The seen and unseen: the unintended impact of a conditional cash transfer program on prenatal sex selection

Ja adeka , Sayli; Saxena, K i ika
A icle — Published Ve sion
The seen and unseen: he unin ended impac o a
condi ional cash ans e p og am on p ena al sex
selec ion
Jou nal o Popula ion Economics
P o ided in Coope a ion wi h:
Sp inge Na u e
Sugges ed Ci a ion: Ja adeka , Sayli; Saxena, K i ika (2025) : The seen and unseen: he unin ended
impac o a condi ional cash ans e p og am on p ena al sex selec ion, Jou nal o Popula ion
Economics, ISSN 1432-1475, Sp inge , Be lin, Heidelbe g, Vol. 38, Iss. 1,
h ps://doi.o g/10.1007/s00148-025-01091-6
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/323177
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h p://c ea i ecommons.o g/licenses/by/4.0/
Vol.:(0123456789)
Jou nal o Popula ion Economics (2025) 38:32
h ps://doi.o g/10.1007/s00148-025-01091-6
ORIGINAL PAPER
The seen andunseen: heunin ended impac
o acondi ional cash ans e p og am onp ena al sex
selec ion
SayliJa adeka 1· K i ikaSaxena2
Recei ed: 12 June 2023 / Accep ed: 4 Feb ua y 2025 / Published online: 22 Feb ua y 2025
© The Au ho (s) 2025
Abs ac
This s udy examines he unin ended consequences o he Janani Su aksha Yojana,
a condi ional cash ans e p og am in India, on p ena al sex-selec i e beha iou
wi hin a son-p e e ence cul u e. This p og am unin en ionally al e ed exis ing ends
in p ena al sex selec ion h ough i s simul aneous p o ision o cash incen i es o
households and communi y heal h wo ke s as well as access o p ena al sex de ec-
ion echnology such as ul asound scans. Using di e ence-in-di e ences and iple
di e ence es ima o s we ind ha he p og am causes an inc ease in he likelihood o
emale bi hs. Fu he mo e, we obse e a ise in unde -5 mo ali y o gi ls bo n a
highe bi h o de s, sugges ing a shi in disc imina ion agains gi ls om p ena al o
pos na al. Ou calcula ions sugges ha he ne impac was app oxima ely 300,000
gi ls su i ing in ea ed s a es be ween 2006 and 2015. Finally, we ind sugges-
i e e idence ha he in ol emen o communi y heal h wo ke s in acili a ing he
p og am is a key d i e o his end. O e all, his s udy sheds ligh on he com-
plex in e play be ween policy in e en ions, cul u al no ms, and gende dispa i ies
in shaping demog aphic ou comes.
Keywo ds P ena al sex selec ion· Missing gi ls· Sex-selec i e abo ions·
Communi y heal h wo ke s· Janani Su aksha Yojana
JEL Classi ica ion J13· J16· J18
Responsible edi o : Kompal Sinha
* K i ika Saxena
k.saxena@ ug.nl
Sayli Ja adeka
ja adeka .say[email p o ec ed]
1 Though wo ks Gmbh, Munich, Ge many
2 Depa men o Economics, Econome ics andFinance, Uni e si y o G oningen, G oningen,
Ne he lands
S.Ja adeka , K.Saxena
32 Page 2 o 34
1 In oduc ion
The long his o y o son p e e ence in India has esul ed in nea ly 63 million women
missing om he coun y’s popula ion, wi h almos 2 million missing ac oss di e -
en age g oups e e y yea .1 This phenomenon o ‘missing women’ has he po en ial
o socioeconomic dis up ion, such as a ma iage ma ke squeeze (Heske h and Xing
2006), an inc ease in c ime a es (Edlund e  al. 2013), social s a i ica ion based
on gende (Edlund 1999), and ewe heal h and educa ional in es men s in women
(Jayachand an and Kuziemko 2011). The Indian go e nmen has in oduced a i-
ous schemes o educe disc imina ion agains women, including p o iding pa en s
wi h inancial incen i es o ha e daugh e s. Howe e , he e ec s o hese policies a e
ambiguous (Anuk i i 2018; Sekhe 2012; Sinha and Yoong 2009). A he same ime,
he li e a u e shows ha access o ul asound echnology inc eases he likelihood o
sex selec ion (Almond e al. 2013; Anuk i i e al. 2022a, b). This pape demons a es
how accessible ul asound echnology, along wi h inancial incen i es p o ided
unde a na ionwide sa e mo he hood p og am, in e ac s wi h he cul u e o son p e -
e ence o in luence he gende imbalance in India. We examine he causal ela ion-
ship be ween a sa e mo he hood p og am and sex-selec i e beha iou among Indian
pa en s and in es iga e he unde lying mechanism ha explains his ela ionship.
The sa e mo he hood p og am, known as Janani Su aksha Yojna (JSY), was
launched by he Indian Go e nmen in 2005 o educe ma e nal and neona al mo -
ali y. Mo he s we e gi en cash paymen s o e e y li e bi h in a heal h acili y.
The p og am also manda ed ha bene icia ies unde go a leas h ee an ena al check-
ups, including ul asound scans (a p ena al sex-de e mina ion echnology). To acili-
a e he p og am, he go e nmen ec ui ed heal h wo ke s. They ecei ed inancial
incen i es o e e y ins i u ional deli e y by egis e ed mo he s in hei neighbou -
hood. The scheme, hus, p o ided simul aneous access o p ena al sex de ec ion
echnology and cash incen i es o ins i u ional bi hs. The e o e, he p og am had
he po en ial o in luence sex-selec i e beha iou among Indian pa en s unin en ion-
ally. To es ima e he impac o he p og am on p ena al sex selec ion we use di -
e ence-in-di e ences (DID) and iple di e ence (DDD) es ima o s ha exploi he
a ia ion in he iming o p og am implemen a ion, eligibili y o bene icia y house-
holds based on hei socioeconomic s a us and geog aphic loca ion, and he na u al
expe imen c ea ed by sex o he i s bo n child.
Be o e he implemen a ion o he p og am, s a es in India we e ca ego ized as
low- o high-pe o ming based on hei s a e-speci ic ins i u ional deli e y a es. The
eligibili y c i e ia o p og am bene i s a ied by household socioeconomic cha ac-
e is ics ac oss his classi ica ion. All women esiding in low-pe o ming s a es we e
eligible o he p og am; howe e , only hose li ing below he po e y line (BPL)
and belonging o he Schedule Cas es o Schedule T ibes (SC/ST) in high-pe o m-
ing s a es could pa icipa e. Taking ad an age o his a ia ion in p og am access,
we compa e women li ing abo e he po e y line and no belonging o SC/ST om
1 Es ima es in Economic Su ey o India h p:// mo app. nic. in: 8080/ econo micsu ey/ pd / 102- 118_ Chap
e _ 07_ ENGLI SH_ Vol_ 01_ 2017- 18. pd
The seen andunseen: heunin ended impac o acondi ional… Page 3 o 34 32
low-pe o ming s a es wi h hei coun e pa s om he high-pe o ming s a es who
we e excluded om he p og am. In o he wo ds, non-BPL, non-SC/ST women om
low- and high-pe o ming s a es composed he ea men and con ol g oup, espec-
i ely.2 Fo ou analysis, we c ea ed a mo he -child panel using he epo ed e il-
i y his o y o mo he s om he Demog aphic and Heal h Su ey o India (DHS)
- 2015/16.
The iden i ica ion o ou es ima es is condi ional on he inclusion o mo he ixed
e ec s ha accoun o he sys ema ic di e ences in he cha ac e is ics o mo he s in
ou wo compa ison g oups. We include child yea o bi h ixed e ec s ha accoun
o he unobse ed ime- a ying ac o s ha may in luence he p opensi y o bi hs
o a mo he o e ime. We also include child bi h o de ixed e ec s o accoun
o he unobse ed he e ogenei y in he p opensi y o sex-selec dissimila ly o di -
e en bi h o de s. To ensu e he obus ness o ou es ima o o any ime- a ian
s a e-speci ic shocks and policies ha could con la e wi h he p og am’s impac , we
include s a e-yea ixed e ec s in ou DDD es ima ion. We measu e sex-selec i e
beha iou as he likelihood o emale bi h a e e y bi h o de o a mo he .
Ou DID es ima o shows ha he p og am inc eased he likelihood o emale
bi hs by 4.8 pe cen age poin s. The DDD es ima es suppo his inding. We show
ha he amilies in he ea men g oup wi h a i s bo n daugh e see an inc ease
in he likelihood o emale bi hs by 12.7 pe cen age poin s o bi h o de s 2 and
abo e. This is a no el esul conside ing he exis ing e idence on he g ea e p e -
alence o p ena al sex selec ion among he o wa d cas e, non-poo amilies, and
amilies wi h i s bo n daugh e s (Bo ke e al. 2017; Anuk i i 2018; Almond e al.
2019; Rosenblum 2013a, b). To es ablish he obus ness o ou esul s, we e i y he
iden i ying assump ion. Ou iden i ying assump ion is ha in he absence o he pol-
icy, he likelihood o emale bi hs e ol es in a pa allel manne in he ea men and
con ol g oups, condi ional on mo he ixed e ec s. To u he alida e ou empi i-
cal s a egy, we pe o m ou analysis on he da a collec ed be o e he launch o he
p og am and ind no p og am e ec s on mo he s who ne e ecei ed he p og am
bene i s. These alsi ica ion es s alida e ou iden i ica ion s a egy and bu ess ou
indings on he causal e ec o he policy.
Fu he mo e, we explo e wha hese esul s mean o he su i al and well-being
o hese addi ional gi ls. We ind sugges i e e idence ha al hough mo e gi ls we e
bo n in ea men households, hey we e mo e likely o die be o e eaching 5 yea s
o age. Su i ing gi ls a e likely o ha e poo e heal h and nu i ional ou comes,
inc easing he gende gap in well-being among child en. Al hough hese esul s
a e no causal e ec s o he p og am, hey p o ide addi ional insigh s in o e ili y
dynamics in India, pa icula ly among mo he s in he ea men s a es. Ou b oad
calcula ions indica e ha he ne e ec o he p og am on emale bi hs is an o e all
inc ease o 300,000 gi ls bo n in ea men s a es be ween 2006 and 2015.
2 Cas e g oups in India a e gi en a hie a chical classi ica ion: uppe / o wa d cas es, o he backwa d
cas es, schedule cas es, and schedule ibes. Non-SC/ST g oup includes uppe / o wa d cas es and o he
backwa d cas es. Uppe and o wa d cas e a e used in e changeably he e.
S.Ja adeka , K.Saxena
32 Page 4 o 34
How did JSY in luence he sex-selec i e p e e ences o Indian households? We
hypo hesize ha he p og am wo ked h ough ou possible channels. Fi s , by man-
da ing a leas h ee an ena al checkups, JSY inc eased access o ul asound echnol-
ogy among households who migh ha e had limi ed o no access. Pa en s wi h s ong
son p e e ences may use hese o sex selec ion by inducing abo ions o unwan ed
emale oe uses. Second, he inancial incen i es gi en o households o e e y li e
bi h lowe ed he cos o bea ing child en. This is a mo i a o o no ca y ou sex
selec ion and o gi e bi h o hei child, pa icula ly du ing pe iods o economic
shock. Thi d, heal h wo ke s’ emune a ion was linked o he numbe o bene icia -
ies egis e ed o he policy and hei deli e ies a heal h cen es. This is an incen i e
o heal h wo ke s o dissuade pa en s om pe o ming sex-selec i e abo ions and
o encou age hem o gi e bi h o hei emale child en. Finally, he heal h wo ke s
main ained a JSY ca d o ack e e y p egnancy in hei neighbou hood. Fe al sex
de e mina ion and sex-selec i e abo ions a e illegal in India. Hence, he egis a-
ion and moni o ing done by he heal h wo ke s could de e he households om sex
selec ing.3 JSY hus could in luence he willingness o pa en s o bea daugh e s by
c ea ing an unin en ional ade-o along hese di e en dimensions o he p og am.4
We use he Heal h Managemen In o ma ion Sys em da a ob ained om he Min-
is y o Women and Child De elopmen , Go . o India. Using his app oach, we
c ea ed a da ase o all heal h wo ke s a he dis ic le el om 2008 o 2015. We
ind sugges i e e idence ha nei he access o ul asound echnology no he inan-
cial incen i es gi en o pa en s explain he inc eased p opensi y o ha ing gi ls in
he ea men s a es. We ind desc ip i e e idence ha he inc ease is explained by
he p esence o heal h wo ke s. This esul has impo an policy implica ions. This
shows ha in e media y heal h wo ke s can play a i al ole no jus in deli e ing
heal h se ices bu also in os e ing desi able ou comes. Ano he key esul is he
shi o he disc imina o y beha iou di ec ed a gi ls om p ena al o pos na al as
a esponse o his policy. This is a e e sal o he p e ailing end whe e access o
ul asound echnology shi ed disc imina ion agains gi ls om pos na al o p ena al
(Bhalo a and Coch ane 2010; Bhaska 2007). Al hough his esul is no encou ag-
ing, i shows ha he e is scope o policy o achie e desi able e ili y ou comes
e en in he p esence o con lic ing cul u al belie s.
This pape con ibu es o he ex ensi e li e a u e on missing women. Se e al
exis ing pape s examine he e ec o ul asound echnology on sex a ios a bi h and
he ela i e well-being o emale child en (Chen e al. (2013); Anuk i i e al. (2022a,
b); Lin e al. (2014); Hu and Schlosse (2015); Almond e al. (2019); Bha adwaj and
Lakdawala (2013); Valen e (2014) Congdon Fo s and Lindskog (2023)). All o hese
s udies show ha he inc eased a ailabili y o e al gende iden i ica ion echnology
induces pa en s o abo unwan ed emale oe uses. The su i ing gi ls, he e o e,
3 P e-Concep ion and P e-Na al Diagnos ic Techniques (PCPNDT) Ac , 1994 is an Ac o he Pa liamen
o India enac ed o s op emale oe icides and a es he declining sex a io in India. The ac banned p e-
na al sex de e mina ion.
4 We es hese a ious mechanisms, howe e gi en he a ailabili y o da a and es ic ions on empi ical
es ima ions, we can only p o ide suppo i e e idence suppo ing o ejec ing hese mechanisms.

The seen andunseen: heunin ended impac o acondi ional… Page 5 o 34 32
a e wan ed and acqui e heal h in es men s. The s uc u e o he p og am c ea es a
adeo be ween access o echnology o p ena al sex selec ion and heal h wo ke s’
pe o mance-based bene i s and bene icia ies’ own cash ans e s o li e deli e -
ies a heal h cen es. We ill he gap in he li e a u e by documen ing he impac o
he p og am on sex-selec i e beha iou o he popula ion amongs whom he p ac-
ice is mos p e alen , he uppe -cas e and weal hy amilies in India (Bhalo a and
Coch ane 2010). We ind ha he p og am causes a decline in sex-selec i e abo -
ions in India.
The second con ibu ion o his pape is o he g owing li e a u e on he unin-
ended consequences o public policies and p og ams (Ebens ein 2010; Buchmann
e al. 2019). This li e a u e e alua es how policies can c ea e pe e se incen i es and
ha e an unin en ional impac on o he socioeconomic ou comes. JSY was imple-
men ed o educe ma e nal and neona al dea hs du ing deli e ies. The scheme did
no a ge imp o ing gende equali y a bi h. Exis ing li e a u e assessing he impac
o JSY has s udied i s impac on he up ake o ma e ni y se ices and ma e nal mo -
ali y (Powell-Jackson e al. 2015), e ili y (Nandi and Laxmina ayan 2016), ma e -
nal ca e (Sen e al. 2020), and he academic pe o mance o child en (Cha e jee and
Podda 2021). This pape is he i s o s udy he impac o JSY on sex selec ion, an
ou come i did no a ge , and unde s and he unde lying mechanism.
The inal con ibu ion o his pape is o he g owing li e a u e on he impo ance
o communi y heal h wo ke s, o achie e desi able ma e nal and child well-being
objec i es. Se e al s udies ha e documen ed he impac o inancial incen i es
gi en o communi y heal h wo ke s on a educ ion in child mo ali y and mo bidi y
(Cohen e al. 2013; Bjö kman Nyq is e al. 2019; Celhay e al. 2019; B enne e al.
2011; Singh and Mas e s 2017). We add o his g owing li e a u e by showing sug-
ges i e e idence o he con ibu ion o heal h wo ke s in he educ ion o p ena al
sex selec ion in he ea men s a es.
In e ms o me hodology, ou pape is closes o Anuk i i e al. (2022a, b), bu
ou pape di e s in h ee ways. Fi s , we s udy how simul aneous access o p ena al
sex de ec ion echnology and inancial incen i es o households and heal h wo k-
e s a ec p ena al sex-selec i e beha iou . The ade-o be ween hese dimensions
o he policy is an unin ended consequence o he in e en ion designed o ackle
low a es o ins i u ional deli e ies. This is he main analysis o ou pape . Second,
ou analysis ocuses on he p ena al sex-selec i e beha iou o he non-SC/ST and
non-poo g oups as opposed o hei wo k which s udies all he socioeconomic
g oups. Al hough ou indings a e o a speci ic socioeconomic g oup, exis ing e i-
dence shows ha p ena al sex-selec i e beha iou is mo e p ominen o his g oup.
Finally, we a emp o explain how policy mechanisms a ec p ena al sex-selec i e
beha iou . The mechanisms ha explain hese espec i e esul s a e dis inc . We ind
ha communi y heal h wo ke s played a p ominen ole in inc easing he likelihood
o emale bi hs. Thei pape inds he decline in desi ed e ili y and lowe bi h
spacing as he d i ing ac o s o he dec ease in he numbe o emale bi hs.
This pape is o ganized as ollows: The ‘Backg ound and Da a’ sec ion p o-
ides backg ound on son p e e ence in India and discusses he da a and desc ip-
i e s a is ics. The “Empi ical S a egy” sec ion in oduces he empi ical s a egy
used in he pape . The ‘4’ sec ion is a discussion o he esul s. The ‘Robus ness
S.Ja adeka , K.Saxena
32 Page 6 o 34
Tes s’ sec ion p esen s he obus ness es s. The ‘Discussion and Addi ional E i-
dence’ sec ion is a discussion on mo ali y and addi ional e idence. The ‘7’ sec-
ion discusses and es s a ious mechanisms ha explain he esul s and ‘8’ sec-
ion concludes he pape wi h some policy ecommenda ions.
2 Backg ound andda a
Disc imina ion agains young gi ls in India is well documen ed, wi h o mal eco ds
a ailable as a back as he Fi s Census o B i ish India in 1871-72 (Wa e ield
1875). Today his disc imina ion is e lec ed in skewed sex a ios a bi h and child
sex a ios. The na u al sex a io a bi h o humans is es ima ed o be be ween 104
and 106 boys pe 100 gi ls (Bhaska 2007; Ande son and Ray 2010); howe e in
India, he sex a io a bi h has inc eased om 108 boys pe 100 gi ls in 1991 o 111
boys pe 100 gi ls in 2011.5 This inc easing sho all in gi ls a bi h is p ima ily due
o he cul u e o son p e e ence. This sho all has also been documen ed in o he
Asian socie ies ha a e known o sha e India’s p e e ence o boys o e gi ls (Cla k
2000; Almond e al. 2019).
India has some eligious and cul u al no ms ha iew sons as asse s and daugh-
e s as liabili ies. Fo ins ance, in Hinduism, he dominan eligion in India, sons a e
expec ed o pe o m une al i es when hei pa en s die. In he absence o social
secu i y, olde pa en s ypically li e wi h hei sons, while hei daugh e s li e wi h
hei husbands’ amilies. Al hough daugh e s ha e a legal igh o an equal inhe i -
ance o he amily weal h, due o s icky social no ms a ound ma iage, households
p e e o keep weal h in he amily by bea ing a son ins ead o bequea hing asse s
o a daugh e who will e en ually mo e o ano he household (Bhalo a e al. 2020;
Roy 2015).6 Paying la ge dow ies o daugh e s (Bo ke e al. 2017) and sa e y con-
ce ns also make i mo e cos ly o pa en s o ha e a daugh e (Al ano 2017; Bo ke
2021; Anuk i i e al. 2022a, b). Fu he mo e, he e is some e idence ha sons ben-
e i om economic ad an ages in he labou ma ke ha daugh e s do no ecei e
(Rosenblum 2013a, b).
These no ms shape households’ e ili y p e e ences and a e in u n e lec ed in
he disc imina o y beha iou o households owa ds daugh e s be o e and a e hei
bi h. Pa en s adjus he gende composi ion o hei amily ia p ena al disc imi-
na ion and pos na al disc imina ion. Be o e ul asound echnology was a ailable
in India, pa en s ollowed a e ili y ule called he s opping ule, o ha ing chil-
d en un il hey eached hei desi ed numbe o boys. As a esul , gi ls we e bo n
in o la ge amilies wi h limi ed esou ces and he e o e ecei ed lowe in es men s
(Jensen 2012; A nold e al. 1998; Das Gup a and Ma i Bha 1997). This pos na-
al disc imina ion esul ed in wo se heal h ou comes and excess mo ali y among
5 The sex a io a bi h among many species including humans is biased owa ds males.
6 In 2005, Hindu Inhe i ance Ac was amended o allow women o inhe i weal h om hei pa en s. Ou
esul s s ay obus o his change. See Appendix C o de ails.
The seen andunseen: heunin ended impac o acondi ional… Page 7 o 34 32
young gi ls. Wi h he ad en o p ena al sex de e mina ion echnology, pa en s can
de e mine he sex o he oe us wi hin 7 weeks o p egnancy.7 This allowed pa -
en s o abo unwan ed emale oe uses (Chen e al. 2013; Bhalo a and Coch ane
2010). Easy access o ul asounds since he mid-1980s and an inc easing p e e ence
o smalle amilies ha e led households o change hei beha iou om pos na al
disc imina ion o p ena al disc imina ion (Goodkind 1996; Kashyap 2019).
A ea u e obse ed since he 1990s in India is ha he sex a io a bi h is highly
skewed owa ds males, pa icula ly a highe bi h o de s (Gella ly and Pe ie 2017;
Visa ia 2005; Das 1987; Na h 2023). Pa en s seldom sex-selec a he i s bi h since
hey p e e o ha e a child o ei he gende o e he possibili y o no ha ing a child.
Howe e , in he p esence o son p e e ence, pa en s whose i s bo n is a daugh e a e
mo e likely o ha e p ena al sex-selec i e abo ions om he second bi h onwa ds
han a e pa en s whose i s bo n is a son. Figu e1 plo s he sex a io a bi h om
2000 o 2016 a a ious bi h o de s. The ho izon al line a 106 is he e e ence line
o he na u al sex a io a bi h. The solid line plo s he sex a io a bi h o chil-
d en bo n a bi h o de one, i.e. he i s -bo n child en. This line closely ollows
he e e ence line indica ing a balanced sex a io o i s bo n child en. The dashed
line and he do ed line plo he sex a io a bi h o child en bo n a bi h o de wo
and bi h o de h ee o abo e, espec i ely. Bo h o hese lines di e ge inc easingly
Fig. 1 Sex a io a bi h, by bi h o de . Sex a io is measu ed as he numbe o males pe 100 emales
7 PNSDT o e al gende iden i ica ion echnology.
S.Ja adeka , K.Saxena
32 Page 8 o 34
om he e e ence line o he na u al sex a io, indica ing ha he sex a io a bi h
o child en bo n a highe bi h o de s is subs an ially dis o ed owa ds males. This
dis o ion a highe pa i y sugges s ha sex selec ion is mo e p e alen o p egnan-
cies a a highe o de . Al hough he sex a io imbalance o child en bo n a highe
bi h o de s is linked o p ena al sex de e mina ion echnology like ul asounds, he
li e a u e also discusses o he channels ha in luence sex-selec i e beha iou among
Indian households, such as he p ice o gold, dow y and ma iage con en ions and
he eligious iden i y o he poli ical leade (Bhalo a e al. 2018, 2020).
2.1 Janani Su aksha Yojna
In 2005, he Go e nmen o India launched Janani Su aksha Yojana, a condi ional
cash ans e p og am sponso ed 100% by he na ional Go e nmen wi h a dual
objec i e o educing he numbe o ma e nal and neona al dea hs na ionwide.8 This
scheme p omo ed sa e mo he hood by p o iding cash incen i es o women i hey
deli e ed hei child en ei he in go e nmen hospi als o in acc edi ed p i a e heal h
ins i u ions o a home unde medical supe ision.9 A u he condi ion o ecei e
he ull cash incen i e was ha he mo he should unde go a leas h ee p ena al
checkups ha include ul asound and amniocen esis, echnologies used o de e mine
e al sex. By manda ing an e-na al checkups, JSY enabled highe access and use o
ul asound echnology e en in a eas ha p e iously did no ha e access o i .
Eligibili y o he condi ional cash ans e was dependen on he place o esi-
dence, income le el, and he cas e o he household. The scheme, implemen ed
na ionwide in Ap il 2005, classi ied s a es as low- and high-pe o ming based on
he a es o ins i u ional deli e ies, i.e. he p opo ion o women who gi e bi h a
heal h cen es as shown in Figu e1. Low-pe o ming s a es we e s a es whe e he
ins i u ional deli e y a e was less han 25%. These included U a P adesh, U an-
chal, Biha , Jha khand, Madhya P adesh, Chha isga h, Assam, Rajas han, O issa,
and Jammu and Kashmi . The emaining s a es we e classi ied as high-pe o ming
s a es. The objec i e o his p og am was o educe ma e nal and child mo ali y
a es by inc easing he numbe o women who ga e bi h sa ely a heal h acili ies
(Joshi and Si a am 2014).
In low-pe o ming s a es, all p egnan women we e p og am bene icia ies and he
bene i s we e paid ega dless o whe he he women deli e ed in a go e nmen hos-
pi al o a p i a e acc edi ed heal h cen e and ega dless o he bi h o de o hei
child en. In high-pe o ming s a es, only women who we e classi ied as li ing below
he po e y line (BPL) o belonging o a scheduled cas e o scheduled ibe (SC/
ST) we e eligible o p og am bene i s. Eligibili y in hese s a es was es ic ed o
8 JSY is a modi ied g aded e sion o he Na ional Ma e ni y Bene i Scheme which uni o mly p o ided
all below po e y line women h oughou he coun y wi h Rs 500 pe li e bi h up o wo li e bi hs.
This Scheme was suspended a e JSY was launched. Since ou compa ison g oups do no comp ise
women below he po e y line, ou es ima es a e no a ec ed by he ea lie scheme.
9 This included go e nmen heal h cen es such as Sub cen es/P ima y Heal h cen es/Communi y
Heal h cen es/Fi s Re e al Uni s/gene al wa ds o he dis ic o s a e hospi als.
The seen andunseen: heunin ended impac o acondi ional… Page 15 o 34 32
o hese a gumen s suppo he case o na u al expe imen s c ea ed by he sex o he
i s bo n and sa is y he i s condi ion men ioned abo e.
Fu he , while ha ing a i s bo n gi l o boy does no inhe en ly con e addi ional
bene i s o accessing he JSY p og am, amilies wi h i s bo n gi ls may ha e an
incen i e o u ilize p ena al sex-selec ion echnologies o e ed h ough he p og am
o sex-selec ion (Akbulu -Yuksel and Rosenblum 2023; Anuk i i e  al. 2022a, b;
Bhalo a and Coch ane 2010). The e o e, he g oup wi h i s bo n gi ls he g oup
po en ially bene i s om he p og am in bo h LPS and HPS s a es compa ed o he
i s bo n boys g oup and hence sa is ies he second condi ion.
The iple di e ence es ima o is equi alen o aking he di e ence be ween wo
di e ence-in-di e ence es ima o s (Olden and Møen 2022). I i s akes he di e -
ence be ween i s gi l and i s boy amilies o e ime in ea men g oup and con ol
g oup sepa a ely and hen akes he di e ence i hese wo di e ences. Since he
i s gi l amilies and he i s boy amilies in he ea men and con ol s a es expe i-
ence he same s a e-speci ic ime- a ying e ec s. By doing his he iple di e ence
Table 2 Balance es
The baseline desc ip i e s a is ics a e o amilies wi h i s bo n gi l and i s bo n boy in u al a eas du -
ing p e-p og am yea s 2000–2005. In he las column, we ha e he coe icien s o he eg ession o he
espec i e a iable on he indica o Fi s Gi l. S anda d e o s clus e ed a he s a e le el. The a iable
sel - epo ed ul asound use has missing alues hence he numbe o obse a ions is di e en han he
es o he a iables
* p < 0.1; **p < 0.05; ***p < 0.01
Fi s gi l amilies Fi s boy amilies Di
NMean SD NMean SD
Hindu 35,539 0.76 0.43 38,284 0.76 0.43 − 0.002
Muslim 35,539 0.11 0.31 38,284 0.11 0.31 0.002
Fo wa d cas e 35,539 0.18 0.38 38,284 0.17 0.38 0.003
OBC 35,539 0.38 0.49 38,284 0.39 0.49 − 0.006**
Mo he ’s educa ion 35,539 4.47 4.59 38,284 4.49 4.58 − 0.017
Sex o household head 35,539 1.12 0.33 38,284 1.12 0.33 0.000
Age o household head 35,539 44.16 13.07 38,284 44.13 13.08 0.032
Sel - epo ed ul asound use 1771 0.29 0.45 1519 0.23 0.42 0.058***
Poo es 35,539 0.22 0.42 38,284 0.22 0.41 0.004
Poo e 35,539 0.21 0.40 38,284 0.21 0.41 − 0.001
Middle 35,539 0.20 0.40 38,284 0.20 0.40 0.004
Riche 35,539 0.19 0.39 38,284 0.19 0.39 − 0.003
Riches 35,539 0.18 0.39 38,284 0.19 0.39 − 0.004
Elec ici y 35,539 0.96 0.94 38,284 0.96 0.93 0.002
T uck 35,539 0.18 0.99 38,284 0.18 0.99 0.000
F idge 35,539 0.32 1.03 38,284 0.32 1.02 − 0.005
Cycle 35,539 0.66 1.03 38,284 0.70 1.02 − 0.040***
TV 35,539 0.66 1.03 38,284 0.67 1.02 − 0.004
Radio 35,539 0.22 1.01 38,284 0.22 1.00 0.002

S.Ja adeka , K.Saxena
32 Page 16 o 34
es ima o allows us o accoun o s a e-speci ic con ounding e ec s which we could
no in ou di e ence in di e ence es ima ion.
Like he DID es ima ion, he DDD es ima o also equi es a pa allel end
assump ion o he es ima ed e ec o ha e a causal in e p e a ion. Al hough DDD
is he di e ence be ween wo di e ence-in-di e ences (di e ence be ween i s
gi l and i s boy amilies o e ime and di e ence be ween ea men and con ol
g oups o e ime), i does no equi e wo pa allel ends assump ion (Olden and
Møen 2022). We discuss his iden i ying assump ion in he ‘Iden i ica ion Assump-
ion’ sec ion.
We un he ollowing iple di e ence speci ica ion whe e T ea × Pos in e ac s
wi h an indica o o i s gi l amilies gi en by Fi s Gi l. The iple di e ence speci-
ica ion es ima ed is as ollows:
The DDD coe icien β1 cap u es he di e ence in he likelihood o emale bi hs
be ween amilies wi h i s bo n daugh e s in he ea men and con ol g oup. We
include mo he ixed e ec s, bi h o de ixed e ec s, yea -o -bi h ixed e ec s, and
s a e-yea -o -bi h ixed e ec s. Fu he mo e, we es ima e he abo e DID and DDD
(2)
Gi l
bis =
𝛽
0+
𝛽
1
T ea
is ×
Pos
×
Fi s Gi l
i
+𝛽2Pos ×Fi s Gi li+𝛽3T ea is ×Pos
+𝛽4Fi s Gi li×T ea is +𝛽5T ea is +𝛽6Pos +𝛽7Fi s Gi l
i
+
S a eyea s
+𝛿
i
+𝜆
+𝜃
b
+
ebi s
Table 3 Main esul s: es ima ion esul s o di e ence-in-di e ences es ima ion
The able epo s he di e ence-in-di e ences es ima ion coe icien o he impac o he JSY on he like-
lihood o obse ing ha he child bo n is a gi l. T ea is he dummy a iable ha akes he alue 1 i he
mo he is om ou ea men g oup. Pos compa es pos -p og am yea s (2006–2015) o he p e-p og am
yea s (2000–2005). Pos 2006−10 and Pos 2011−15 a e he ea ly (2006–2010) and la e di usion (2011–2015)
pe iods o he p og am. The main FEs include mo he , bi h o de , and yea o bi h ixed e ec s as indi-
ca ed. Season FEs a e bi h mon h ixed e ec s and Season-Yea FEs a e mon h and yea -speci ic bi h
mon h ixed e ec s. All s anda d e o s a e clus e ed boo s apped (wi h 1000 eps) a he s a e le el and
epo ed in pa en heses
* p < 0.1; **p < 0.05; ***p < 0.01
(1) (2) (3) (4) (5)
T ea × Pos 0.048** 0.048** 0.048**
(0.023) (0.023) (0.023)
T ea × Pos 2006-10 0.041** 0.041*
(0.020) (0.021)
T ea × Pos 2011-15 0.086** 0.086**
(0.036) (0.038)
No. o Obs 150,757 150,757 150,757 150,757 150,757
Main FE Yes Yes Yes Yes Yes
Season FE No No Yes Yes No
Season and Yea FE No No No No Yes
The seen andunseen: heunin ended impac o acondi ional… Page 17 o 34 32
Table 4 Main esul s: Es ima ion esul s o iple di e ence es ima ion
The able epo s iple di e ence es ima ion coe icien o he impac o he JSY on he likelihood o obse ing he child bo n o i s gi l amilies is a gi l. Dependen
a iable is gi l. T ea is he dummy a iable ha akes he alue 1 i he mo he is om ou ea men g oup. Simila ly, Fi s Gi l is an indica o o i he woman’s i s bo n
child was a gi l. Pos compa es pos -p og am yea s (2006–2015) o he p e-p og am yea s (2000–2005). Pos 2006−10 and Pos 2011−15 a e he ea ly (2006–2010) and la e di -
usion (2011–2015) pe iods o he p og am. The main FEs include mo he , bi h o de , and yea o bi h ixed e ec s as indica ed. The s a e-yea end is he s a e speci ic
ime end, and he s a e yea FE is he S a e Yea speci ic ixed e ec . Season FEs a e bi h mon h ixed e ec s, and Season Yea FEs a e bi h mon h and yea -speci ic
ixed e ec s. All s anda d e o s a e clus e ed boo s apped (wi h 1000 eps) a he s a e le el and epo ed in pa en heses
* p < 0.1; **p < 0.05; ***p < 0.01
(1) (2) (3) (4) (5) Dep (6) (7) (8) (9)
T ea × Pos × Fi s Gi l 0.126** 0.105* 0.114** 0.127** 0.126**
(0.056) (0.058) (0.057) (0.055) (0.056)
T ea × Pos 2006-10 × Fi s Gi l 0.116** 0.097* 0.107* 0.116**
(0.056) (0.058) (0.056) (0.056)
T ea × Pos 2011-15 × Fi s Gi l 0.183*** 0.152** 0.163** 0.184***
(0.069) (0.069) (0.067) (0.070)
T ea × Pos − 0.011 0.000 − 0.080 − 0.012 − 0.013
(0.050) (0.000) (0.053) (0.051) (0.053)
Pos × Fi s Gi l − 0.181*** − 0.147*** − 0.157*** − 0.181*** − 0.180***
(0.053) (0.056) (0.055) (0.053) (0.054)
T ea × Pos 2006-10 − 0.017 0.000 − 0.063 − 0.018
(0.049) (0.000) (0.048) (0.048)
T ea × Pos 2011-15 0.014 0.000 − 0.079 0.012
(0.064) (0.000) (0.060) (0.064)
No. o Obs 63,250 63,250 63,204 63,204 63,250 63,250 63,232 63,232 63,250
Main FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
S a e Yea FE No No Yes Yes No No No No No
S a e Yea T end No No No No Yes Yes No No No
Season FE No No No No No No Yes Yes No
Season Yea FE No No No No No No No No Yes
S.Ja adeka , K.Saxena
32 Page 18 o 34
speci ica ions by classi ying he pos -JSY yea s in o he ea ly and la e di usion pe i-
ods. This is done o wo easons. Fi s , as addi ional ea u es we e added o JSY
in 2011, we can see how he impac changed o e he wo di usion pe iods. Sec-
ond, we ha e in o ma ion on he an h opome ic ou comes o child en bo n in he
la e di usion pe iod. By classi ying he e ec s in o di usion pe iods we can ie he
e ec o he p og am on he sex a io a bi h o his coho o hei a e age an h o-
pome ic wel a e ou comes.
4 Resul s
Table3 p esen s he esul s o he DID es ima ion, and Table4 shows he esul s o
ou iple di e ence es ima o . In he i s column o Table3, he pos -p og am yea s
2006 o 2015 a e compa ed wi h he p e-p og am yea s 2000 o 2005. In he second
column, he pos -p og am yea s a e di ided in o a la e di usion pe iod (2011–2015)
and an ea ly di usion pe iod (2006–2010) and compa ed o he e e ence p e-p o-
g am yea s. The key a iables o in e es a e T ea is × Pos , T ea is × Pos 2006-10, and
T ea is × Pos 2011-15.
Columns 1, 3, and 5 o Table 3 show ha he likelihood o a emale bi h
inc eased by 4.8 pe cen age poin s in he ea men g oup. This ansla es o a nea ly
10% inc ease in he numbe o gi ls bo n o mo he s in he ea men g oup. When
we look a he ea ly and la e di usion pe iods o he policy, we see ha in he ea ly
di usion pe iod, his likelihood inc eases by 4 pe cen age poin s while in he la e
pe iod inc eases by 8.6 pe cen age poin s. This esul is in e es ing because i shows
a educ ion in sex-selec i e beha iou among he g oups ha ha e been known in
he li e a u e o sex selec , i.e. non-SC/ST and non-BPL g oups.
The key coe icien s o in e es a e he iple di e ence es ima o s. Simila o
he DID speci ica ion, we i s look a he pos -policy pe iod om 2006 o 2015 in
Table4 in columns 1,3,5,7, and 9 and hen we di e en ia e be ween he ea ly and
la e di usion pe iods in columns 2,4,6, and 8 o Table4. We see ha he p og am
led o an inc ease in he likelihood o emale bi hs om bi h o de 2 onwa ds o
amilies wi h a i s bo n emale child in he ea men g oup by 12.6 pe cen age
poin s. The e was an inc ease o almos 18.3 pe cen age poin s in he la e di usion
pe iod and 11.6 pe cen age poin s in he ea ly di usion pe iod (column 2). We add
s a e-yea ixed e ec s and s a e-yea ends o ou speci ica ions. A e including
s a e-yea ixed e ec s his es ima e educes o 10.5 pe cen age poin s (column 3)
wi h inc eases o 9.7 and 15.2 pe cen age poin s in he likelihood o emale bi hs
in he ea lie and la e di usion pe iods (column 4). This is a mo e conse a i e
speci ica ion as i con ols o s a e-speci ic ime- a ying con ounde s. This sugges s
ha o amilies wi h i s -bo n daugh e s in he ea men g oup, he inc ease in he
numbe o gi ls a e 2005 was nea ly 23%, compa ed o amilies wi h a i s -bo n
boy. In columns 7 and 8 we include mon h o bi h ixed e ec and in column 9 we
addi ionally include bi h mon h-yea speci ic ixed e ec s o accoun o seasonali y
o bi hs (Boland e al. 2020; K ombholz 2023).
The seen andunseen: heunin ended impac o acondi ional… Page 19 o 34 32
Though ou iple di e ence es ima e sugges s inc eased likelihood o bi h o
gi ls, he coe icien on Pos × Fi s Gi l in all ou speci ica ions shows ha in he
con ol g oup i s gi l amilies a e he yea 2005 we e signi ican ly less likely o
ha e second bi h o a gi l compa ed o i s boy amilies be o e 2005. O e all, ou
esul s sugges ha an unin en ional impac o he p og am is he educ ion in sex
selec i e abo ions and an inc ease in he p obabili y o gi ls being bo n, in amilies
eligible o ea men . We also see ha mos o he posi i e esul s a e d i en by he
la ge impac s in he la e di usion pe iods.
5 Robus ness es s
5.1 Iden i ica ion assump ion
A key assump ion o a DID es ima ion is ha in he absence o he p og am, he
ou come a iable in he ea men and con ol g oups has pa allel ends, i.e. he ou -
come a iable would ha e e ol ed in he same way o bo h g oups. Fo alidi y o
ou analysis, he p obabili y o ha ing a gi l a he nex bi h should no be signi i-
can ly di e en ac oss mo he s in he ea men and con ol g oups du ing he p e-
p og am yea s. To es his we un a speci ica ion whe e he e ec o he p og am
is allowed o a y by yea , as in an e en s udy analysis. This app oach is ecom-
mended and widely used o de ec ing p e ends (Ro h e al. 2023). Fo us o be con-
iden ha he p og am had a causal impac on he sex-selec i e beha iou o mo h-
e s, we should no obse e any signi ican di e ences in he p obabili y o ha ing a
gi l in he compa ison g oups p io o he p og am. Signi ican di e ences, i any,
should only occu a e he p og am i he p og am has any e ec on sex-selec i e
abo ions. To check his, we es ima e he ollowing speci ica ion o a DID and a
DDD:
Fig. 2 Tes o pa allel ends. a Plo s es ima ed di e ence in he likelihood o gi l bi hs be ween ea -
men and con ol g oups, condi ional on mo he ixed e ec s. b Plo s he es ima ed di e ences in he
likelihood o gi l bi hs o mo he s wi h i s gi ls in ea men g oups wi h hei coun e pa s in he con-
ol g oup. The dashed ed line ep esen s he yea o he JSY p og am. The join es o he signi icance
o he lead yea s o he p og am yielded a F-S a is ic o 1.26 and 1.74 espec i ely, implying ha be o e
he p og am, he numbe o gi l bi hs e ol ed simila ly in he wo g oups
S.Ja adeka , K.Saxena
32 Page 20 o 34
Figu e2a shows he likelihood o a gi l being bo n o a mo he in he ea men
o con ol g oups is no signi ican ly di e en o yea s p io o 2005. Simila ly,
Figu e2b shows he likelihood o gi ing bi h o a gi l is no signi ican ly di e -
en o i s -gi l amilies be ween he ea men and con ol g oups. Condi ional on
mo he ixed e ec s, we ind no signi ican di e ences in he likelihood o bi h o
gi ls be ween he ea men and con ol g oups be o e p og am implemen a ion in
2005. The di e ences in ou comes become signi ican only a e 2009. The join es
o he signi icance o he lead yea s o he p og am yielded p- alue o 0.3 and 0.163,
implying ha p io o he p og am, he numbe o gi l bi hs e ol ed simila ly in he
wo g oups.
While es ing o p e ends alone is no su icien we also suppo ou esul s wi h
he placebo es s discussed in he nex subsec ion.
Ano he ac o ha could bias ou esul s and h ea en ou iden i ica ion s a egy
is whe he he implemen a ion o he JSY was an icipa ed be o e 2005. We would
hen be con la ing ou es ima e wi h he households’ expec a ions. I his we e he
case, hen households in he ea men g oup should ha e changed hei e ili y
beha iou p io o 2005 and we should see a dec ease in emale bi hs. Howe e ,
i households in he ea men g oup did no change hei beha iou p io o 2005
di e en ly han did households in he con ol g oup, i.e. he p obabili y o emale
bi hs was simila in bo h g oups p io o 2005, we can say ha households did no
an icipa e he p og amme and ha he yea o implemen a ion was exogenous. In
Figu e2, we show ha he di e ence in he numbe o emale bi hs p io o 2005
(3)
Gi l
bis =𝛽0+
∑2015
j
=
2000
𝛽jT ea is ×Yea j+𝛿i+𝜆 +𝜃b+e
bis
(4)
Gi l
bis =𝛽0+
2015
∑
j
=
2000
𝛽jT ea is ×Yea j×Fi s Gi li+𝜙s +𝛿i+𝜆 +𝜃b+e
bis
Fig. 3 Falsi ica ion es s using DHS—2015/16. This igu e epo s coe icien s o he di e ence-in-di e -
ence analysis assuming yea s om 1990 o 2004 as p og am yea s and checks i he likelihood o emale
bi h is di e en ac oss he ea men and con ol g oup. All eg essions con ain mo he , bi h, and yea
ixed e ec s. S anda d e o s a e clus e ed a he s a e le el. The F-s a is ics o he p e- ea men yea s
join ly is 1.04

The seen andunseen: heunin ended impac o acondi ional… Page 21 o 34 32
was no signi ican , indica ing ha households did no an icipa e he p og am and
change hei e ili y beha iou .
5.2 Falsi ica ion Tes s
I ou empi ical s a egy iden i ies he causal impac o he p og am on he e ili y
decisions o mo he s, hen we should no be able o see any e ec on mo he s who
ne e ecei ed he p og am. Ou i s alsi ica ion es in ol es indi idual assump-
ions o each yea om 1990 o 2004, i.e. yea s p io o 2005, o be he p og am
yea . Assuming ha each o he yea s was he yea when he JSY was implemen ed,
we checked he impac o he p og am on he likelihood o gi l bi hs ac oss ea -
men and con ol mo he s. Figu e3 plo s he coe icien o each yea and we can
see ha he p obabili y o gi l bi h ac oss ea men and he con ol g oups is no
signi ican ly di e en o any o he yea s excep 1996 and 1997. The signi ican di -
e ence in hese 2 yea s could be due o he s uc u al b eak in 1995 when ul asound
echnology became widely a ailable in India (Bhalo a and Coch ane 2010). How-
e e , he e ec o his s uc u al b eak did no las long and dissipa ed a e 1997 as
can be seen in Figu e3. The coe icien s o he emaining yea s a e no signi ican
and he di e ences in he ou come only appea a e 2005, i.e. a e he JSY was
implemen ed, sugges ing ha wha we a e cap u ing is he causal e ec o he JSY.
The second alsi ica ion es is o un ou iple DID speci ica ion on DHS-
2005/06. Since his su ey was comple ed by 2005–2006, he women in e iewed
in his sample ne e ecei ed he p og am. This idea is simila o he es abo e.
We should no ind any e ec o he p og am on women who ne e ecei ed he
p og am. He e, we assume 1995 as he yea he JSY was implemen ed and com-
pa e child en bo n up o 10 yea s a e 1995 wi h child en bo n up o 5 yea s p io
o 1995. Ou sample consis s o mo he s who made hei e ili y decisions om
1990 onwa ds, since we assume 1995 o be he yea ha he p og am was olled
ou . We compa e child en bo n be ween 1996 and 2000 (ou assumed ea ly di u-
sion pe iod) and be ween 2001 and 2005 (ou la e di usion pe iod) wi h hose bo n
be ween 1990 and 1995. One eason o his is ha i he e a e any epo ing biases
in e ili y o child en bo n mo e han 10 yea s p io o he su ey yea , hen hese
biases should be he same in any DHS sample. Hence, i ou main esul s a e d i en
by epo ing bias, hen we will also see signi ican di e ences in he ou come o ou
DHS-III es ima ion esul s.
We es ima e he ollowing speci ica ion o DHS-III:
Table5 shows he esul s o ou alsi ica ion es on mo he s whose e ili y deci-
sions we e made in 1990.15 In bo h columns, we see ha he likelihood o gi ing
(5)
Gi l
bi s =
𝛽
0+
𝛽
1
T ea
is ×
Pos
1996−00, ×
Fi s Gi l
i+
𝛽
2
T ea
is ×
Pos
2001−05, ×
Fi s Gi li
+𝜙
s
+𝛿
i
+𝜆
+𝜃
b
+e
bi s
15 An addi ional alsi ica ion es assuming he yea 2000 o be he ea men yea o he DHS III sample
is shown in he Appendix.
S.Ja adeka , K.Saxena
32 Page 22 o 34
bi h o a gi l is no signi ican ly di e en o amilies whose i s child was a gi l
ac oss he ea men and he con ol g oups. A lack o signi icance will indica e ha
ou empi ical s a egy is o cap u e only he p og am e ec .
Bo h alsi ica ion es s suppo ou claim o causal iden i ica ion o he p og am
e ec on he likelihood o gi l bi hs wi h he empi ical s a egy we employ.
6 Discussion andaddi ional e idence
The p e ious sec ion desc ibed he causal impac o he JSY on sex-selec i e abo -
ions in India. The p og am caused an inc ease in he numbe o gi ls bo n o ami-
lies eligible o ecei e he JSY bene i s, indica ing ha he mechanism o access o
p ena al sex de e mina ion echnologies was no dominan . P e ious wo k has shown
ha in socie ies wi h a p e e ence o male child en, gi ls su e om lowe wel a e
in amilies ha ollow he s opping ule and ha e mo e gi ls han hey desi e. This
disc imina ion is s a ke o gi ls a highe bi h o de s. In his sec ion, we he e o e
es he hypo hesis ha gi ls bo n in amilies wi h son p e e ence will be wo se o .
Table 5 Falsi ica ion es : iple
di e ence es ima ion using DHS
2005–06
The able epo s he iple di e ence esul s o he alsi ica ion
es s using DHS 2005–06 da a collec ed p io o he implemen a ion
o he p og am. In columns (1) and (2), we assume 1995 o be he
yea o p og am implemen a ion. In column (1), we conside yea s
1996–2000 and yea s 2001–2005 as ea ly and la e di usion pe iods.
These a e compa ed o he p e-p og am pe iod 1990–1995. In col-
umn (2), we assume yea s 1996–2005 as pos -p og am yea s. In col-
umn (3), we assume 2000 as he yea o p og am implemen a ion.
Pos -p og am yea s 2001–2005 a e compa ed o p e-p og am yea s
1996–2000. T ea is he dummy a iable ha akes he alue 1 i he
mo he is om he ea men g oup. Simila ly, Fi s Gi l indica es i
he woman’s i s bo n child was a gi l. FE con ains mo he , bi h,
and yea ixed e ec s. All iple di e ence es ima es a e o child en
a pa i y 2 onwa d. Main FE con ains mo he , bi h, and yea ixed
e ec s. S anda d e o s in pa en hesis a e clus e ed a he s a e le el
* p < 0.1; **p < 0.05; ***p < 0.01
(1) (2) (3)
T ea × Pos 1996-00 × Fi s Gi l − 0.036
(0.64)
T ea × Pos 2001-05 × Fi s Gi l − 0.092
(0.82)
T ea × Pos 1995-05 × Fi s Gi l − 0.048
(0.67)
T ea × Pos 2001-05 × Fi s Gi l − 0.052
(0.04)
Main FE Yes Yes Yes
No. o Obs 15,524 15,524 11,987
The seen andunseen: heunin ended impac o acondi ional… Page 23 o 34 32
Table 6 Es ima ion esul s o mo ali y o child en unde 1yea
The able epo s he mo ali y ou comes o child en below age 1. Columns 1 and 2 eco d he likeli-
hood o gi ls dying be o e eaching age 1. Columns 3 and 4 eco d he likelihood o gi ls bo n a pa i y
2 and abo e dying be o e eaching age 1. Columns 5 and 6 eco d he likelihood o gi ls bo n a pa i y
3 and abo e dying be o e eaching age 1. T ea ha akes he alue 1 i he mo he is om ou ea men
g oup. Pos compa es pos -p og am yea s (2006–2015) o he p e-p og am yea s (2000–2005). Pos 2006-10
and Pos 2011-15 a e he ea ly (2006–2010) and la e di usion (2011–2015) pe iods o he p og am. The
main FEs include mo he , bi h, and yea ixed e ec s. S anda d e o s in pa en heses a e clus e ed boo -
s apped (1000 eps) a he s a e le el
* p < 0.1; **p < 0.05; ***p < 0.01
All Bi hs Pa i y > 1 Pa i y > 2
(1) (2) (3) (4) (5) (6)
T ea × Pos × Gi l 0.007 0.020 0.042
(0.007) (0.013) (0.030)
T ea × Pos 2006-10 × Gi l 0.009 0.025** 0.071**
(0.007) (0.012) (0.033)
T ea × Pos 2011-15 × Gi l 0.005 0.017 0.009
(0.008) (0.018) (0.038)
No. o Obs 150,757 150,757 63,250 63,250 23,275 23,275
Main FE Yes Yes Yes Yes Yes Yes
Table 7 Es ima ion esul s o mo ali y o child en unde 5yea
The able epo s he likelihood o a gi l in ea men g oup dying be o e she eaches age 5. Columns 3
and 4 eco d he likelihood o gi ls bo n a pa i y 2 and abo e dying be o e eaching age 5. Columns 5
and 6 eco d he likelihood o gi ls bo n a pa i y 3 and abo e dying be o e eaching age 5. T ea ha
akes he alue 1 i he mo he is om ou ea men g oup. Pos compa es pos -p og am yea s (2006–
2015) o he p e-p og am yea s (2000–2005). Pos 2006-10 and Pos 2011-15 a e he ea ly (2006–2010) and
la e di usion (2011–2015) pe iods o he p og am. The main FEs include mo he , bi h, and yea ixed
e ec s. S anda d e o s in pa en heses a e clus e ed boo s apped (1000 eps) a he s a e le el
* p < 0.1; **p < 0.05; ***p < 0.01
All Bi hs Pa i y > 1 Pa i y > 2
(1) (2) (3) (4) (5) (6)
T ea × Pos × Gi l 0.005 0.013 0.058*
(0.007) (0.014) (0.034)
T ea × Pos 2006-10 × Gi l 0.008 0.020 0.091**
(0.007) (0.015) (0.036)
T ea × Pos 2011-15 × Gi l 0.002 0.009 0.018
(0.009) (0.018) (0.039)
No. o Obs 150,757 150,757 63,250 63,250 23,275 23,275
Main FE Yes Yes Yes Yes Yes Yes
S.Ja adeka , K.Saxena
32 Page 24 o 34
6.1 Impac onIn an Mo ali y
We look a he unde 5 mo ali y o child en bo n o women in ou sample. Biologi-
cally, mo ali y is g ea e among boys han gi ls be ween he age o 0 and 1 (K ae-
me 2000); he e o e, i we obse e highe mo ali y o gi ls han boys, i would
indica e ha gi ls a e being neglec ed.
Using ou di e ence-in-di e ences es ima o , we es ed whe he he p og am
inc eased child mo ali y o gi ls. We es ima e he model:
The esul s in Tables 6 and 7 show whe he he e a e disp opo iona ely mo e
gi ls among in an s who died in hei i s yea o be o e eaching i e yea s o age.
Fo each o hese samples, he i s wo columns show he esul s o all in an s in
u al India i espec i e o hei bi h o de . Columns 3 and 4 show he esul s o
all in an s who we e bo n a bi h o de g ea e han 1. The las wo columns show
esul s o all in an s bo n a bi h o de g ea e han 2. We make his dis inc ion by
bi h o de because gi ls a a highe bi h o de end o die mo e han boys, due o
neglec and disc imina ion.
We ind ha o bo h age g oups, he p obabili y ha he deceased child is a gi l
is posi i e o all bi h o de s. Fo gi ls bo n a a pa i y g ea e han 1, he likelihood
o a gi l dying is 2 pe cen age poin s g ea e in he ea men g oup. This is mo e
p ominen in he ea lie di usion pe iod and o gi ls bo n a a pa i y g ea e han
2. The likelihood o a gi l dying be o e eaching he age o 5 is nea ly 6 pe cen age
poin s g ea e in he ea men g oup a e he p og am. An in e es ing obse a ion
is ha he signi ican di e ence in mo ali y be ween gi ls and boys disappea s when
we look a he la e di usion pe iod. This could be due o he addi ional ea u e o
p o iding nu i ional supplemen s o in an s ha we e added o he p og am in 2011.
7 Mechanisms
7.1 Ul asound access channel
Acco ding o he li e a u e, one o he main channels ha impac s households’ sex-
selec i e e ili y decisions is access o p e-na al sex de e mina ion echnologies
such as ul asounds. All p og am bene icia ies we e expec ed o unde go h ee an e-
na al checkups ha included ul asound scans. Al hough disco e ing he gende o
he oe us was no he pu pose o he scans, pa en s migh use his in o ma ion and
abo unwan ed emale oe uses. Since we canno obse e who uses he echnology
o de e mine he sex o he oe us and who uses i o sa is y he p og amme con-
di ion, we can hypo hesize ha i mo e people we e using his aspec o he p o-
g amme o sex selec , we should see his channel o lead o on a e age a signi ican ly
lowe p obabili y o gi ls being bo n on a e age in he ea men g oup.
Using he DHS- 2015/16 da a, we ob ained in o ma ion on which mo he s
epo ed ha ing used ul asound echnology. Column 2 in Table8 shows he esul s
(6)
Deadbi
=
𝛽0
+
𝛽1T ea i
×
Pos
×
Gi li
+
𝛽2T ea i
+
𝛽3Pos
+
S a eyea s
+
𝛿i
+
𝜆
+
𝜃b
+
ebi
The seen andunseen: heunin ended impac o acondi ional… Page 31 o 34 32
8 Conclusion andpolicy ecommenda ions
This pape examined he impac o he JSY condi ional cash ans e p og am on he
e ili y decisions o mo he s in u al India. Mo e speci ically, his s udy p o ides
causal e idence o he impac o he JSY on sex-selec i e beha iou among Indian
households. The esul s show ha , con a y o p e ious wo k on sex selec ion, his
p og am led o an inc ease in he p obabili y o ha ing a gi l a each bi h o de o
mo he s eligible o he p og am. The magni ude is especially la ge in amilies who
acco ding o he li e a u e ha e a g ea e incen i e o sex selec , i.e. hose whose i s
child is a daugh e . Al hough o e all in he coun y, he e has been an inc ease in
he p e alence o sex-selec i e abo ions, he JSY managed o educe his p ac ice
among amilies who quali ied o he p og am.
The esul s also showed ha while he e we e mo e gi ls being bo n o amilies in
LPS; hese gi ls a e also mo e likely o die be o e eaching he age o 5 yea s. These
indings indica e ha hough he e a e imp o emen s in bi h ou comes o gi ls as a
esul o he p og am; i may be he case ha disc imina ion agains hem con inues
and shi s om p ena al o pos na al disc imina ion.
Ou esul s show ha in he 0–4 age g oup, 424,825 women we e missing om
he popula ion. Howe e , his is an imp o emen o nea ly 300,000 women com-
pa ed o 724,997 missing women in he same age g oup a decade p io o he p o-
g am. While he e s ill is a e y la ge numbe o missing gi ls in he coun y, he
policy con ibu ed o educing his numbe . The channel ha leads o his esul is
he one d i en by communi y heal h wo ke s (ASHA) ha we e appoin ed as pa
o he p og am o assis p egnancies in hei neighbou hood. Since hese wo ke s
eco d each p egnancy o bene icia ies o he p og am and ge inancial incen i es
o e e y li e bi h o bene icia ies a heal h ins i u ions, hey ac as de e en s o
couples o selec i ely abo hei oe uses. This esul suppo s he eme ging e i-
dence on he ole ha heal h wo ke s play in e icien public good dis ibu ion and in
suppo ing heal h p og ams.
The e ec i eness o communi y heal h wo ke s in educing he p ac ice o p e-
na al sex-selec i e abo ions ei he due o pa en al ea o being epo ed i hey
unde go a sex-selec i e abo ion o ASHA’s p essu e on pa en s o no abo he
child as he paymen is condi ional on a bene icia y’s deli e y in a hospi al. This
is an impo an piece o e idence in a coun y ha has been unsuccess ully ying
o educe emale oe icide h ough laws agains sex-selec i e abo ions o inancial
incen i es o bea gi ls. Howe e , ou esul s should be aken wi h cau ion as we do
no claim ha he in e en ion o heal h wo ke s shi ed pa en al son p e e ence in
India. Ou esul s poin o he ac ha pa en al son p e e ence was me ely subs i-
u ed by pos na al excess gi l mo ali y be ween he ages o 24 and 59 mon hs.
Supplemen a y In o ma ion The online e sion con ains supplemen a y ma e ial a ailable a h ps:// doi.
o g/ 10. 1007/ s00148- 025- 01091-6.
Acknowledgemen s The au ho s hank Giacomo de Gio gi and Lo e Vandewalle o hei supe ision
and suppo . This pape bene i ed om commen s by Ch is elle Dumas, Eliana La Fe a a, Jaya K ishna-
kuma , T a is Lybbe , Tobias Muelle , Dilip Mukhe jee, Michele Pellizza i, Deb aj Ray, Alessand o
Ta ozzi, and Aleksey Te eno . We a e hank ul o all he pa icipan s o he b own bag semina a he

S.Ja adeka , K.Saxena
32 Page 32 o 34
Uni e si y o Gene a and he G adua e Ins i u e o In e na ional and De elopmen S udies, NCDE 2019,
SSDe 2019, DENS Swi ze land 2019, he SSES 2019 6 h DENeB Be lin 2019, and 2019 ISI con e ence
New Delhi. Biplob Biswas’s Py hon expe ise was indispensable. We a e hank ul o Vik am Bahu e o
sha ing he CHIRPS da a. Finally, we hank he h ee anonymous e e ees and edi o Kompal Sinha o
hei aluable commen s.
Da a a ailabili y The da a and do ile will be made a ailable upon eques .
Decla a ions
Con lic o in e es The au ho s decla e no compe ing in e es s.
Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License,
which pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long
as you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e
Commons licence, and indica e i changes we e made. The images o o he hi d pa y ma e ial in his
a icle a e included in he a icle’s C ea i e Commons licence, unless indica ed o he wise in a c edi line
o he ma e ial. I ma e ial is no included in he a icle’s C ea i e Commons licence and you in ended
use is no pe mi ed by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mis-
sion di ec ly om he copy igh holde . To iew a copy o his licence, isi h p://c ea i ecommons.o g/
licenses/by/4.0/.
Re e ences
Akbulu -Yuksel M, Rosenblum D (2023) Es ima ing he e ec s o expanding ul asound use on sex selec-
ion in India. J De S ud 59(4):516–532
Al ano M (2017) Daugh e s, dow ies, deli e ies: he e ec o ma i al paymen s on e ili y choices in
India. J De Econ 125:89–104
Almond D, Li H, Zhang S (2019) Land e o m and sex selec ion in China. J Poli Econ 127(2):560–585
Almond, D., H. Li, and S. Zhang (2013). Land e o m and sex selec ion in China. Technical epo ,
Na ional Bu eau o Economic Resea ch
Ande son S, Ray D (2010) Missing women: age and disease. Re Econ S ud 77(4):1262–1300
Anuk i i S (2018) Financial incen i es and he e ili y-sex a io ade-o . Am Econ J Appl Econ
10(2):27–57
Anuk i i S, Bhalo a S, Tam EH (2022a) On he quan i y and quali y o gi ls: e ili y, pa en al in es -
men s and mo ali y. Econ J 132(641):1–36
Anuk i i S, Kwon S, P akash N (2022b) Sa ing o dow y: e idence om u al India. J De Econ
154:102750
A nold F, Choe MK, Roy TK (1998) Son p e e ence, he amily-building p ocess and child mo ali y in
India. Popul S ud 52(3):301–315
Bane jee S, Sen G (2024) Pe sis en e ec s o a condi ional cash ans e : a case o empowe ing women
h ough Kanyash ee in India. J Popul Econ 37(4):66
Be and M, Du lo E, Mullaina han S (2004) How much should we us di e ences-in-di e ences es i-
ma es? Q J Econ 119(1):249–275
Bhalo a S, Chak a a y A, Gulesci S (2020) The p ice o gold: dow y and dea h in India. J De Econ
143:102413
Bhalo a SR, Coch ane T (2010) Whe e ha e all he young gi ls gone? Iden i ica ion o sex selec ion in
India. h ps:// doi. o g/ 10. 2139/ ss n. 17311 85
Bhalo a SR, Clo s-Figue as I, Iye L (2018) Religion and abo ion: he ole o poli ician iden i y. h ps://
doi. o g/ 10. 2139/ ss n. 31172 96
Bha adwaj P, Lakdawala LK (2013) Disc imina ion begins in he womb: e idence o sex-selec i e p ena-
al in es men s. J Human Resou 48(1):71–113
Bhaska V (2007) Pa en al choice and gende balance. Uni e si y College London
The seen andunseen: heunin ended impac o acondi ional… Page 33 o 34 32
Bjö kmanNyq is M, Gua iso A, S ensson J, Yanagizawa-D o D (2019) Reducing child mo ali y in he
las mile: expe imen al e idence on communi y heal h p omo e s in Uganda. Am Econ J Appl Econ
11(3):155–192
Boland MR, Fiede M, John LH, Rijnbeek PR, Hube S (2020) Female ep oduc i e pe o mance and
ma e nal bi h mon h: a comp ehensi e me a-analysis explo ing mul iple seasonal mechanisms. Sci
Rep 10(1):555
Bo ke G, Eeckhou J, Luke N, Minz S, Munshi K, Swamina han S (2017) Weal h, ma iage, and sex
selec ion.Unpublished manusc ip
Bo ke G (2021) Sa e y i s : pe cei ed isk o s ee ha assmen and educa ional choices o women
B enne JL, Kabakyenga J, Kyomuhangi T, Wo on KA, Pim C, N a o M, Bagenda FN, Gad NR, Godel J,
Kayizzi J e al (2011) Can olun ee communi y heal h wo ke s dec ease child mo bidi y and mo -
ali y in sou hwes e n Uganda? An Impac E alua ion Plos One 6(12):e27997
Buchmann N, Field EM, Glenne s e R, Hussam RN (2019) Th owing he baby ou wi h he d inking
wa e : unin ended consequences o a senic mi iga ion e o s in Bangladesh(No. w25729). Na ional
Bu eau o Economic Resea ch
Celhay PA, Ge le PJ, Gio agnoli P, Ve mee sch C (2019) Long- un e ec s o empo a y incen i es on
medical ca e p oduc i i y. Am Econ J Appl Econ 11(3):92–127
Cha e jee S, Podda P (2021) F om sa e mo he hood o cogni i e abili y: explo ing in ahousehold and
in e gene a ional spillo e s. Economica 88(352):1075–1106
Chen Y, Li H, Meng L (2013) P ena al sex selec ion and missing gi ls in China: e idence om he di u-
sion o diagnos ic ul asound. J Human Resou 48(1):36–70
Cla k S (2000) Son p e e ence and sex composi ion o child en: e idence om India. Demog aphy
37(1):95–108
Cohen A, Dehejia R, Romano D (2013) Financial incen i es and e ili y. Re Econ S a 95(1):1–20
Congdon Fo s H, Lindskog A (2023) Son p e e ence and educa ion inequali ies in India: he ole o gen-
de -biased e ili y s a egies and p e e en ial ea men o boys. J Popul Econ 36(3):1431–1460
Das N (1987) Sex p e e ence and e ili y beha io : a s udy o ecen Indian da a. Demog aphy
24(4):517–530
Das Gup a M, Ma i Bha P (1997) Fe ili y decline and inc eased mani es a ion o sex bias in India.
Popul S ud 51(3):307–315
Ebens ein A (2010) The “missing gi ls” o China and he unin ended consequences o he one child pol-
icy. J Human Resou 45(1):87–115
Edlund L (1999) Son p e e ence, sex a ios, and ma iage pa e ns. J Poli Econ 107(6):1275–1304
Edlund L, Li H, Yi J, Zhang J (2013) Sex a ios and c ime: e idence om China. Re Econ S a
95(5):1520–1534
Gella ly C, Pe ie M (2017) P ena al sex selec ion and emale in an mo ali y a e mo e common in India
a e i s bo n and second-bo n daugh e s. J Epidemiol Commun Heal h 71(3):269–274
Goodkind D (1996) On subs i u ing sex p e e ence s a egies in Eas Asia: does p ena al sex selec ion
educe pos na al disc imina ion? Popul De Re 22(1):111–126
Heske h T, Xing ZW (2006) Abno mal sex a ios in human popula ions: causes and consequences. P oc
Na l Acad Sci 103(36):13271–13275
Hu L, Schlosse A (2015) P ena al sex selec ion and gi ls’ well-being: e idence om India. Econ J
125(587):1227–1261
Jayachand an S, Kuziemko I (2011) Why do mo he s b eas eed gi ls less han boys? E idence and impli-
ca ions o child heal h in India. Q J Econ 126(3):1485–1538
Jensen R (2012) Ano he mou h o eed? The e ec s o (in) e ili y on malnu i ion. Cesi o Econ S ud
58(2):322–347
Joshi S, Si a am A (2014) Does i pay o deli e ? An e alua ion o India’s sa e mo he hood p og am.
Wo ld De 64:434–447
Kashyap R (2019) Is p ena al sex selec ion associa ed wi h lowe emale child mo ali y? Popul S ud
73(1):57–78
K aeme S (2000) The agile male. BMJ 321(7276):1609–1612
K ombholz H (2023) Global pa e ns o seasonal a ia ion in human bi h a es-in luence o clima e and
economic and social ac o s.h ps:// doi. o g/ 10. 23668/ psych a chi es. 13203
Lin M-J, Liu J-T, Qian N (2014) Mo e missing women, ewe dying gi ls: he impac o sex-selec i e
abo ion on sex a bi h and ela i e emale mo ali y in Taiwan. J Eu Econ Assoc 12(4):899–926
Nandi A, Laxmina ayan R (2016) The unin ended e ec s o cash ans e s on e ili y: e idence om he
sa e mo he hood scheme in India. J Popul Econ 29(2):457–491
S.Ja adeka , K.Saxena
32 Page 34 o 34
Na h S (2023) Explaining hi d bi h pa e ns in India: causal e ec s o sibling sex composi ion. J Popul
Econ 36(4):2169–2203
Olden A, Møen J (2022) The iple di e ence es ima o . Econome J 25(3):531–553
Powell-Jackson T, Mazumda S, Mills A (2015) Financial incen i es in heal h: new e idence om India’s
Janani Su aksha Yojana. J Heal h Econ 43:154–169
Ri chie H, Rose M (2019) Gende a io. Ou Wo ld in Da a. h ps:// ou wo ldin da a. o g/ gende - a io .
Accessed May 2019
Rose E (1999) Consump ion smoo hing and excess emale mo ali y in u al India. Re Econ S a
81(1):41–49
Rosenblum D (2013a) The e ec o e ili y decisions on excess emale mo ali y in India. J Popul Econ
26(1):147–180
Rosenblum D (2013b) Economic incen i es o sex-selec i e abo ion in India. Canadian Cen e o
Heal h Econ 2014–13
Ro h J, San ’Anna PH, Bilinski A, Poe J (2023) Wha ’s ending in di e ence-in-di e ences? A syn hesis
o he ecen econome ics li e a u e. Jou nal o Econome ics 235(2):2218–2244
Roy S (2015) Empowe ing women? Inhe i ance igh s, emale educa ion and dow y paymen s in India. J
De Econ 114:233–251
Sekhe TV (2012) Ladlis and Lakshmis: inancial incen i e schemes o he gi l child. Economic and
Poli ical Weekly 58–65
Sekh i S, S o eyga d A (2014) Dow y dea hs: esponse o wea he a iabili y in India. J De Econ
111:212–223
Sen S, Cha e jee S, Mohan y SK e al (2020) Unin ended e ec s o Janani Su aksha Yojana on ma e nal
ca e in India. SSM-Popula ion Heal h 11:100619
Singh P, Mas e s WA (2017) Impac o ca egi e incen i es on child heal h: e idence om an expe imen
wi h anganwadi wo ke s in India. J Heal h Econ 55:219–231
Sinha N, Yoong J (2009) Long- e m inancial incen i es and in es men in daugh e s: e idence om con-
di ional cash ans e s in No h India. The Wo ld Bank
Valen e C (2014) Access o abo ion, in es men s in neona al heal h, and sex-selec ion: e idence om
Nepal. J De Econ 107:225–243
Visa ia L (2005) Female de ici in India: ole o p e en ion o sex selec i e abo ion ac . In: CEPED-
CICREDINED Semina on Female De ici in Asia: T ends and Pe spec i es, Singapo e, pp. 5–7.
Ci esee
Wa e ield H (1875) Memo andum on he census o B i ish India 1871–72.Ey e and Spo iswoode
Publishe ’s No e Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in published maps
and ins i u ional a ilia ions.