofthe explanatory variables, holding other variables constant. These are thusadjustedprobabilities,adjustedfortheeffectsofothervariables,asinMultipleClassificationAnalysis(MCA).Fromtheseprobabilities,thecorrespondingvaluesofadjustedSRB(femalebirthsper1000malebirths)forcategoryjofvariableihavebeencomputedsimplyas,AdjustedSRBij=1000x(1-Adjustedpij)/Adjustedpij,whereAdjustedpij is theadjustedprobability of a birthbeingamale forcategoryjofvariableicomputedfromthecoefficientsofthelogisticregression

31

equation.Twinbirthshavebeenexcluded intheanalysis.Further,onlythosebirthswhere thefirst childwas survivingare included in the analysis of thesexofthesecondbirthandonlythosebirthswherethefirsttwochildrenweresurviving are included in the analysis of the sex of the third birth. Sampleweightshavebeenapplied.

Table 8 presents the results for the sex of the first birth. The columns showthe unadjusted SRB, the adjusted SRB, and the odds ratio for each category.TheadjustedSRBsvarysomewhatfromtheunadjustedSRBsbutgenerallythedifferencesaresmall.Hardlyanyoddsratiosaresignificant(thatis,thelogisticregressioncoefficientsforthecategoriesarenotsignificant).Thisistrueofboththetimeperiods2005-09and2010-14.Asnotedabove,grossdifferentialsarealsominorandshowednopattern.Thus,theSRBatthefirstbirthdoesnotseemtobenotablyinfluencedbyanyofthebackgroundfactors.

Analysisforthesexofsecondbirthincasethefirstisadaughterrevealsthattherichestwealthcategory shows lowerSRBcompared to thepoorest inboththe periods and the higher education category compared to those with noeducationduring2010-14(Table9).MuslimreligionhassignificantlyhigherSRBthanHindusduring2010-14.SocialgroupsgenerallydonotshowsignificantnetdifferencesexceptforSCsduring2005-09.SomeregionshavehigheradjustedSRBthanthecentralregion(referencecategory);thesoutherninboththeperiodsandtheeasternandthenortheasterninonlyoneperiod.Overall,socioeconomicdifferencesarenotveryconsistentexceptthelowvaluesforthehigheducationandwealthindexgroups.NotethatthelackofsignificantlyhighoddsratiosdoesnotimplythattheSRBisnothigherthanthenatural.Infact,forthesecondbirthafteradaughter,theSRBishighlymasculineformostcategories.

Forthethirdbirthafteronedaughterandoneson,theoddsratiosbyandlargedo not differ significantly from 1 (Table 10). Thus, the adjusted SRB formostcategoriesdoesnotdifferfromthatforthereferencecategory.Thereareonlyafewexceptions,richestcategoryduring2010-14,andthesouthernandwesternregionsduring2005-09.Incontrast,notabledifferencesinadjustedSRBareseeninthecaseoftheSRBatthethirdbirthfollowingtwodaughters,(Table11).Theoddsratioforthehigheducationcategoryissignificantlyhighinboththetimeperiods and for the richest category in oneperiod. The adjusted SRBs for thetopeducationandwealthcategoriesarehigherthantheunadjustedSRBs;thusthenet effect isnot as strong as seen in the gross differentials; yet even theadjustedSRBsarequitelowforthesecategories.Ontheotherhand,theSRBforMuslimsissignificantlyhigherthanHindus inboththeperiods.Noclearnetdifferencesareseenacrosssocialgroups.ThenorthernandwesternregionsshowsignificantlylowerSRBthanthecentralregion(whichtooshowsalowSRB)inoneorbothperiods.TheSRBathighmediaexposureislowerthanatnoorlowlevelofexposure.TheoverallSRBatthisstage,thatis,atthethirdbirthaftertwodaughters,islow(thatis,highlymasculine),andthisissoforthereferencecategoriesofall thesevenvariables inthe2010-14periodandexceptfor thosewithlowlevelsofeducationandthepoorest,inthe2005-09periodaswell.AhighvalueoftheoddsratioforanycategorythenimpliesaveryhighlymasculineSRBevenaftereffectsofotherfactorsareadjusted.This isseenfromtheadjustedSRBs shown in the table.

Overall, as assessed from the SRBs at different stages of family building, sexselectionatthefirstbirthdoesnotseemtobeatanotablelevel.Thereissomeindication of this in the northern region and for the richest quintile of thepopulationbuttheeffectsseemtobecomeweakeronceotherfactorsarecontrolled.

Analysis for the sex of second birth in case the first is a daughter reveals that the richest wealth category shows lower SRB compared to the poorest in both the periods and the higher education category compared to those with no education during 2010-14

32

Ontheotherhand,sexselectionisquiteconspicuousatthesecondbirthincasethe first was female. This cuts acrossmost of the regions and socioeconomicbackgrounds.Thedegreeisquitehighfortherichestquintileandtosomeextentforthericherquintileaswellasforthosewithhighereducationbutrelativelylower forMuslims. There is some sex selection at the third birth in case thefirsttwoincludeadaughterandason.Thisisespeciallyseeninthenorthernregionandtherichestquintile.Butnetdifferencesaregenerallynotsignificant.Sexselectionishighatthethirdbirthforthosewhosefirsttwochildrenweredaughters.Thisisseenregardlessofbackgroundcharacteristicsandcutsacrossalltheregionsintherecentyears.Moreover,thenorthernandwesternregions,thewealthiestsectionsofpopulation,andthosewithhighereducationshowveryhighdegreeof sex selection.The relativelyhigherprevalenceof sex selectioninhighereducationaswellaswealth/incomeclassesinIndiacould,atleastinpart,beattributedtothegreaterawarenessof,accessto,andaffordabilityofthetechnologyofsexselectionforthesesectionsofpopulation.Religiousandmoralconsiderationsand societalnormsmayprohibit access to abortionand this isprobablythereasonfortheabsenceorverylowprevalenceofsexselectionamongMuslims.

33

Theprimemotive for resorting to sex selection, specifically of amale, isastrongpreferenceforsonsoverdaughters.Sonpreference isknowntobewidelyprevalentaroundtheworld(Williamson,1976)andespeciallyin

partsofAsia(DasGuptaetal.2003).Manycoupleswouldlikethefamilylinetocontinueandinpatrilinealsocietiesmaleoffspringareneededforthispurpose.Besides,certainrituals,especiallythoserelatedtodeath(burialorlightingthefuneralpyre)andancestorworshipare traditionallyperformedby sons.Suchvaluesmaybeconsideredsentimentalorritual.Formorepracticalandmaterialneedssuchasoldagesecurity,traditionallysonsarevaluedoverdaughters.Thismayinvolveresidenceduringoldage,especiallywhenthecouplecannolongerwork or take care of day-to-day household management, as well as financialsupport. Inpatrilocal (or virilocal) arrangements, sonsand their families areexpectedtoresidewithparentswhereasamarrieddaughterwouldresidewiththeparentsofherhusband.Childrenalsocontributetofamilylabour(labelledas‘productionutility’byLeibenstein,1974)andformanyagriculturaloperationsandenterprisessonsarevaluedoverdaughtersthoughdaughtersdocontributetohouseholdworkandalsotoagriculturalandsimilaractivities.However,overtime,theextentofchildlabourhasdeclinedandthelabourvalueofchildrenisnolongeramajorissue.Expensesonmarriagesofdaughters,especiallydowryandcostsofthecelebration,ifexpectedtobebornebythefamilyofthebride,leadto‘disvalues’ofdaughters.Anotherfactoristheconsiderationofsecurityofunmarrieddaughtersandapprehensionsonsafetyofunmarriedgirlscouldleadto‘daughteraversion’.Overall,forvariousreasons,theperceivedvalueofsonsdiffersfromthatofdaughters.Thedegreeofsonpreferencedependsonhowhighistheperceivedvalueofsonsvis-à-visdaughters.Suchvaluesmayvaryacrosssocial,cultural,andeconomicsettings.

8.1 Evidence on values of sons vis-à-vis daughters in India

Thereisahugebodyofliteratureonperceivedvaluesofadaughtervis-à-visasoninIndia.KaurandKapur(2018)provideacomprehensivereviewofthestudieswithfocusonrecentchangesinvariousaspects,namely,education,marriage,work, rituals, and property. A field investigation in five states examinedfactorsthatleadtosonpreference(Johnetal,2008).InamorerecentstudyinHaryanaandMaharashtra,John(2018)adoptedaqualitativeapproachtoinquireinto recent changes in attitudes towards daughters and sons. Further, threestudiescommissionedby theUNFPA inTamilNadu,Maharashtra,andPunjabalsoaddressed these issuesasperceivedby ‘daughtersonly’ families followingqualitativemethods (Gokhale InstituteofPoliticsandEconomics, 2017).Thesestudies do show that attitudes have changed in the recent years. There isnowa strongdesire among couples to educatedaughters. Though the level ofeducationdesiredfordaughtersmaynotbeashighasthatdesiredforsons,oftenat least high school education and inmany cases higher education is sought

Why Sex Selection? 8

34

fordaughters.This is inordertoenabledaughterstoseekemploymentinthemodernsector(sothattheyareselfreliantandcan‘standontheirfeet’)andalsotoimproveprospectsoffindingagoodgroom.Theneedofasonforcontinuinglineageandforritualsincludingfuneralandancestorworshipseemstobenotasstrongasinthepastandthereissomeacceptanceofdaughtersperformingsuch rituals.Thoughdaughtersare legally entitled to inherit familypropertyincludingland,generallymanywomenforgotheirshareinancestralpropertyandletbrothershaveit.Itisunderstoodthatdaughterswouldgetsomeshareatthetimeofmarriageinaformotherthanthefamilyhouseorlandandhencewouldnotexerciserighttoparentalpropertylater.Incaseofacouplewithnosons,adaughterinheritsthepropertyifherhusbandmovesinwithherparents(thehusbandisthencalled‘gharjamai’)andthusthedaughter’shusbandandchildrenresidewithherparents.Butparentsgenerallydonotprefertomovetothedaughter’smaritalhome.Financialsupportfromdaughtersisgenerallynotexpected.However,many‘daughtersonly’couplesnotedthatsonstoomaynotprovidesupportastheymaygetinto‘undesirablehabits’andcannotbereliedupon.While thiscouldbepost-factoattitudeorrationalization,complaintsofyoungmengettingintodrinkingordrugusearecommon.Thoughthemarriageofadaughterinvolvesexpensesincludingdowry,thegroom’sfamilyalsohastoincurexpendituresandheavycostsofperformingadaughter’smarriagedidnotfigureasanotablefactorcausingdaughteravoidanceinresponsesofparticipantsof these studies.

Inadditiontothequalitativestudies,theIndiaHumanDevelopmentSurvey-II,carriedoutduring2011-12inIndia(IHDS,2018),alsocollectedinformationonoldage security expectations from children. The survey covered 42152householdsand39253evermarriedwomenof reproductiveageswere interviewed. In thesurvey,marriedwomenwereaskedaseriesofquestionsonexpectationsofoldagesecurityandfromtheresponsesonecanseewhetherthedegreeofdependenceonsonsvis-à-visdaughtersvariesacrossbackgroundcharacteristicsandregions.Thewomenwereasked:17.41:Whodoyouexpecttolivewithwhenyougetold?:‘Son’, ‘Daughter’, ‘Both’, and ‘Others’ were listed as responses. Those whoseresponsedidnotinclude‘Daughter’werefurtherasked:17.42Wouldyouconsiderlivingwithyourdaughter?Similarquestionswereaskedonfinancial supportduringoldage:17.43Whodoyouexpectwillsupportyoufinanciallywhenyougetolder?Andagainthosewhoseresponsedidnotinclude‘Daughter’wereasked:17.44:Wouldyouconsiderbeingfinanciallysupportedbyyourdaughter?Inthecaseoflivingarrangements,thosewhosefirstresponse(Q.17.41)didnotinclude‘Daughter’ were subdivided into ‘Sons only’ (response as ‘Son’ to Q.17.41 and‘No’toQ.17.42)and‘SonorDaughter’(Response‘Son’to17.41and‘Yes’to17.41).Similarly,thecategories‘Othersonly’and‘Othersordaughter’werecreated.Inalikemanner,responsestoquestionsonfinancialsupportduringoldagewerecategorized.

The responses are tabulatedbykeybackground characteristics inorder to seeifperceivedvaluesvarybysocioeconomicbackground.Thecharacteristicsusedare:placeofresidence(rural/urban),educationallevelofwoman(no,primary,secondary,higher),percapitaincomequintiles(poorest,poorer,middle,richer,andrichest),workingstatusofwoman(Yes:working,No:notworking),religion(Hindu, Muslim, other), social group (SC, ST, OBC, and Non-SC/ST/OBC), andownershipofagriculturallandbythehousehold(No,Yes).Besides,tabulationsarealsomadebyregion,forthesixregionsasintablesbasedonNFHSpresentedearlier.Further,thetabulationsbyregionarealsoprovidedseparatelyforthosewithnolivingsonandthosewithatleastonelivingsoninordertoseewhattheexpectationsofthosewithoutasonare.Sampleweightshavebeenapplied.

Though daughters are legally entitled to inherit family property including land, generally many women forgo their share in ancestral property and let brothers have it

35

ItisseenfromTable12thatamajorityofwomen,65percent,expecttoliveonlywithsons inoldage incontrast to less than 10percentexpecting to livewithdaughters.Thisistobeexpectedgiventhepatrilinealandpatrilocaltraditionsinmostofthecountry(somecommunitiesinpartsofthenortheastandthesouth-westerncoastdonot followpatrilocal traditionsand,according to tabulationsbystates,notshownhere,86percentofwomeninMeghalayaexpecttolivewithdaughtersinoldage).Variationsbysocio-economicbackgroundfactorsarenotsoconspicuousexceptthatrelianceonsonsisslightlyhigherthanaverageamongnon-SC/ST/OBCwomen.Nodifferentials are seenbyplace of residence (rural-urban)andbyownershipofagriculturalland.Theexpectationtolivewithsonsishigherthanthenationalaverageinthewesternandnorthernregionsandlowerinthesouthernregion.Separatetabulationsbythenumberofsonsatthetimeofthesurveyshowthatevenamongthosewithnosonsatthetimeofsurvey,about25percentmentionedthattheyexpecttoliveonlywithsons(showninthelowerpanelsofthetable);thispercentageishigher,over40percent,inthewesternandnorthernregions.

Forfinancialsupportduringoldage,amajorityofwomen(57percent)expecttorelyonlyonsons(Table13).Again,thislevelishighinthewesternandnorthernregionsthanintheotherregions.Suchexclusiverelianceonsonsistheleastinthenortheasternregionandlowerthanaverageinthesouthern(inMeghalaya,relianceexclusivelyonsonsisonly11percent).Differentialsbysocioeconomicfactorsarenarrow.Amongthosewithnosonsatthetimeofsurvey,aboutafifthexpectedfinancialsupportonlyfromsons.

36

Table 12:Table 12: Expected living arrangements in old age by background characteristics, ever married women of ages 15-49, India, IHDS-II

Characteristic Category

(Percent who expect to live with)

Sons only Daughters Both S & D

Others only

Son or Daughter

Others or daughter

Total

PlaceofresidenceRuralUrban

65.065.3

7.39.0

5.34.8

4.04.3

18.016.3

0.50.4

100100

EducationlevelcompletedNoPrimarySecondaryHigher

65.067.265.060.3

6.17.39.212.2

4.95.15.45.3

2.94.54.87.0

20.915.115.214.9

0.30.70.50.3

100100100100

PercapitaincomequintilesPoorestPoorerMiddleRicherRichest

63.364.265.466.066.6

6.87.38.08.48.7

5.55.25.84.74.7

3.63.93.94.34.7

20.419.116.616.314.5

0.40.30.30.30.8

100100100100100

WorkingpresentlyNoYes

66.163.6

7.09.0

5.44.9

4.33.8

16.818.3

0.50.4

100100

ReligionHinduMuslimOtherreligions

65.763.158.5

7.76.911.8

4.77.48.0

4.24.12.8

17.418.116.8

0.40.32.0

100100100

SocialgroupNonSC/ST/OBCOBCSCST

70.062.965.060.3

6.68.37.89.3

4.74.85.28.8

4.33.64.54.8

13.920.216.916.2

0.50.30.50.6

100100100100

OwnagriculturallandNoYes

64.665.6

9.16.4

5.15.2

4.14.1

16.718.2

0.40.4

100100

All 65.1 7.8 5.2 4.1 17.4 0.4 100

RegionCentralSouthernWesternNorthernEasternNortheasternAll

63.653.381.772.563.360.865.1

5.916.93.43.35.513.87.8

5.75.52.13.05.719.75.2

5.03.23.03.35.72.44.1

19.820.49.017.619.62.717.4

0.10.70.80.30.30.70.4

100100100100100100100

Amongthosewithnolivingson

RegionCentralSouthernWesternNorthernEasternNortheasternAll

20.910.047.242.426.722.824.7

37.066.120.924.129.743.641.0

8.63.92.65.310.018.16.7

23.811.217.514.219.810.016.7

9.26.46.812.312.12.78.7

0.42.45.01.81.72.72.2

100100100100100100100

Amongthosewithatleastonelivingson

RegionCentralSouthernWesternNorthernEasternNortheasternAll

72.366.387.877.170.771.573.7

0.22.30.30.20.65.30.9

5.16.02.02.64.820.14.8

0.50.60.31.52.80.21.2

21.924.79.518.421.12.719.3

0.00.10.10.10.00.10.1

100100100100100100100

Source:ComputedfromIHDS-IIdatafiles.

37

Table 13:Table 13: Expected financial support in old age by background characteristics, ever married women of ages 15-49, India, IHDS-II

Characteristic Category

(Percent who expect support from)

Sons only Daughters Both S & D

Others only

Son or Daughter

Others or daughter Total

PlaceofresidenceRuralUrban

56.657.8

7.59.3

8.27.1

7.58.2

19.416.6

0.80.9

100100

EducationlevelcompletedNoPrimarySecondaryHigher

57.958.356.649.5

6.37.89.312.4

8.07.97.87.7

5.58.09.013.0

21.617.316.415.5

0.60.71.01.9

100100100100

PercapitaincomequintilesPoorestPoorerMiddleRicherRichest

55.756.356.458.058.3

7.17.08.48.69.3

8.58.78.27.16.8

6.67.17.68.29.2

21.319.718.517.615.5

0.71.10.90.60.9

100100100100100

WorkingpresentlyNoYes

57.556.0

7.39.2

8.27.4

8.56.6

17.520.0

0.90.7

100100

ReligionHinduMuslimOtherreligions

57.854.848.3

8.06.413.4

7.011.513.3

7.87.76.6

18.618.617.2

0.81.01.1

100100100

SocialgroupNonSC/ST/OBCOBCSCST

60.455.856.652.3

7.58.18.68.5

7.37.18.213.0

8.47.17.78.7

15.421.218.016.6

1.10.60.90.9

100100100100

OwnagriculturallandNoYes

56.857.0

9.36.7

7.68.2

8.07.4

17.419.8

0.90.8

100100

All 56.9 8.1 7.9 7.7 18.5 0.8 100

RegionCentralSouthernWesternNorthernEasternNortheasternAll

56.246.175.965.054.431.556.9

6.016.65.33.65.911.58.1

9.28.03.24.87.732.37.9

6.47.65.87.010.79.27.7

22.020.99.218.720.111.018.5

0.20.80.50.81.24.50.8

100100100100100100100

Amongthosewithnolivingson

RegionCentralSouthernWesternNorthernEasternNortheasternAll

18.38.538.829.418.99.719.5

33.056.723.217.624.036.934.4

9.65.64.87.09.819.68.1

29.521.626.029.532.621.327.2

8.55.24.612.59.54.37.6

1.02.52.64.05.28.23.3

100100100100100100

Amongthosewithatleastonelivingson

RegionCentralSouthernWesternNorthernEasternNortheasternAll

65.158.984.672.464.339.666.9

0.43.11.20.70.92.31.3

9.18.92.84.47.136.97.8

0.42.70.92.34.54.62.3

25.126.210.320.123.113.521.5

0.00.20.10.20.13.10.2

100100100100100100100

Source:ComputedfromIHDS-IIdatafiles.

38

Afactormentionedinthecontextofdaughteraversionistheperceivedharassmentofunmarrieddaughters intheneighborhood.TheIHDShadaskedaquestion:How frequently are unmarried girls harassed in your village/neighborhood?;theresponseswere: ‘Rarely’, ‘Sometimes’,and ‘Often’.Less than 10percentofwomenreportedthat thisoccursoften intheirvillageorneighbourhood.Thequalitativestudiescitedabovealsorevealedthatwhilecouplesareawareofsuchharassment,manysaidthat theyheardofsuch incidences throughthemediaoroccurrencesatotherplaces.Daughteravoidanceduetothisfactordoesnotappeartobeamajorfactor.

TheIHDSalsocollecteddataonmarriageexpensesincludinggiftsanddowryatthetimeofmarriageasprevalentinthecommunity.ItiscustomarytogivesomegoldintheformofJewellerytothebride.Dowryincashisalsogiventhoughthisislessprevalentinthenortheasternandnorthernregionsandtheamountsarerelativelyhigherinthesouthernregion.Overallweddingexpensesarehigherforthebride’sfamilybutthegroom’sfamilyalsoincursexpenditures;theratioofthemedianexpenditurebybrides’familytothatbythegroom’sfamilyisintherange1to2.

Ifdaughtersinheritparentalpropertytothesameextentassons,manywomenshouldbeownersofpropertysuchasahouseorland.IntheNFHS-4,womenwereaskedquestionsonownershipofhouseorland.Itisseenthatabouttwothirdsof themarriedwomen interviewed in the survey didnot own ahouse eitherindividuallyorjointly(intheNFHS-4thesequestionswereaskedonlytomarriedwomeninasub-sample,thestatemodule,ofthemainsample,andcovered121118women).Hardlyanydifferentialsby socioeconomic factorsare seen (Table 14)butthenorthernandwesternregionsshowlowerlevelsofownershipbywomencomparedtotheotherregions.Similarly,over70percentofwomendidnotownanylandindividuallyorjointly.Thelevelof‘non-ownership’oflandishigherinthewesternandnorthernregionsandlowerinthenortheasternregionthanaverage.Separatetabulationsfortheruralpopulation(forwhichownershipoflandmattersmorethanfortheurbanpopulation)giveafairlysimilarpicturethough,asexpected,thenon-ownershipoflandislowerintheruralpopulationthan urban, yet substantial, 69 percent. Given that inheritance laws entitledaughterstohaveashareinancestralproperty,ownershipofwomenisinsisteduponincertainhousingschemesandlandallotments,andfurtherthatthereareincentivesforfemaleownershipsuchaslowerstampdutiesinsomestates,theleveloffemaleownershipappearstobequitelow.

Fertility surveys generally ask direct questions on ideal number of childrenandnotethesebysex.ResultsonthisfromtheNFHS-4showamildpreferenceforsons;atthenationallevel,themeanidealnumberofsonswas1.1comparedto0.9daughters (IIPSandICF,2017:Table4.16,p.104).Further, 18.8percentofinterviewedwomenstatedmoresonsthandaughtersintheidealcombination;thisislowerthanthefigureintheNFHS-3,22.4percent,andshowssomedeclineovertimeinthedegreeofsonpreference.Butthepercentagewhomentionedmoresonsthandaughtersisrelativelyhigherforwomenwithnoschoolingandthelowestwealthindexquintilethanthosewithhigherlevelsofeducationandwealthquintilesrespectivelywhereasthedegreeofsexselection,asseenfromtheSRB,ishigherforthosewithhighereducationandwealthlevels.Further,mostof the states in thenorthernandwestern regionswhichhaveexhibitedhighsexselection inouranalysis showlowersonpreferenceandmanystatesintheeasternandnortheasternregionshowhighersonpreferenceinresponseto thequestionon idealnumberof sons anddaughters (seeTable 4.17 in IIPSandICF,2017).Ofcourse,onedoesnotexpectaperfectcorrespondencebetween

It is customary to give some gold in the form of jewelry to the bride. Dowry in cash is also given though this is less prevalent in the northeastern and northern regions and the amounts are relatively higher in the southern region

39

son preference and sex selection since thewillingness and ability to practisesexselectionalsomatter.Butitisquitelikelythatthereisatendencytogivenormativeandsociallyandpoliticallycorrectresponsestothequestiononidealnumberofchildrenbysexandhenceonemaynotseeagreementbetweentheresponses todirectquestionson sonpreferenceanddegreeof sex selectionasascertainedfromtheanalysisofdataonSRB.

The NFHS also askedmarried women whether they want anymore childrenandtheseresponsesaretabulatedintheNFHSreportsbythenumberoflivingchildrenaswellasbythenumberoflivingsons.ForIndia,amongwomenwithtwochildren,87percentofthosewithoneson(thatis,thosewithonesonandonedaughter)didnotwantanymorechildrenincontrastto62percentofthosewithnoson,thatis,thosewithonlytwodaughters(seeTable4.14inIIPSandICF,2017).Thefactthat62percentofwomenwithnosonwantedtostopchildbearingshowsthatamajoritydoesnotinsistonason,yetthedifferencebetweenthetwopercentagesisconspicuousandshowsthatmanydesiretohaveatleastoneson.Thegapbetweenthetwopercentagesisverywideinmanystatesinthenorthernregion,around50percentagepointsinHaryanaandRajasthanandaround40pointsinPunjabandUttarakhand(thefiguresaretakenfromTable17ofNFHS-4statereportsforlargestates;IIPSandICF,2017a).Thegapisalsowide,closeto40pointsinUttarPradeshandJharkhandandbetween30and40pointsinMadhyaPradesh,Bihar,Chhattisgarh,JammuandKashmir,andGujarat.Attheotherend,thegapisnarrow,below20pointsinallthesouthernstatesandinWestBengalandmoderate,20to30percent,inAssam,Maharashtra,Odisha,andHimachalPradesh.Thus,sonpreferenceseemstobequitestronginthenorthernandcentralstatesandsomeeasternandwesternstates.Notethatdesiretostopchildbearingafteraparticularsexcompositionmayleadtostoppingstrategies,ifthedesireistranslatedintopracticeofcontraception,butdoesnotamounttosexselection.Butthegeneralpatternofthelevelofsonpreferenceasascertainedfromthisinformationisquitesimilartothatseeninsexselection,thoughnotidentical.

40

Table 14:Table 14: Women’s ownership of house and land, by background characteristics, women ages 15-49, India, NFHS-4, State module

Categories of background characteristics

Owns a house individually or jointly (Percentage distribution)

Owns land individually or jointly(Percentage distribution)

Ruralpopulation

Does not own

Indivi-dually

Jointly Both indivi-dually and jointly

Does Not own

Indivi- dually only

Jointly only

Both indivi Dually and jointly

Percent of women not Owning any land

UrbanRural

66.161.0

10.610.6

13.916.1

9.412.3

77.168.6

5.97.5

10.213.4

6.810.4

-68.6

NoeducationPrimarySecondaryHigher

55.662.165.767.4

13.111.19.59.4

18.115.814.113.5

13.311.010.69.7

66.372.173.675.1

8.57.16.26.3

14.611.811.510.9

10.69.18.77.7

63.869.571.172.5

PoorestPoorerMiddleRicherRichest

56.961.163.665.365.5

11.210.211.210.610.1

17.516.515.014.214.0

14.412.310.29.810.5

66.068.971.774.675.3

7.66.77.86.66.2

14.113.812.211.310.5

12.210.68.47.58.0

65.368.169.671.072.0

DidnotworkWorked

63.361.8

9.213.9

15.714.2

11.810.1

71.272.8

6.38.4

12.711.1

9.77.7

67.970.1

HinduMuslimOtherreligions

62.265.964.7

11.18.110.1

15.415.313.7

11.310.811.5

71.075.173.6

7.35.26.8

12.511.610.6

9.38.19.1

68.271.070.6

NonSC/ST/OBCSCSTOBC

65.063.259.862.0

9.510.811.811.0

14.415.017.215.6

11.111.011.211.4

73.373.268.070.8

6.36.28.17.4

11.511.714.512.5

8.98.89.49.4

70.566.667.370.5

CentralSouthernWesternNorthernEasternNortheastern

64.960.069.772.254.948.4

5.118.08.85.013.17.6

15.416.111.811.517.626.5

14.65.99.711.414.317.4

73.170.779.178.463.458.0

3.311.94.33.29.56.0

11.812.99.39.114.821.9

11.94.57.39.312.314.1

70.566.877.760.855.570.5

MediaexposureNoorlowModerateHighAll

58.262.064.862.9

10.19.611.010.6

17.116.714.315.3

14.611.79.911.2

66.870.373.871.7

7.06.57.06.9

13.913.611.412.2

12.39.67.89.1

65.667.570.868.6

Source:ComputedfromNFHS-4datafiles.

Overall,sonscontinuetobemorevaluedthandaughtersforoldageresidenceandsupport.Thoughsomechangesinpreferencesandattitudesareseen,thesearenot large.Perhaps changes ineconomyand societymightbring ina shifttowardspost-maritalneo-localresidentialarrangementsandthenparentsmayhavelesshesitationinco-residingwithmarrieddaughters;butsuchchangesaregenerallyslow.Thelowleveloffemaleownershipoflandandhouseshowsthatinspiteoftheentitlementunderinheritancelaws,itisprimarilythesonsandnotdaughterswho inherit familyproperty. Thehigh level of sex selection inthewesternandnorthernregionsisconsistentwiththehighrelianceonandimplicitlyhighvalueattachedtosonsintheseregions.

41

The north-south dichotomy in kinship structure and female autonomy inIndiahas longbeenrecognisedandhasbeenlinkedtoregionaldifferences indemographicbehaviour(Karve,1965;DysonandMoore,1983).Itmakessensetoarguethatthedegreeofautonomywomenenjoywouldcorrespondtothevalueofgirls toa fairly largeextent.Wehaveseenabove that thesouthernregion,inwhichwomenareknowntohavegreaterautonomythaninnorthernIndia,haslowlevelsofsexselection.Butsodotheeasternandnortheasternregions.Further, the western region has shown a high level of sex selection. Thus,insteadofanorth-southdichotomy,weseeapatternwiththenorthernregion(ascategorizedintheNFHSandinthisstudythisisessentiallythenorthwesternpartofIndia)andthewesternregionsshowinghighdegreeofsexselectionandthesouthern,eastern,northeasternregionsshowinglowsexselectionwiththecentralregionfallinginthemiddle.

42

ThesexratioatbirthinIndiahasbeenmoremasculinethannaturalforsome timeand it iswell recognised that this is causedprimarilyby thepracticeofgenderbiasedsexselection.EstimatesoftheSRBareavailable

from a number of sources; civil registration, sample registration, surveys,censuses,andhealthmanagementinformationsystem.Theevidenceisclearonthe point that in the recent decades the SRBhas beenmuchmoremasculinethanthenaturallevel.However,theestimatesdifferandatthenationalleveltheSRBvariesinawiderangeof860to960excludingsomeoutliers.ThisstudyscannedvarioussourcesofdatatohaveaclearerideaofthevalueofSRBandinturnestimatethemagnitudeofgenderbiasedsexselection.Further,theeffectofpost-nataldiscriminationwasalsoassessed intermsofmissinggirlsduetoexcess femalechildhoodmortality.Ananalysisof surveydatawascarriedouttoseehowtheSRBvariesbythestageoffamilybuilding;mostoftheestimatespertaintotimeperiodsafter2000.Thestudyalsoexaminedsocioeconomicandspatialdifferentialsinthesexratioofbirth.Finally,factorsassociatedwithsonpreferenceandrecentchanges in the situationwerediscussedon thebasisofevidencefromsomerecentstudies.Theprincipalfindingsarenotedbelow.

1. 1. Onthebasisoftheassessmentofvariousestimates,itcanbesaidthatthecensus based indirect estimate obtained by reverse survival is the mostplausibleone.Atthenationallevel,thiswas923femalebirthsper1000malebirthsfortheperiod2004-2011.

2. 2. TheSRBinIndiaisclearlymoremasculinethanthenaturallevelbutnotashighassomeoftheestimatesindicate.TheSRSestimateoftheSRB(intermsoffemalesper1000males)seemstobeanunderestimatebyabout2percentatthenationallevelandneedstobecorrected;thecorrectionfactorvariessomewhatforstates.TheSRBhasbeenfluctuatingintherange900to930femalebirthsper1000malebirthssince2000forIndiawithnocleartrend.

3. 3. TheregionalpatternintheSRBiswellrecognized.Statesinthenorthern-westernregionshowmuchmoremasculineSRBthanintheotherregions;somestatesinthecentralregionalsoshowlowratiosbutnottothelevelsof thenorthern-westernregions.Theeastern,northeastern,andsouthernregionsgenerallyshowratiosnearnatural.InPunjab,JammuandKashmir,andHimachalPradeshtheSRBseemstohaverisenbutisstilllowerthanthenaturallevel.

4. 4. Thehighermasculinityisonaccountofthewide-scaleprevalenceofgenderbiasedsexselection.Closeto400thousandfemalebirthsaremissedinIndiaannually, amounting to about three percent of female births. The degree(number ofmissing female births as percent of female births) is high inmoststatesinthenorthernandwesternregions,moderateinUttarPradesh,HimachalPradeshandMadhyaPradesh,andlowornegligibleinmoststatesintheeasternandsouthernregions.

Principal Findings9

On the basis of the assessment of various estimates, it can be said that the census based indirect estimate obtained by reverse survival is the most plausible one. At the national level, this was 923 female births per 1000 male births for the period 2004-2011

43

5. 5. At the 2011 census enumeration, about fourmillion girls of ages 0-6maybeconsideredtohavebeenmissing;2.5milliononaccountofsexselection(pre-natal discrimination) and 1.5 million due to excess femalemortality(post-nataldiscrimination).Thissituationhaspersistedbeyond2011aswell.Further,whilepre-nataldiscriminationisconcentratedinthenorthernandwestern regions, post-natal discrimination is common across the country;thesouthernregionandafewotherstatesshowrelativelylowlevelsbuttheregionaldifferencesinpost-nataldiscriminationarenotaswideasseeninthecaseofpre-nataldiscrimination.

6. 6. AnalysisofSRBbybirthorderandbysexcompositionofpreviouschildrenshowsthatathigherordersandamongthosewhohavenoson, theratiosare veryhighlymasculine in thenorthern,western, and central regions.In thenorthern region, the SRBat thefirst order is alsomoremasculinethannaturalimplyingthatthereissomesexselectionatthefirstbirthitselfindicatingthatsomecouplesdesiretoavoidthebirthofevenonedaughter.Moreover, sex selection at the third birth following two daughters seemstobeverywidelyprevalent.This is in linewith thefindings fromJohn’sresearch(2018),thatmostfamiliesareincreasinglyaversetothepossibilityofbeingadaughter-onlyfamily.

7.7. SomedifferencesintheSRBbysocioeconomicbackgroundareseenespeciallyatthesecondandthirdbirths.Forthesecondbirthafterfirstdaughter,theSRBisgenerallymoremasculinethanaverageinthehighesteducationandwealthclasses.Atthethirdbirthfollowingtwodaughters,theSRBishighlymasculine;thisismoresointhemostrecentperiodof2010-14.Further,theSRBishighlymasculineatthehighestwealthandeducationlevels,inthenorthernandwesternregions,andathighmediaexposure,butnotamongMuslims.

8.8. Evidence on perceived values of sons vis-à-vis daughters shows that sonsare valued for old age support, financial as well as for residence; suchrelianceisrelativelyhigherinthenorthernandwesternregionscomparedtootherregions.Thoughsomechanges intheattitudesareseen inrecentinvestigations,thesearenotlargeenoughandparentsbyandlargecontinueto expect such support primarily from sons rather than from daughters.Besides,inspiteofthelegalentitlementsandprovisions,itisnotcommonfordaughterstoinheritparentalproperty.Sonpreference,clearly,persists.

Beforeclosing,itisnecessarytonotetwomajorconcerns.Inordertocurbthepracticeofgenderbiasedsexselection,effortshavebeenmadebyenactmentoflawsandcampaignsbythegovernmentandcivilsocietyorganisations.Financialincentiveshavealsobeenintroducedtodissuadecouplesfromresortingtosexselection (for a review, see Sekher, 2012). However, the persistence of highlymasculine SRB clearly shows that sex selection has persisted. Some changehasbeenseeninstateswithveryhighratios,Punjab,Haryana,andHimachalPradesh,buteveninthesestatestheSRBisfarfromthenaturallevelandtheSRBinthenorthernregionishighlymasculineevenatthefirstbirthorder.Clearly,themultitude of efforts at preventing sex selection, legalmeasures,financialincentives,andcampaignshasnotbeensuccessfulineradicatingthepracticeofgenderbiasedsexselectionsofar.Besides,inrecentyears,theSRBinsomestatesoutside the northern-western region has also becomemore masculine. GiventhatsonpreferenceiswidelyprevalentinIndia,thereisthepossibilityofthepracticeofsexselectionspreadingtoareaswhichhavehithertonotshownitonalargescale,oncetheavailabilityofsonographicscanfacilitiesandaffordabilityoftheservicesrise.

Analysis of SRB by birth order and by sex composition of previous children shows that at higher orders and among those who have no son, the ratios are very highly masculine in the northern, western, and central regions

44

Moreover,childhoodmortalityishigherforfemalesthanformalesindicatingthatneglectofthegirlchild,orpost-natalgenderdiscrimination,persists.WhilethematterofgenderbiasedsexselectionhasbeenreceivingmediaandpolicyattentioninIndia,andrightlyso,post-nataldiscriminationfindslittlespaceinthepublicdiscourse.Itisimperativethatcivilsocietyandpolicymakersaccorddueattentiontothisconcernaswellandadoptappropriatemeasurestoaddressit.

45

References

Arnold, Fred, Sunita Kishor, and T.K. Roy, 2002. “Sex Selective Abortions inIndia”,PopulationandDevelopmentReview28(4):759-85.

ArokiasamyP.andSrinivasGoli,2012.“ExplainingtheSkewedChildSexRatioinRuralIndia-Revisitingthelandholding-patriarchyhypothesis”,EconomicandPoliticalWeekly,47:85-94.

Bhalla, Surjit S., Ravinder Kaur, and Manoj Agrawal, 2013. Son Preference,Fertility Decline and the Future of the SexRatio at Birth,UNFPA,NewDelhi.

Bhat,P.N.Mari,2002.“OntheTrailofMissingIndianFemalesI:SearchforClues,II:IllusionandReality”,EconomicandPoliticalWeekly,37:5105-18,5244-63.

Bhat, P.N.Mari andA.J. Francis Zavier, 2007. “Factors Influencing theUse ofPrenatal Diagnostic Technique and the Sex Ratio at Birth in India”,EconomicandPoliticalWeekly42:2292-2303.

Bongaarts, John and Christophe Z. Guilmoto, 2015. “HowManyMoreMissingWomen?ExcessFemaleMortalityandPrenatalSexSelection,1970–2050”,PopulationandDevelopmentReview,41(2):241-269.

Coale, Ansley J. 1991. “Excess FemaleMortality and the Balance of Sexes: AnEstimateoftheNumberof‘MissingFemales’”,PopulationandDevelopmentReview17(1):35-51.

DasGupta,MonicaandP.N.Mari.Bhat,1997,“FertilityDeclineandIncreasedManifestationofSexBiasinIndia”,PopulationStudies,51,No.3:307-315.

DasGupta,Monica, Jiang Zhenghua, Li Bohua,Xie Zhenming,Woojin Chung,BaeHwa-Ok,2003.“WhyIsSonPreferenceSoPersistentinEastandSouthAsia?ACross-CountryStudyofChina, Indiaand theRepublicofKorea”,JournalofDevelopmentStudies40(2):153-187

Desai, Sonalde, Amaresh Dubey, B.L. Joshi, Mitali Sen, Abusaleh Shariff andReeveVanneman.2010.IndiaHumanDevelopmentinIndia:ChallengesforaSocietyinTransition.NewDelhi:OxfordUniversityPress.

Dubuc,SylviaandDavidColeman,2007.“AnIncreaseintheSexRatioofBirthsto India-bornMothers in England andWales: Evidence for Sex-SelectiveAbortions,”PopulationandDevelopmentReview33(2):383-400.

Dyson, T., & Moore, M. 1983. “On kinship structure, female autonomy, anddemographicbehaviourinIndia”.PopulationandDevelopmentReview,9,35-60.

GokhaleInstituteofPoliticsandEconomics,2017.ReversingSonPreference:AStudy of Contributory Factors: Tamil Nadu, Maharashtra, Punjab, DraftReports,GokhaleInstituteofPoliticsandEconomics,Pune.

46

George, Sabu, Rajaratnam Abel and B.D. Miller, 1992. “Female Infanticide inRuralSouthIndia”,EconomicandPoliticalWeekly,27:1153-1156.

Goodman,LeoA.,1961.”SomePossibleEffectsofBirthControlontheHumanSexRatio”,AnnalsofHumanGenetics,225.

GuilmotoCZ, 2008. “Economic, social and spatialdimensionsof India’s excesschildmasculinity,“Population-E.2008;63(1):91-118.

Guilmoto,ChristopheZ.2009.“TheSexRatioTransitioninAsia”,PopulationandDevelopmentReview,35(3):519-549.

Guilmoto,ChristopheZ.2012.“SkewedSexRatiosatBirthandFutureMarriageSqueezeinChinaandIndia,2005-2010”,Demography,49(1):77–100.

Guilmoto C. Z., 2015. “The masculinization of births: Overview and CurrentKnowledge,”,Population.70(2):201-264.

Guilmoto C. Z.; Dudwick N; Gjonça A; Rahm L, 2018. “How do demographictrendschange?theonsetofbirthmasculinizationinAlbania,GeorgiaandVietnam,1990-2005”PopulationandDevelopmentReview.44(1):37-61.

Guilmoto,C.Z. and I.Attane, 2007. “Thegeographyofdeteriorating child sexratio in China and India,” In: Watering the neighbour’s garden: Thegrowing demographic female deficit in Asia, edited by Isabelle Attaneand Christophe Z. Guilmoto. Paris, France, Committee for InternationalCooperationinNationalResearchinDemography[CICRED]:109-129.

Guilmoto, Christophe Z. and S. Irudaya Rajan, 2013. “Fertility at the DistrictLevelinIndia,”EconomicandPoliticalWeekly,48(23).

Guilmoto, Christophe Z, Nandita Saikia, Vandana Tamrakar, Jayanta KumarBora,2018,“Excessunder-5femalemortalityacrossIndia:aspatialanalysisusing2011censusdata”,LancetGlobHealth,6:e650–58.

ChristopheZ.Guilmoto,FengqingChao&PurushottamM.Kulkarni,2020.“Onthe estimation of female births missing due to prenatal sex selection”,PopulationStudies,74(2):283-289.

HealthManagementInformationSystem(HMIS),MinistryofHealthandFamilyWelfare, Government of India, 2018. www.nrhm-mis.nic.in accessed onDecember6,2018.

India Human Development Survey (IHDS), 2018. India Human DevelopmentSurvey.http://ihds.umd.eduaccessedonDecember6,2018.

InternationalInstituteforPopulationSciences(IIPS)andMacroInternational.2007.NationalFamilyHealthSurvey (NFHS-3),2005–06: India:VolumeI.Mumbai:IIPS.

International Institute for Population Sciences (IIPS) and ICF. 2017. NationalFamilyHealthSurvey(NFHS-4),2015-16:India.Mumbai:IIPS.

International Institute for PopulationSciences (IIPS) and ICF. 2017s.NationalFamilyHealthSurvey(NFHS-4),2015-16:India,Reportsforvariousstates.Mumbai:IIPS.

JamesK.S.,N.Kavitha,AnnieGeorge.P.M.Kulkarni,SardaPrasadandSanjayKumar,2015.AnAssessmentofQualityofCivilRegistrationSystemDatain

47

StatesofIndia,PRC,InstituteforSocialandEconomicChange,Bangalore,CSRD, Jawaharlal Nehru University, New Delhi, and United NationsPopulation Fund, New Delhi.

Jha, Prabhat, RajeshKumar, Priya Vasa,Neeraj Dhingra, Deva Thiruchelvam,RahimMoineddin,2006.”Lowmale-to-femalesexratioofchildrenborninIndia:nationalsurveyof1·1millionhouseholds,”Lancet,2006:367:211:18.

Jha, Prabhat, Maya A Kesler, Rajesh Kumar, Faujdar ram, Usha Ram, LukaszAleksandrowicz,DiegoG.Bassani,ShailajaChandraandJayantK.Banthia,2011.“TrendsinselectiveabortionsofgirlsinIndia:Analysisofnationallyrepresentativebirthhistoriesfrom1990to2005andcensusdatafrom1991to2011”,Lancet377:9781:1921-1928.

John,MaryE.,RavinderKaur,RajniPalriwala,SaraswatiRajuandAlpanaSagar,2008, Planning Families Planning Gender: The Adverse Child Sex Ratioin Selected Districts of Madhya Pradesh, Rajasthan, Himachal Pradesh,HaryanaandPunjab.BooksforChange:Bangalore.

John,Mary E., 2018. Political and Social Economy of Sex Selection- ExploringFamilyDevelopmentLinkages,ReportofastudyundertakenbytheCWDSforUNFPA,NewDelhi:UNFPAandUNWomen.https://india.unfpa.org/sites/default/files/pub-pdf/The%20Political%20Social%20Economy%20of%20Sex%20selection.pdf

Karve, Irawati, 1965. Kinship Organization in India, Asia Publishing House:Bombay.

Kashyap, Ridhi, 2019. “Is prenatal sex selection associatedwith lower femalechildmortality?”PopulationStudies,73(1):57-78.

Kaur,Ravinder,2004,“AcrossRegionMarriages-Poverty,FemaleMigrationandtheSexRatioinEconomicandPoliticalWeekly,Vol.XXXIXNo.25:2595-2603.

Kaur,Ravinder,SurjitBhalla,ManojK.AgarwalandPrasanthiRamakrishnan,2016.SexRatioatBirth:TheRoleofGender,ClassandEducation.Delhi,UNFPA.

Kaur,Ravinder,SurjitBhalla,ManojK.AgarwalandPrasanthiRamakrishnan,2017. Sex Ratio Imbalances and Marriage Squeeze in India: 2000–2050.Delhi, UNFPA.

Kaur,RavinderandTaanyaKapoor,2018.ReportonChangingParentalAttitudesandShiftingValueofDaughters,UNFPA.

Kulkarni,P.M.,2007.“EstimationofMissingGirlsatBirthandJuvenileAgesinIndia”,ApaperpreparedfortheUnitedNationsPopulationFund(UNFPA),India

Kulkarni P. M., 2009. “Tracking India’s Sex Ratio at Birth: Evidence of aTurnaround,”inPopulation,GenderandHealthinIndia-Methods,ProcessesandPolicies,K.S.James,ArvindPandey,DhananjayW.BansodandLekhaSubaiya(eds.),NewDelhi:AcademicFoundation:191-210.

Kulkarni P.M., 2012. “India’s Child Sex Ratio:Worsening Imbalance”, IndianJournalofMedicalEthicsIX(2):112-114.

48

Kumar,SanjayandK.M.Sathyanarayana,2012.“DecadalTrendsinDistrictLevelEstimatesofImpliedSexRatioatBirthinIndia,”DemographyIndia,41,No.1&2:.1-30

Leibenstein,Harvey,1974.“AnInterpretationoftheEconomicTheoryofFertility:PromisingPathorBlindAlley?”JournalofEconomicLiterature,12,No.2:457-479.

Premi, Mahendra K., 2001. “The Missing Girl Child”. Economic and PoliticalWeekly,36:1875-80.

Rajan,SIrudaya,SharadaSrinivasan,ArjunSBedi,2017.“UpdateonTrendsinSexRatioatBirthinIndia”,EconomicandPoliticalWeekly,LII(11):14-16.

RegistrarGeneral,2001.ProvisionalPopulationTotals,Paper1of2001,CensusofIndia2001,Series-1India.Delhi:ControllerofPublications.

Registrar General, various years: 2002-2019. Sample Registration SystemStatistical Report, various years: 1999…2017. Office of Registrar General,MinistryofHomeAffairs:NewDelhi.

RegistrarGeneral,2013a.VitalStatisticsinIndiabasedontheCivilRegistrationSystem2009.OfficeofRegistrarGeneral,MinistryofHomeAffairs:NewDelhi.

RegistrarGeneral, 2014a.ReportonPostEnumerationSurvey,Censusof India2011.OfficeofRegistrarGeneral,MinistryofHomeAffairs:NewDelhi.

RegistrarGeneral,2018a.VitalStatisticsinIndiabasedontheCivilRegistrationSystem2016.Office ofRegistrarGeneral,Ministry ofHomeAffairs:NewDelhi.

RegistrarGeneral,2019a.VitalStatisticsinIndiabasedontheCivilRegistrationSystem 2017. Office ofRegistrarGeneral,Ministry ofHomeAffairs:NewDelhi.

RegistrarGeneral,n.d.a).Datafrom2001Census:TableF1:Numberofwomenandevermarriedwomenbypresentage,parityandchildreneverbornbysex.

RegistrarGeneral,n.d.b)Datafrom2001Census:TableF9:Numberofwomenandcurrentlymarriedwomenbypresentage,numberofbirthslastyearbysexandbirthorder..

RegistrarGeneral,n.d.c)Datafrom2011Census:TableF1:Numberofwomenandevermarriedwomenbypresentage,parityandchildreneverbornbysex.

RegistrarGeneral,n.d.d)Datafrom2011Census:TableF9:Numberofwomenandcurrentlymarriedwomenbypresentage,numberofbirthslastyearbysexandbirthorder.

Registrar General, n.d. e) Data from 2011 Census: Table C-13: Single year agereturnsbyresidenceandsex.

Retherford, Robert D. and T.K. Roy, 2003. “Factors Affecting Sex-SelectiveAbortion in India and 17Majors States”,National FamilyHealth SurveySubject Reports, Number 21, January 2003, International Institute forPopulation Sciences, Mumbai, India and East-West Centre program onPopulation,Honolulu,Hawaii.U.S.A.

49

Sekher,T.V.2012.“LadlisandLakshmis:FinancialIncentiveSchemesfortheGirlChildinIndia”,EconomicandPoliticalWeekly,Vol.47(17)..

Sen,Amartya,1990.“Morethan100womenaremissing”,NewYorkReviewofBooks20:61-66.

United Nations, 2015. World Population Prospects: The 2015 Revision. U.N.PopulationDivision,NewYork;UnitedNations.

United Nations, 2017. World Population Prospects: The 2017 Revision. U.N.PopulationDivision,NewYork;UnitedNations.

Visaria, Leela, 2007. “Deficit ofGirls in India: Can It BeAttributed to FemaleSelectiveAbortion?” inTulsiPatel (ed.), Sex-SelectiveAbortion in India,NewDelhi:Sage.

Visaria,PravinM.1968.SexRatioofIndia’sPopulation,1961CensusMonograph,Delhi:ControllerofPublications.

Williamson, Nancy. 1976. Sons or Daughters? Beverley Hills, California: SagePublications.

50

Appendix

Estimation of the numbers of missing female births, missing girls, and excess female childhood mortality

Estimation of the number of missing female births during 2004-2011

Themethodologyfortheestimationofmissingfemalebirthsisstraightforward.First, female and male life tables for the seven-year period 2004-2010 wereconstructedbasedontheannualdataonage-specificdeathratesfromtheSRS(RegistrarGeneral,SampleRegistrationSystemStatisticalReport,variousyears)andaveraging therates for thesevenyears.Thenumbersof femaleandmalebirthsduringthisperiodwerethencomputedusingtheenumeratedpopulationsofages0-6inthe2011censusandapplyingreversesurvival.NotethatsincethecensusenumerationwasonMarch1,2011,ideallythedeathratesfortheperiodMarch2004-February2011areneededfortheconstructionoflifetableswhereastheSRSgivesratesforcalendaryearsandthusthelifetablefromtheSRSratesfor2004 to2010refers to the seven-yearperiod January2004-December2010.Butthedisplacementofonlytwomonthsisnotexpectedtochangethelifetablefunctionsmaterially since changes inmortality are generally gradual. If thefemaleandmalepopulationsintheagegroup0-6in2011censusenumerationaredenotedbyFP

0-6 (2011)andMP

0-6 (2011)respectively,and

7L

0 female and

7L

0 male , the

standardlifetablefunctionsforpersonyearsforfemalesandmalesrespectively,then thenumbers of female andmale birthsduring 2004-2011, denoted byFB(2004-2011)andMB(2004-2011)are

FB(2004-2011)=FP0-6 (2011)/(

7L

0 female//700000)and

MB(2004-2011)=MP0-6 (2011)/(

7L

0 male//700000)respectively. (1)

Populationsinages0-6bysexaretakenfromthedatasetfromRegistrarGeneral,Census of 2011, Table C-13.

TheSRB(expressedasnumberoffemalebirthsper1000malebirths)impliedbythecensusviareversesurvivalisthengivenas

SRB=1000xFB(2004-2011)/MB(2004-2011). (2)

Incidentally,thisissamemethodasthatemployedtoestimateSRBbyKumarandSathyanarayana(2012).

Now,iftheSRBhadbeennatural,thenumberfemalebirthsduringtheperiod,calledexpectedfemalebirthsanddenotedasEFB(2004-2011),isgivenby

EFB(2004-2011)=NSRB*MB(2004-2011)/1000 (3)

whereNSRBisthenaturalSRBexpressedasfemalebirthsper1000malebirths.Inthecalculationsinthisstudy,NSRBhasbeenassumedtobe952.

51

Theestimatednumberofmissingfemalebirthsduringtheperiod,denotedbyMiFB(2004-2011),isthencomputedas

MiFB(2004-2011)=EFB(2004-2011)-FB(2004-2011). (4)

Estimation of the number of missing girls at the 2011 census enumeration

Fortheestimationoftheeffectofexcessfemalemortalityintermsofmissinggirls at a point in time, it is necessary to first estimate ‘expected’ level offemale childhood mortality in the absence of post-natal discrimination. Inmost populations, female mortality is lower than male mortality and thefemalemortalitylevelcorrespondingtoagivenmalemortalitylevelinnormalsituations(thatis,intheabsenceofpost-nataldiscrimination)isthe‘expected’femalemortality.Inanearlierwork(Kulkarni,2007),thesystemofPrinceton(Coale-Demeny)WestModellifetableswasinvoked,andfemalemortalityatthesame level in themodel tables as the level ofmalemortalitywas acceptedastheexpectedleveloffemalemortality.ArecentpaperbyGuilmotoetal.(2018)hasprovidedaregressionrelationshipbetweenmaleunder-fivemortalityrate(U5MR)andfemaleU5MRonthebasisofdatafromcountriesthatdonotshowevidenceofpost-nataldiscrimination.Theequationis:

5qf

0 =Ax(

5qm

0 )2 + Bx (

5qm

0 ) + C. (5)

where5qf

0 f and

5qm

0 arefemaleandmaleU5MRs(expressedasdeathsbelowage5

per1000births)andA=0.0006,B=0.8013andC=-0.3462.

Using this regression equation, the value of expected U5MR for female wascomputedfromthemaleU5MR.Fromthis,theexpectedvalueoffemalepersonyearsbetweenages0and 7,denotedby

7L*

0 female , was obtained. Theexpected

femalepopulationofages0-6atthe2011census,denotedbyEFP0-6

(2011) was then computedas:

EFP 0-6 (2011)=EFB(2004-2011)x(

7L*

0 female /700000). (6)

Thisisthenumberofgirlsofages0-6whichwouldhavebeenpresentatthe2011censusenumeration intheabsenceofpre-nataldiscrimination(genderbiasedsexselection)andpost-nataldiscrimination(excessfemalechildhoodmortality).

Thetotalnumberofmissinggirlsofages0-6atthe2011censusissimply

Missinggirls(0-6)in2011=EFP0-6

(2011) - FP 0-6 (2011). (7)

Further,thenumberofmissinggirlsduetoexcessfemalechildhoodmortalityis

FB (2004-2011) x (7L*

0 female -

7L

0 female )/700000. (8).

Thenumberofgirlsmissingduetogenderbiasedsexselectionisthengivenbythedifferencebetweenthesetwoterms.Thiscanalsobeobtaineddirectlyas

MiFB(2004-10)x(7L*

0 female /700000). (9)

52

Estimation of the number of missing female births during 2011-2016

Forestimationofthenumbersofmissingfemalebirthsduringaspecifiedperiod,informationonthenumbersofbirthsbysexisessential.Fortheperiodbeforethecensus,thereversesurvivalmethodwasemployedforthispurposebutthisapproachcannotobviouslybeusedforperiodafter thecensus.Since thecivilregistration system does not provide a complete coverage, onemust resort tothe estimates of the crude birth rate from the SRS and in combinationwiththeSRBfromtheSRSandprojectedpopulationsize,estimate thenumbersofbirthsby sex.Normally, theOfficeofRegistrarGeneralof Indiapreparesandpublishespopulationprojectionssometimeaftereverycensusbutsofarnosuchprojectionshave been released after the 2011 census.Hence, projectionsmadeby the author independently have been used. Of course, making populationprojectionsinvolveassumptionsonfuture(thatis,forperiodsafter2011)levelsoffertility,mortality,andmigrationwith2011censusasthebaselineandthesewouldvary indifferentprojectionsbut for a shortperiod, suchvariationsdonotinfluencetheprojectedtotalsizeofpopulationsubstantially.Therefore,thepopulationsizeintheprojectionsbytheauthorcombinedwithcrudebirthratesfromtheSRSareusedheretoestimatethenumbersofbirthsover2011-2016andthen, by applying the SRB, thenumber ofmale and female births computed.Forthispurpose,populationsatmid-yearswereinterpolatedfromthe2011and2016 projected populations and the SRS crude birth rates applied to these tocomputethenumbersofbirthsineachyear.TheSRSestimatesoftheSRB,whichareavailableasthree-yearaverageswereusedforthemiddleyearandfurtheradjustedbythefactorshowninTable3(thisistheratioshowninthecolumnSRSEst./CensusbasedestimateoftheTableforIndiaandforindividualstates;since such ratiowasnot available forUttarakhand, the adjustment factor forIndiawasapplied).ForboththecrudebirthsratesandtheSRB,theestimatesareforcalendaryearsbutwereusedfortheperiodsMarch2011-February2012andsoonignoringthedisplacementoftwomonths.ApplyingtheadjustedSRBtothenumberofbirths,thenumbersoffemaleandmalebirthswerecomputed.Thestepsare:

Let Pi=Projectedpopulationatmid-yearforyeari,

wherei=1for2011-2012,2for2012-2013,3for2013-2014,4for2014-2015,5for2015-2016.

CBRi=crudebirthrateforyeari(per1000population),and

SRBi = SRB for year i, taken from the three-year average SRB for the

years,i-1,i,i+2.

Then,thenumberofbirthsinyeari,thenumberoffemalebirthsinyeari,andthenumberofmalebirthsinyeari,denotedbyB

i, FB

i,andMB

i respectively,are

givenby

Bi=P

i x CBR

i/1000,

FBi=B

ix(SRBi/A)/((SRB

i/A)+1000),and

MBi=B

i–FB

i, (10)

whereAistheadjustmentfactorforSRB(thisistheratioshowninthecolumnSRSEst/CensusbasedestimateofTable3).

Now,

53

FB(2011-16)= Numberoffemalebirthsduring2011-16=Σ FB

i , and

MB(2011-16)= Numberofmalebirthsduring2011-16=ΣMB

i ,

thesumbeingoveri=1,5. (11)

Then,Expectednumberoffemalebirthsduring2011-16,denotedbyEFB(2011-16),isgivenby

EFB(2011-16)=NSRB*MB(2011-16)/1000. (12)

Thenumber ofmissing female births during 2011-2016,MiFB (2011-16), is thengivenby

MiFB(2011-16)=EFB(2011-16)–FB(2011-16). (13)

Estimation of the number of excess female under-five deaths out of births during 2011-16

In order to estimate the impact of excess under-five female mortality, it isnecessarytofirstcomputefemaleandmaleunder-fivemortalityrates.Averagesofage-specificdeathsratesfromtheSRSfortheyears2011-2015wereobtainedbysexandlifetablesconstructedfortheperiodandfromthesethevaluesofU5MRformalesandfemalesweretaken.TheexpectedU5MRforfemalescorrespondingtomaleU5MRwascomputedusingtheregressionequationgivenbyGuilmotoetal(2018),reproducedaseq(5)above,andtheestimatednumberofexcessfemaledeathsunderage5obtainedas

FB(2011-16)x[FemaleU5MR–ExpectedFemaleU5MR]/1000. (14)

Ideally,U5MRfromcohortlifetablesshouldbeusedforthispurpose.However,changesinmortalityovertimeareveryslowwhenthelifeexpectancyiscloseto70yearsorhigherandhenceperiodlifetablesservethepurposequitewell.

54

App

endi

x Ta

bles

App

endi

x Ta

ble

1: S

ex R

atio

at

Birt

h fr

om t

he C

ivil

Reg

istr

atio

n Sy

stem

, Ind

ia, S

tate

s, a

nd U

nion

Ter

rito

ries

, 20

00

-20

16(f

emal

ebi

rths

per

10

00

mal

ebi

rths

bas

edo

nre

gist

ered

bir

ths)

Stat

e/ Y

ear

200

020

01

200

220

03

200

420

05

200

620

07

200

820

09

2010

2011

2012

2013

2014

2015

2016

IND

IA8

86

875

872

86

88

728

768

91

90

39

04

89

88

579

09

90

88

98

88

78

81

877

And

hra

Prad

esh

936

934

936

973

972

98

797

397

49

86

98

59

83

98

39

85

954

955

971

80

6

Aru

nach

alP

rade

sh72

88

87

80

68

128

64

799

792

854

875

80

679

28

978

1997

89

93

89

59

64

Ass

amna

nana

750

na73

28

768

348

539

3112

44

920

872

90

99

02

88

58

88

Biha

rna

98

477

98

268

01

793

nana

na71

5@

naN

a9

248

68

870

837

Chh

attis

garh

933

925

920

926

919

939

919

919

88

69

249

06

915

89

59

259

349

389

80

Del

hi8

208

09

831

823

823

822

831

84

810

04

915

90

18

93

88

68

95

89

68

98

90

2

Goa

918

919

929

931

931

931

96

39

47

931

921

90

89

349

299

46

94

79

289

00

Guj

arat

816

80

28

268

358

248

46

972

879

88

39

05

90

29

01

90

29

01

88

6na

na

Har

yana

792

789

829

814

796

827

857

86

08

548

538

388

338

328

40

84

38

518

65

Him

acha

lPra

desh

857

856

86

68

778

728

66

88

59

03

90

49

229

119

189

169

06

89

69

03

910

Jam

mu

&K

ashm

irα

879

916

923

925

934

923

90

69

229

06

918

917

913

Na

923

914

912

914

Jhar

khan

dna

nana

na8

40

830

835

86

58

788

64

86

5na

84

78

85

88

68

798

63

Kar

nata

ka9

5597

19

569

248

589

43

952

100

410

47

1011

1025

98

397

19

43

926

89

38

96

Ker

ala

99

89

68

94

49

519

46

94

99

559

44

952

94

29

399

399

559

42

94

89

48

954

Mad

hya

Prad

esh

86

98

41

823

84

18

64

855

855

90

59

139

389

228

979

129

04

90

89

04

90

9

Mah

aras

htra

84

78

198

268

08

811

nana

na8

708

68

854

86

18

94

90

19

118

83

90

4

Man

ipur

914

753

94

99

2310

04

1011

96

39

139

509

3477

08

1679

770

06

84

68

6na

Meg

hala

ya9

349

42

976

875

1011

918

96

58

64

959

94

89

129

42

94

797

89

68

975

na

Miz

oram

911

927

913

926

94

59

40

952

938

94

29

49

94

897

29

68

954

96

397

39

64

Nag

alan

d8

208

94

98

097

197

79

80

80

48

238

02

88

38

718

738

738

738

60

897

967

Odi

sha

94

79

399

88

950

927

934

927

919

935

925

911

90

28

96

88

68

80

86

68

58

Punj

ab74

775

477

778

979

479

08

138

208

208

228

248

528

44

876

88

08

91

857

Raj

asth

an8

218

258

06

833

820

80

58

288

3278

58

61

839

911

86

18

5979

979

48

06

Sikk

im97

39

3612

529

359

569

549

41

98

19

519

559

61

94

797

49

569

68

973

99

9

Tam

ilN

adu

922

929

926

920

936

935

939

935

928

933

935

90

59

04

853

834

818

84

0

55

App

endi

x Ta

ble

1: S

ex R

atio

at

Birt

h fr

om t

he C

ivil

Reg

istr

atio

n Sy

stem

, Ind

ia, S

tate

s, a

nd U

nion

Ter

rito

ries

, 20

00

-20

16(f

emal

ebi

rths

per

10

00

mal

ebi

rths

bas

edo

nre

gist

ered

bir

ths)

Stat

e/ Y

ear

200

020

01

200

220

03

200

420

05

200

620

07

200

820

09

2010

2011

2012

2013

2014

2015

2016

Tela

ngan

a(i

nclu

ded

inA

ndhr

aPr

ades

h)9

549

61

834

88

1

Trip

ura

931

914

1121

120

2na

nana

nana

na9

96

98

29

80

1055

88

210

00

917

Utt

arP

rade

shna

nana

nana

nana

nana

nana

na9

308

83

88

18

778

85

Utt

arak

hand

nana

nana

84

58

64

873

86

98

60

855

858

86

98

47

84

38

65

86

28

25

Wes

tBe

ngal

932

90

48

82

88

99

03

90

68

90

918

914

929

931

924

926

913

897

919

911

Uni

onT

errit

orie

s

A&

Nic

obar

94

69

68

927

100

39

359

679

339

49

96

197

39

349

549

349

47

1031

925

98

7

Cha

ndig

arh

751

771

827

80

778

68

268

60

86

08

728

678

788

358

87

90

48

708

98

90

1

Dad

ra&

NH

avel

i9

03

86

58

339

328

799

188

95

917

94

69

289

05

96

09

548

768

90

100

19

45

Dam

an&

Diu

897

86

08

769

159

06

86

98

47

90

89

169

349

128

578

86

96

19

169

2497

4

Laks

hadw

eep

90

79

08

96

89

91

84

38

69

89

99

119

348

66

1015

897

Na

96

910

43

89

19

45

Pudu

cher

ry9

42

94

69

379

42

936

950

955

950

923

927

917

912

90

99

109

119

399

31

Leve

lofc

over

age

ofb

irth

sun

der

regi

stra

tion

(per

cent

)

Indi

a56

.258

.059

.557

.76

0.4

62.

56

9.0

74.5

76.4

81.

38

2.0

83.

68

4.4

85.

68

8.8

88

.38

6.0

@Theva

lueofSRBfor

Bih

arin2010asgiven

intherepo

rtdoe

sis323

andisprimafaciein

correct.

α:I

ncludingLa

dakh

Source:Reg

istrar

Gen

eral(20

13a,2018a

)

56

App

endi

x Ta

ble

2: E

stim

ates

of

Sex

Rat

io a

t Bi

rth

for

Indi

a an

d La

rge

Stat

es F

rom

SR

S, 1

99

8 o

nwar

ds

(fem

ale

birt

hsp

er1

00

0m

ale

birt

hs)

Peri

od

199

8-

200

019

99

-20

01

200

0-

200

220

01-

200

320

02-

200

420

03-

200

520

04

-20

06

200

5-20

07

200

6-

200

820

07-

200

920

08

-20

1020

09

-20

1120

10-

2012

2011

-20

1320

12-

2014

2013

-20

1520

14-

2016

Mid

-yea

r19

99

200

020

01

200

220

03

200

420

05

200

620

07

200

820

09

2010

2011

2012

2013

2014

2015

IND

IA8

98

89

48

92

88

38

82

88

08

92

90

19

04

90

69

05

90

69

08

90

99

06

90

08

98

And

hra

Prad

eshβ

96

09

349

45

932

916

917

917

915

917

919

920

915

914

916

919

918

913

Ass

am9

84

96

29

45

90

49

159

07

920

939

933

937

928

926

922

920

918

90

08

96

Biha

r8

96

873

870

86

18

63

86

58

81

90

99

149

179

129

109

09

911

90

79

169

08

Chh

attis

garh

96

49

47

96

19

69

975

98

09

85

99

197

997

097

39

61

96

3

Del

hi8

358

318

47

871

877

88

28

84

88

08

84

88

78

768

69

857

Guj

arat

851

837

84

48

62

855

84

48

65

89

18

98

90

49

03

90

99

09

911

90

78

548

48

Har

yana

797

80

38

04

80

78

218

298

378

43

84

78

49

84

88

548

578

64

86

68

318

32

Him

acha

lPr.

90

28

588

268

03

851

858

872

931

938

94

49

42

938

939

94

39

389

249

17

Jam

mu.

&K

ashm

irα

816

824

838

854

86

28

708

738

80

89

59

02

89

98

99

90

6

Jhar

khan

d8

65

872

88

89

279

229

219

199

159

189

139

109

02

918

Kar

nata

ka9

42

935

952

94

39

239

159

179

269

359

44

94

39

45

950

958

950

939

935

Ker

ala

930

927

911

89

28

89

912

922

958

96

49

68

96

69

65

96

69

66

974

967

959

Mad

hya

Pr.

90

79

159

209

229

169

119

139

139

199

269

219

209

219

209

279

199

22

Mah

aras

htra

913

915

89

98

87

878

872

879

871

88

48

96

89

58

93

89

69

02

89

68

788

76

Oris

sa9

289

209

44

934

94

49

329

349

339

379

41

938

94

69

48

956

953

950

94

8

Punj

ab79

277

577

577

679

78

01

80

88

378

368

368

328

41

86

38

678

708

89

89

3

Raj

asth

an8

778

85

89

08

558

388

398

558

65

870

875

877

878

89

38

93

89

38

61

857

Tam

ilN

adu

931

926

926

953

94

69

43

955

94

49

369

299

279

269

289

279

219

119

15

Utt

arP

rade

sh8

68

870

86

48

538

598

62

874

88

18

778

748

708

758

748

788

69

879

88

2

Wes

tBe

ngal

952

956

94

99

379

319

269

319

369

41

94

49

389

41

94

49

43

952

951

937

Sour

ce:R

egis

trar

Gen

eral

(va

rious

yea

rs,2

00

1-20

18).

β:IncludingTe

langa

na;α

:IncludingLa

dakh

57

Appendix Table 3: Estimates of Sex Ratio at Birth for India and States, 2001 and 2011 Census Data (femalebirthsper1000malebirths)

2001 Census 2011 Census Indirect estimate from

Based on Based on Child sex ratio (0-6 years) #

CEB $ BLY @ CEB $ BLY @ 2001 census 2011 Census

Reference period 2000-01 2010-11 1994-2000 2004-10

INDIA 939 906 928 899 935 923

AndhraPradeshβ 961 951 929 924 959 938

Assam 964 948 963 930 968 961

Bihar 935 917 929 892 954 943

Chhattisgarh 975 928 965 948 983 971

Delhi 903 852 1001 869 865 873

Gujarat 896 834 904 868 890 895

Haryana 864 786 878 824 838 845

HimachalPradesh 919 845 969 948 898 916

Jammu&Kashmirα 915 951 888 774 na 866

Jharkhand 962 907 941 903 977 958

Karnataka 953 936 951 922 944 949

Kerala 964 969 966 977 959 965

MadhyaPradesh 942 903 944 908 941 923

Maharashtra 927 877 905 862 915 896

Orissa 966 928 941 910 951 942

Punjab 851 787 882 843 809 854

Rajasthan 918 864 903 899 924 897

TamilNadu 954 935 927 934 945 942

UttarPradesh 937 901 919 890 936 914

WestBengal 962 976 947 937 953 954

ArunachalPradesh 978 997 954 935 965 972

Goa 943 921 903 908 931 943

Manipur 969 976 929 905 957 937

Meghalaya 990 958 982 978 960 967

Mizoram 997 994 977 966 949 970

Nagaland 954 984 956 965 na 944

Sikkim 994 937 956 965 948 960

Tripura 974 973 963 956 964 958

Utttarakhand 938 853 931 869 928 895

$:CEB:Childreneverborntowomenofages20-29atcensus;@:BLY:Birthslastyear

#:Estimatedfromchildsexratioages0-6byKumarandSathyanarayana(2012)

β:includingTelangana;α:IncludingLadakh.na:Notavailable.

Source:CEBandBLYestimatescomputedfrom2001and2011Censusfertilitytables.

58

Appendix Table 4: Trends in Sex Ratio at Births based on NFHS-3 and NFHS-4(femalebirthsper1000malebirths)

Survey NFHS-3 NFHS-4

Period 1995-1999 2000-2004 2000-2004 2005-2009 2010-2014

Mid-year 1997 2002 2002 2007 2012

India 930 919 917 919 911

Jammu&Kashmirα 900 913 887 928 909

HimachalPradesh 938 898 923 943 928

Punjab 751 742 816 853 845

Uttarakhand 912 878 913 908 878

Haryana 910 723 777 810 846

Delhi 874 857 921 759 821

Rajasthan 931 897 886 881 869

UttarPradesh 931 915 893 917 896

Bihar 927 921 944 939 931

Sikkim 1000 929 904 955 800

Arunachal 974 1000 1008 915 911

Nagaland 896 922 924 957 953

Manipur 921 1017 931 959 964

Mizoram 962 1083 981 956 982

Tripura 966 976 973 952 941

Meghalaya 945 935 976 929 990

Assam 952 980 945 975 901

WestBengal 953 947 931 927 947

Jharkhand 1027 1063 949 942 930

Odisha 862 887 957 948 941

Chhattisgarh 969 890 922 923 913

MadhyaPradesh 993 1014 922 923 913

Gujarat 895 898 864 946 873

Maharashtra 944 865 935 869 911

AndhraPradeshincludingTelangana 909 872 946 951 888

AndhraPradesh na na 943 948 899

Telangana na na 951 954 874

Karnataka 878 971 970 918 909

Goa 893 929 977 867 945

Kerala 1005 934 972 941 1027

TamilNadu 933 992 909 955 950

Source:ComputedfromNFHS-3andNFHS-4datafiles.

na:Notavailable.

α:IncludingLadakh

59

Appendix Table 5: Sex Ratio at Birth from HMIS reports, India, States and Union Territories, 2008/09 to 2017/18 (femalebirthsper1000malebirths)

State/Year Year

2008-09

2009-10 2010-11 2011-12 2012-13 2013-14 2014-

152015-

162016-

172017-

18

India 900 927 913 917 915 918 918 923 926 929

A&NIslands na 956 923 971 954 959 967 890 1003 897

AndhraPradesh 915 946 938 943 931 926 921 951 946 958

ArunachalPradesh 995 911 942 918 931 921 916 951 936 956

Assam 927 885 903 924 917 928 920 922 936 938

Bihar 797 1129 935 946 932 941 936 928 918 910

Chandigarh 806 898 881 867 889 899 874 906 921 897

Chhattisgarh 969 968 964 930 920 923 930 931 946 961

Dadra&NagarHaveli 945 919 901 928 947 936 939 951 934 919

Daman&Diu na Na 1062 936 915 960 894 906 972 894

Delhi 900 871 882 891 888 893 901 904 908 917

Goa 865 919 923 963 934 905 939 918 937 942

Gujarat 904 904 894 890 891 900 901 907 910 910

Haryana 874 854 853 865 865 883 876 887 902 914

HimachalPradesh 886 898 880 896 894 894 897 908 916 931

Jammu&Kashmirα 924 891 909 921 913 937 936 942 947 958

Jharkhand 896 914 908 912 921 918 920 924 918 921

Karnataka 937 948 923 935 942 941 945 943 948 940

Kerala 914 957 960 950 955 952 959 953 958 964

Lakshadweep Na 1054 951 891 866 1021 1000 832 955 885

MadhyaPradesh 928 929 936 936 932 924 926 929 937 929

Maharashtra 886 881 870 889 910 921 920 924 922 940

Manipur 978 957 978 932 941 918 933 936 952 914

Meghalaya 1012 959 940 953 960 953 938 952 949 936

Mizoram 904 922 962 936 953 948 971 955 980 958

Nagaland 994 928 955 900 925 912 948 904 923 921

Odisha 992 936 927 924 932 940 948 943 940 936

Puducherry 912 914 924 903 927 898 916 948 931 939

Punjab 902 878 884 885 884 890 892 891 902 907

Rajasthan 905 901 893 900 906 924 929 929 938 945

Sikkim 1014 942 946 920 985 959 957 998 954 928

TamilNadu 950 947 943 933 924 923 917 935 938 947

Telangana IncludedinAndhraPradesh 925 947 941 925

Tripura 899 935 922 924 936 940 958 930 954 946

UttarPradesh 870 909 907 909 891 888 885 902 906 911

Uttarakhand 947 901 910 894 911 907 903 906 914 922

WestBengal 846 922 928 927 935 931 942 937 936 942

Source:HMIS(2018).

na:Notavailable.

α:IncludingLadakh

60

61

62

United Nations Population Fund (UNFPA)55 Lodi Estate, New Delhi 110003 IndiaE-mail:india.office@unfpa.org