Broken homes and empty pantries: the impact of couple separation on living standards in France

Mink, Julia
A icle — Published Ve sion
B oken homes and emp y pan ies: he impac o couple
sepa a ion on li ing s anda ds in F ance
In e na ional Re iew o Economics
P o ided in Coope a ion wi h:
Sp inge Na u e
Sugges ed Ci a ion: Mink, Julia (2025) : B oken homes and emp y pan ies: he impac o couple
sepa a ion on li ing s anda ds in F ance, In e na ional Re iew o Economics, ISSN 1863-4613,
Sp inge , Be lin, Heidelbe g, Vol. 72, Iss. 2,
h ps://doi.o g/10.1007/s12232-025-00499-6
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/323902
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)
In e na ional Re iew o Economics (2025) 72:26
h ps://doi.o g/10.1007/s12232-025-00499-6
RESEARCH ARTICLE
B oken homes andemp y pan ies: heimpac o couple
sepa a ion onli ing s anda ds inF ance
JuliaMink1
Recei ed: 3 Sep embe 2024 / Accep ed: 30 June 2025
© The Au ho (s) 2025
Abs ac
This s udy sheds new ligh on he impac o couple sepa a ion on he li ing s and-
a ds o women by conside ing he e ec s o sepa a ion on measu es e lec ing he
adequacy o ood consump ion in addi ion o mo e commonly s udied income and
expendi u e measu es. Using panel da a om F ance, I examine changes in dis-
posable income, ood expendi u es and quan i ies pu chased, die quali y and body
weigh a ound he ime o sepa a ion and up o 8yea s la e , compa ed o a con ol
g oup o households ha did no sepa a e. Li ing s anda ds decline o such an ex en
ha some households canno main ain hei ood consump ion, esul ing in weigh
loss.
Keywo ds Sepa a ion· Di o ce· Li ing s anda ds· Income· Food consump ion·
Weigh loss· E en s udy
JEL Classi ica ions I12· J12· D12
1 In oduc ion
Family dissolu ion due o couple sepa a ion has become an e en ha a la ge pa
o he popula ion will expe ience a some poin in hei li es. In F ance, he sha e
o cohabi ing couples who b oke up hei i s union a e less han 8yea s mo e
han doubled, om 12% o unions o med in he 1970s o 29% o unions o med
be ween 1997 and 2005 (INSEE 2015). A couple who ma ied in 1970 has a 12%
chance o e e ge ing di o ced, compa ed o 44% o couples who ma ied in 2014
(INSEE 2016). T ends a e simila in many coun ies a ound he wo ld (OECD
2011). Sepa a ion has many consequences o hose in ol ed, especially inancially.
Di o ce has been associa ed wi h a educ ion in income (see o example McKee e
* Julia Mink
[email p o ec ed]
1 Depa men o Economics, Ins i u e o Applied Mic oeconomics, Uni e si y o Bonn,
Lennés aße 43, 53113Bonn, Ge many
J.Mink
26 Page 2 o 56
and Wol inge (2001); A ella and Smock (2005); Tach and Eads (2015)). In F ance,
i is es ima ed ha women’s a e age s anda d o li ing d ops by 14–25% in he yea
ollowing a di o ce, whe eas men’s li ing s anda ds a e es ima ed o all by 3–7%
(Cos emalle 2017).
Since sepa a ion means he end o he economies o scale associa ed wi h cohabi-
a ion and he sha ing o ce ain expenses, a decline in li ing s anda ds a e sepa-
a ion is o be expec ed. Wha emains an open ques ion, howe e , is he ex en o
which he income shock o sepa a ion a ec s households’ abili y o main ain a mini-
mum le el o consump ion necessa y o mee basic needs. The s udies ha look a
he e ec o sepa a ion on income p o ide in o ma ion on ela i e changes in he
inancial si ua ion, bu no so much on whe he households ha e di icul y mee ing
hei basic needs in absolu e e ms. Households could a y in hei abili y o adap o
lowe budge s by being mo e o less able o u n o less expensi e op ions. The ques-
ion o households’ abili y o main ain a minimum le el o consump ion is impo -
an . When households a e unable o mee hei basic needs, he e can be cos ly neg-
a i e impac s on human capi al, such as poo e heal h. Lowe economic esou ces
ha e been associa ed wi h poo e adul and child ou comes, including poo e men al
and physical heal h and lowe educa ional a ainmen (McLanahan e al. 2013; Tach
and Eads 2015).
This s udy add esses his gap in he li e a u e by examining whe he couple sepa-
a ion a ec s households’ abili y o mee basic consump ion needs, wi h a ocus on
ood consump ion as a undamen al necessi y. Mo e p ecisely, his s udy es ima es
he e ec s o couple sepa a ion on household disposable income, ood expendi u es
and quan i ies pu chased, die quali y, and body weigh . Looking a he quan i y o
ood pu chased, die quali y, and body weigh is c ucial because i is mo e in o ma-
i e abou whe he households a e ha ing di icul y mee ing hei basic die a y
needs han changes in income and e en ood expendi u es would sugges . A all in
income and ood expendi u es does no necessa ily signal a dec ease in ac ual ood
consump ion as households migh be able o shop o ba gains and subs i u e o
cheape p oduc s ins ead.
A majo obs acle o s udying his issue is he di icul y in ob aining indi idual-
le el da a on household composi ion and income, as well as eliable and de ailed
da a on ood consump ion. Fo his s udy, I use consume panel da a om Kan a
Wo ldpanel o F ance, which include basic demog aphic in o ma ion on house-
hold membe s, household composi ion, and de ailed in o ma ion on ood pu chases
en e ed by households ia scanne s a e each g oce y pu chase. Al hough men a e
included in he sample, mos pos -sepa a ion obse a ions in ol e emale-headed
households. The analysis he e o e concen a es on women, which is also pa icu-
la ly ele an gi en exis ing e idence ha hey end o bea he inancial conse-
quences o sepa a ion. In an e en s udy app oach, I examine how he ou come a i-
ables e ol e in he yea s be o e and a e he sepa a ion, ela i e o he pe iod jus
be o e sepa a ion and compa ed o households whe e no sepa a ion is obse ed. I
con ol o ime-in a ian di e ences ac oss households and he impac o mac o-
le el shocks by including household and yea ixed e ec s. To iden i y ulne able
households and explo e po en ial adap a ion mechanisms, he analysis also exam-
ines he e ogenei y by p e-sepa a ion income, amily composi ion, and employmen
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 3 o 56 26
s a us. I pe o m se e al obus ness checks by selec ing con ol households di -
e en ly, omi ing con ol households al oge he and calcula ing es ima es ha a e
obus o ea men e ec he e ogenei y, as p oposed in he ecen wo-way ixed
e ec s li e a u e.
Se e al key indings eme ge om his s udy. Fi s , I ind ha women su e signi -
ican and pe manen losses o economic esou ces a e sepa a ion. In he i s yea
a e sepa a ion, hei disposable income is a ound 20% lowe ela i e o p e-sepa-
a ion le els and emains 30% lowe un il he end o he obse a ion window up o
7yea s a e he sepa a ion. Second, women appea o cu back on ood consump ion
a e a sepa a ion. Food expendi u es and quan i ies pu chased decline by 20–40%
ela i e o p e-sepa a ion le els. Thi d, he changes in income and ood consump ion
a e accompanied by a 1.5% decline in he women’s body mass index (BMI1) du ing
he i s 3yea s a e sepa a ion. I ind e idence o an inc ease in he p obabili y o
being unde weigh (BMI o less han 18.5) as well as being o no mal weigh , while
he p obabili y o being o e weigh (BMI o o e 25) dec eases. The weigh loss
occu s despi e an inc ease in he sha e o unheal hy ood p oduc s pu chased con-
sis ing o mo e sal y, swee , a y, and con enience oods, which a e calo ie-dense.
Impo an ly, he e ec s on income and ood pu chases a e no simply a mechanical
consequence o he depa u e o a household membe . I , ins ead o adjus ing o
household composi ion, pe capi a income and ood pu chases a e used, he esul s
a e quali a i ely equi alen . I ind no e ec when an adul o he han he pa ne ( o
example, an adul child) lea es he household. I he e ec we e pu ely mechanical,
I would expec o ind simila esul s when any adul household membe lea es he
household. Fou h, he e ec s on income a e s onge in high-income households,
bu ood pu chases and body weigh decline mo e s ongly in low-income house-
holds and single-pa en households. The p obabili y o being unde weigh a e
sepa a ion inc eases in low-income households, bu no in high-income households.
Labou ma ke pa icipa ion does no appea o signi ican ly bu e he e ec s o
sepa a ion. In con as , en e ing a new ela ionship seems o mi iga e he impac ,
al hough he es ima es a e imp ecise. I also ind e ec s o men, hough hey a e
mo e mode a e. Income declines by app oxima ely 8%, and ood pu chases all by
10–20%, hough hese es ima es a e no s a is ically p ecise.
This s udy con ibu es o he li e a u e on he economic consequences o sepa-
a ion in h ee key ways. Fi s , i goes beyond commonly used indica o s such as
income and expendi u es by inco po a ing di ec measu es o nu i ional well-being,
including ood quan i ies pu chased, die quali y, and body weigh . These ou comes
p o ide a mo e comple e pic u e o li ing s anda ds and ulne abili y, ecognizing
ha households may adap o lowe budge s wi hou immedia ely educing calo ie
in ake o nu i ional adequacy. Second, i p o ides medium and long- e m es ima es
o sepa a ion e ec s using de ailed longi udinal da a on household consump ion.
While many s udies ha e documen ed sho - e m income declines ollowing di o ce,
1 De ined as body weigh di ided by heigh squa ed. A commonly used measu e o s anda dise weigh
o heigh and a ule o humb o b oadly ca ego ise a pe son as unde weigh , no mal weigh , o e weigh ,
o obese. The esul s a e quali a i ely he same when using body weigh di ec ly.
J.Mink
26 Page 4 o 56
ypically anging om 23% o 40% in he yea a e sepa a ion (Ho man 1977; Dun-
can and Ho man 1985b; Bianchi and McA hu 1991; Holden and Smock 1991;
McLanahan and Sande u 1994; Pe e son 1996; Gala neau and S u ock 1997; McK-
ee e and Wol inge 2001; A ella and Smock 2005; Tach and Eads 2015), ewe
s udies examine longe - e m consequences. Whe e such s udies exis , hey o en ely
on da ed samples o limi ed co a ia e adjus men (Weiss 1984; Duncan and Ho -
man 1985a; Fishe and Low 2009; De Vaus e al. 2014). This s udy add esses hese
gaps by acking ou comes o e se e al yea s and accoun ing o bo h ime- a ying
household cha ac e is ics and unobse ed he e ogenei y. Thi d, i p o ides new e i-
dence om F ance, a con ex ha di e s signi ican ly om he Uni ed S a es, whe e
mos p io esea ch has been conduc ed. In 2017, public spending on amily bene i s
in F ance, including cash ans e s, se ices, and ax elie , ep esen ed o e 3.5%
o GDP, compa ed o only 1% in he Uni ed S a es (OECD 2017). This di e ence
makes F ance a aluable se ing o examining whe he mo e ex ensi e social p o-
ec ion al e s he consequences o sepa a ion. This s udy is mos closely ela ed o
Page and S e ens (2004), who examine he e ec s o amily s uc u e on income and
ood spending in he Uni ed S a es, and o Bonne e al. (2021), who s udy di o ce
e ec s on li ing s anda ds and labou supply in F ance. The p esen analysis builds
on his wo k by inco po a ing nu i ional and an h opome ic ou comes, o e ing a
b oade pe spec i e on how sepa a ion a ec s household well-being.
O e all, he s udy p o ides e idence ha couple sepa a ion leads o a signi i-
can and las ing decline in li ing s anda ds o women, pa icula ly hose in low-
income and single-pa en households. Reduc ions in ood quan i ies pu chased and
body weigh sugges ha some households s uggle o mee basic nu i ional needs,
e en in a coun y wi h a compa a i ely gene ous social sa e y ne . While he e is
also some e idence o ad e se e ec s o men, hese a e mo e mode a e and less
p ecisely es ima ed. Taken oge he , he indings highligh he impo ance o going
beyond income-based indica o s o iden i y ulne abili y and sugges ha policy
suppo may be needed o ensu e ha all households, especially he mos economi-
cally agile, can main ain access o adequa e nu i ion and basic li ing s anda ds
ollowing sepa a ion.
The emainde o he pape is as ollows. Sec ion 2 desc ibes he da a and
de ails he empi ical s a egy, Sec .3 p esen s and discusses he esul s, and Sec .4
concludes.
2 Me hod
2.1 Da a
This s udy d aws on de ailed da a om a ep esen a i e sample o F ench house-
holds collec ed by Kan a Wo ldpanel, co e ing he pe iod om 2005 o 2014. The
da ase is well sui ed o his analysis because i combines ich in o ma ion on house-
hold cha ac e is ics and income wi h de ailed eco ds o ood pu chases, allowing
o he cons uc ion o bo h inancial and consump ion-based measu es o li ing
s anda ds. Kan a Wo ldpanel is a p i a e i m specializing in consume panels and

B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 5 o 56 26
ma ke esea ch, simila o Nielsen in he Uni ed S a es. Pa icipa ing households a e
equipped wi h scanne s o eco d ood pu chases wi h ba codes. I ems wi hou ba -
codes a e en e ed manually by household membe s.2 The da ase p o ides in o ma-
ion on indi idual and household cha ac e is ics, including household composi ion,
household disposable income bins whe e income includes any income and ans e s
(ea nings, pensions, alimony paymen s, go e nmen ans e s), he sociop o essional
ca ego y, age, sex, heigh , weigh , educa ion le el, and labou ma ke s a us o each
household membe . In o ma ion on ood pu chases includes p oduc ype, quan i y,
p ice and pu chase da e.
As he da a on indi idual and household cha ac e is ics a e upda ed annually, he
ime in e al o analysis in his s udy is he yea . I cons uc household ood quan i-
ies pu chased as he sum o all p oduc s pu chased annually, measu ed in kilog ams,
and household ood expendi u e as he sum o annual ood expendi u es, exp essed
in eu os. The sha e o unheal hy ood p oduc s is de ined as he a io o he annual
quan i y in kilog ams o p epa ed ood p oduc s (such as pizza, saue k au , cas-
soule ), sal y- a y p oduc s (such as inge ood, c isps, c acke s, appe ise s), and
swee - a y p oduc s (such as candy, chocola e, cookies, pas ies, ice c eams, jams)
o he o al amoun o annual ood pu chases, also measu ed in kilog ams.
I use in o ma ion on body weigh and heigh o calcula e he body mass index
(BMI) o each household membe . BMI is de ined as weigh di ided by heigh
squa ed. I is a commonly used measu e o s anda dise weigh o heigh and se es
as a ule o humb o ca ego ise indi iduals as unde weigh (BMI unde 18.5kg/
m
2
), no mal weigh (18.5–24.9), o e weigh (25–29.9), o obese (30 o mo e). BMIs
unde 20 and o e 25 ha e been associa ed wi h highe all-cause mo ali y, wi h he
isk inc easing wi h dis ance om he 20–25 ange.
I do no ha e da a on ood consumed ou side he home, bu households epo he
numbe o meals ypically ea en a home o each day o he week. I use his in o -
ma ion o calcula e he weekly a e age numbe o meals ea en a home in a gi en
yea , in o de o in es iga e whe he sepa a ion a ec s ea ing habi s a home as a
po en ial mechanism behind he obse ed changes in ood pu chases.
A limi a ion o he Kan a Wo ldpanel da a is ha i does no include di ec in o -
ma ion on he ma i al o pa ne ship s a us o household membe s. Households
a e classi ied in o young, mid-age and old single o couple households and amily
households. The s a us changes om couple o single household when a pa ne
lea es, bu amily households emain classi ied as amily as long as child en a e p e-
sen . To o e come his limi a ion, I exploi he ac ha each indi idual is assigned
a s a us code ha is no eassigned o o he household membe s in case he membe
lea es he household. S a us 1 is always assigned o he emale household mem-
be who is in cha ge o he household, as opposed o s a us 3, which designa es a
dependen emale, such as a child. S a us 2 is always assigned o a male household
membe who is assumed o be in cha ge o he household, as opposed o s a us 4,
which deno es a dependen male, such as a child. I de ine sepa a ion as he depa u e
2 Fo mo e in o ma ion, see he Kan a Wo ldpanel websi e h ps:// www. kan a wo l dpanel. com/ global/
Consu me - Panels.
J.Mink
26 Page 6 o 56
om he household o an indi idual wi h s a us 1 o 2. I am he e o e no dis in-
guishing be ween he sepa a ion o cohabi ing o ma ied couples. This app oach
may also cap u e some ins ances o pa ne dea h a he han sepa a ion. Howe e ,
his is likely o be a e in he wo king-age popula ion. In F ance, only abou 6% o
single-pa en amilies esul om he dea h o a pa ne , and ewe han 5% o am-
ily dissolu ions be o e age 45 a e due o dea h a he han sepa a ion (INSEE 2015;
INED 2018). I show in Sec .3.2 ha esul s a e obus when es ic ing he sample
o indi iduals younge han 45yea s, o whom pa ne dea h is an unlikely sou ce
o bias.
I obse e a o al o 42,000 households o an a e age du a ion o 5.9yea s, while
he maximum obse a ion window is 10yea s. The esul s p esen ed in his s udy a e
based on analyses using a sample o ea men households ha is un es ic ed, excep
o he equi emen ha he household be obse ed a leas one yea be o e sepa a-
ion, in he yea o sepa a ion, and a leas one yea a e sepa a ion. This ensu es
ha he e ec s o sepa a ion can be iden i ied. As his es ic ion has no e ec on he
es ima ed coe icien s, I a oid imposing u he limi a ions on he sample o sepa a -
ing households, in o de o e ain as many sepa a ion e en s as possible.
The con ol g oup is es ic ed o households obse ed o a leas 7 consecu i e
yea s. This es ic ion aims o a oid he appea ance o p e- ends in ood pu chases.
Compa ed o o he ou come a iables, ood pu chases a y mo e s ongly ac oss
households and o e ime. Using a panel composed o households obse ed o sho
pe iods may in oduce sample composi ion e ec s ha could bias he esul s. None-
heless, he esul s a e quali a i ely simila e en wi hou his es ic ion on he con-
ol sample. See Sec .3.2 o u he discussion and addi ional sensi i i y analyses
using al e na i e e en ime windows o e i y ha esul s a e no d i en by changes
in sample composi ion.
Table1 epo s summa y s a is ics o bo h he un es ic ed sample and he sub-
sample es ic ed o households obse ed o a leas 7 consecu i e yea s, wi h
obse a ions pooled ac oss households and yea s. The wo samples a e b oadly
simila , al hough he es ic ed sample has highe a e age ood pu chases and olde
household membe s. Households included in he panel o only sho pe iods may
epo ood pu chases less consis en ly. The highe a e age age in he es ic ed sam-
ple may e lec di e en ime cons ain s, wi h olde indi iduals po en ially ha ing
mo e a ailabili y o emain in he panel. The inal sample comp ises 10,033 house-
holds, obse ed o an a e age o 8.6yea s.
Table 2 p o ides na ional s a is ics o F ance o compa e wi h he s udy da a.
In 2010, a e age household disposable income in F ance was EUR 2919, which is
close o he sample a e age o EUR 2646. The inac i i y a e in he sample is also
simila o he na ional le el. Household size in he sample is sligh ly la ge (2.6 s.
2.3 in he gene al popula ion). The a e age age, pa icula ly o men, is somewha
highe in he sample. Howe e , he esul s emain obus when es ic ing he sample
o indi iduals unde age 45 (see Sec .3.2 o esul s and discussion).
Table3 compa es summa y s a is ics o ea men households in he yea be o e
sepa a ion wi h non-single households in he con ol g oup. In sepa a ing house-
holds, bo h income and ood quan i ies pu chased a e lowe , and household membe s
a e younge on a e age han in he con ol g oup. Fo his eason, I apply household
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 7 o 56 26
ixed e ec s and adjus o pa ne age in all eg essions. The esul s a e also quali-
a i ely simila when he con ol g oup is omi ed en i ely o cons uc ed di e en ly,
including using p opensi y sco e ma ching (see Sec .3.2).
Ou o he 1038 households in which a sepa a ion is obse ed, 854 in ol e he
depa u e o a male pa ne (indi idual wi h s a us 2), and 184 in ol e he depa u e
o a emale pa ne (s a us 1). The “ emaining” household is de ined as he one in
which he pa ne con inues o espond o he Kan a Wo ldpanel su ey. In Kan a
Wo ldpanel, he panelis esponsible o g oce y shopping and epo ing household
pu chases is ypically a woman, and is o en he one who con inues o be obse ed
a e sepa a ion. A he ime o sepa a ion, 475 o he households a e couple-only
households. The emaining 563 include a leas one addi ional household mem-
be . Among hem, 306 households include a leas one child unde age 18, and 206
include a leas one child unde age 12. The es a e composed o o he adul mem-
be s in addi ion o he sepa a ing couple.
Table 4 shows he numbe o households obse ed a each yea ela i e o he
yea o sepa a ion. Since no all households a e obse ed h oughou he ull 10-yea
window, he sample composi ion a ies by e en ime. A o al o 1038 households
a e obse ed in he yea be o e, du ing, and a e sepa a ion, bu ewe obse a ions
a e a ailable in ea lie and la e yea s. The esul s emain quali a i ely simila when
households wi h ewe obse a ions a e g adually excluded, and when applying he
es ima o p oposed by De Chaisema in and d’Haul oeuille (2020), which accoun s
o unbalanced panels. See Sec s.3.2 and 3.3 o u he esul s and discussion.
2.2 Empi ical s a egy
To es ima e he e ec s o couple sepa a ion on household li ing s anda ds, I use
an e en s udy amewo k ha acks he e olu ion o ou come a iables be o e and
a e he sepa a ion. The analysis compa es households ha expe ience a sepa a ion
o hose ha do no , and con as s ou comes each yea o he yea immedia ely p e-
ceding he sepa a ion. Speci ically, I es ima e he ollowing model:
whe e
Yh
deno es he ou come o householdh in yea  ,
𝛼h
and
𝛾
a e household
and yea ixed e ec s,
Dj
h
deno es he ea men indica o o a sepa a ion occu ing
j∈[j,

j
]
pe iods away om ,and
Xh
is a ec o o ime- a ying household cha ac-
e is ics. The e o e m
𝜖h
is clus e ed a he household le el. The a iables o in e -
es a e he e en ime coe icien s
𝛽j
, which measu e he impac o sepa a ion ela i e
o he households ha do no sepa a e, and, since he e en ime dummy
j=−1
is
omi ed, ela i e o he yea jus be o e he sepa a ion.
The ou come a iables include household disposable income, ood expendi-
u es, he quan i y o ood pu chased in kilog ams, he sha e o unheal hy oods
in o al pu chases, and he body weigh o household membe s. While income and
expendi u e a e commonly used measu es o li ing s anda ds, hey may no ully
Y
h =𝛼h+𝛾 +

j
∑
j=j
𝛽jDj
h +𝜌Xh +𝜖h
,
J.Mink
26 Page 8 o 56
e lec a household’s abili y o mee basic needs. Income can unde es ima e ac ual
esou ces, and a educ ion in expendi u e may no imply educed consump ion i
households swi ch o cheape al e na i es. Measu es o ood quan i y, die quali y,
and body weigh o e a mo e di ec iew o nu i ional su iciency and ma e ial
ha dship.
Following Page and S e ens (2004), he main speci ica ions use he log o income
and ood pu chases, while con olling o household size in
Xh
. This accoun s o
changes in household composi ion om he a i al o depa u e o membe s in bo h
ea ed and con ol households. The es ima ed e en ime e ec s he e o e cap u e
he addi ional impac o losing a pa ne , o e and abo e he gene al e ec o a
change in household size. In obus ness checks, I es ima e models using pe capi a
ou comes and consump ion uni adjus men s. In he la e , he i s adul coun s as 1,
addi ional indi iduals aged 15 o olde as 0.7, and child en unde 15 as 0.5, o e lec
age-based needs and economies o scale. The esul s a e obus o hese al e na i e
speci ica ions.
Iden i ying causal e ec s o couple sepa a ion is challenging because sepa a ion
is unlikely o be exogenous. Fi s , households ha expe ience a sepa a ion may di -
e sys ema ically om hose ha do no . Second, ime- a ying shocks may bo h
inc ease he likelihood o sepa a ion and di ec ly in luence he ou come a iables.
To add ess po en ial biases, I include bo h household and yea ixed e ec s. House-
hold ixed e ec s abso b all ime-in a ian di e ences ac oss households, and yea
ixed e ec s accoun o mac oeconomic shocks common o all households. The
eg essions also include households ha do no sepa a e as con ols, a s anda d ea-
u e in e en s udy designs ha imp o es s a is ical p ecision by le e aging bo h
he iming o e en s and compa isons wi h un ea ed uni s (Mille 2023). While he
e en ime coe icien s a e iden i ied using wi hin-household a ia ion in he ea ed
g oup, yea ixed e ec s and he coe icien s on ime- a ying con ols a e iden i-
ied using a ia ion om bo h ea ed and con ol households. To assess whe he
di e ences be ween ea ed and con ol households migh bias esul s despi e ixed
e ec s, I conduc wo u he obus ness checks. Fi s , I eweigh he con ol g oup
using p opensi y sco es o make i mo e compa able o he ea ed g oup on obse -
able cha ac e is ics. Second, I es ima e models using only ea ed households, ely-
ing exclusi ely on he iming o sepa a ion. Bo h app oaches p oduce esul s ha a e
quali a i ely consis en wi h he main indings.
To add ess conce ns abou ime- a ying unobse ed shocks, I es o p e- ends
in he ou comes. I , o example, a heal h shock educes income and la e leads o
sepa a ion, I would expec o see a decline in income o ood pu chases be o e sep-
a a ion. In he da a, I ind no such p e- ends in income, ood quan i ies, o body
weigh . The e is a modes inc ease in he sha e o unheal hy ood pu chases be o e
sepa a ion, bu o he ou comes emain s able. This suppo s he iew ha he es i-
ma ed e ec s a e no p ima ily d i en by shocks occu ing be o e sepa a ion. Ne -
e heless, I canno ule ou bias om shocks ha simul aneously cause sepa a ion
and a ec ou comes wi hin he same yea . I e u n o his poin in he discussion o
esul s.
In all eg essions, I use he longes a ailable e en window o cap u e bo h sho -
and long- un e ec s. T ea ed households a e obse ed o up o 7yea s be o e and
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 15 o 56 26
% di e ence ela i e o j -1
Yea s since b eak-up
30% poo es households
% di e ence ela i e o j -1
Yea s since b eak-up
30% iches households
A. Income
% di e ence ela i e o j -1
Yea s since b eak-up
30% poo es households
% di e ence ela i e o j -1
Yea s since b eak-up
30% iches households
B. Food quan i ies pu chased
% di e ence ela i e o j -1
Yea s since b eak-up
30% poo es households
% di e ence ela i e o j -1
Yea s since b eak-up
30% iches households
C. Spouse's body weigh index
Di e ence ela i e o -1
Yea s since b eak-up
30% poo es households
Di e ence ela i e o -1
Yea s since b eak-up
30% iches households
D. Sha e o unheal hy ood pu chased
Fig. 2 T end in ou come a iables a ound sepa a ion, by p e-sepa a ion pe capi a income. The igu e shows e en ime
coe icien s o he ea men g oup o households whe e he male pa ne lea es he emale pa ne , ela i e o he con-
ol g oup o households whe e no sepa a ion occu s and ela i e o he yea jus be o e sepa a ion (
j
=−
1
). The le -
hand panels p esen esul s o he 30% o households wi h he lowes p e-sepa a ion pe capi a incomes, while he
igh -hand panels show esul s o he 30% o households wi h he highes incomes. The dependen a iables a e he
loga i hm o income, ood quan i ies pu chased, emaining pa ne ’s BMI, and he sha e o unheal hy ood p oduc s
pu chased by household i in yea . In addi ion o household and yea ixed e ec s and household size, all eg essions
include dummies o bo h pa ne s’ ages. The da a co e he pe iod om 2005 o 2015. The 95% con idence in e als
a e based on s anda d e o s clus e ed a he le el o he household

J.Mink
26 Page 16 o 56
also become s a is ically signi ican when using a single pos -sepa a ion dummy o
imp o e p ecision.4
The inding ha income declines a e mo e p onounced in households wi h highe
p e-sepa a ion income is consis en wi h p e ious esea ch. Fo example, Fishe and
Low (2016) show ha women om he highes -income households expe ience he
la ges and mos pe sis en declines in hei s anda d o li ing ollowing sepa a-
ion. These high-income households may he e o e appea mo e ulne able. How-
e e , he sha pe d op in income among high-income households likely e lec s he
highe ea nings o he pa ne who lea es, a he han g ea e ulne abili y. By con-
as , he mo e immedia e and se e e educ ions in ood pu chases and body weigh
among low-income households sugges ha hese households a e less able o smoo h
consump ion and may in ac bemo e exposed o he economic shock o sepa a-
ion. Suppo ing his in e p e a ion, I ind ha he p obabili y o being unde weigh
inc eases only among low-income households. In highe -income households, he
shi in BMI dis ibu ion is d i en by a educ ion in he p obabili y o being o e -
weigh o obese, a he han an inc ease in unde weigh s a us. See Fig.15 in he
Appendix. This sugges s ha weigh loss among low-income households may be
less olun a y and mo e likely o e lec ma e ial ha dship. This pa e n is consis en
wi h di e ences in sa ing capaci y. Low-income households may ha e ewe inan-
cial bu e s and less oom o ealloca e spending. In con as , high-income house-
holds may ha e accumula ed asse s ha help cushion he e ec s o sepa a ion. These
esul s unde sco e he impo ance o looking beyond income o consump ion-based
measu es when e alua ing changes in well-being and iden i ying he mos ulne -
able popula ions.
3.3.2 He e ogenei y o  ea men e ec s in ela ion o hep esence o child en
Figu e3 shows esul s sepa a ely o couple-only households (le -hand panels) and
households wi h child en unde age 18 a he ime o sepa a ion ( igh -hand pan-
els). Income declines by a ound 20% in couple-only households, compa ed o a 40%
decline in households wi h child en. Food pu chases also all mo e sha ply in amily
households, wi h educ ions o up o 40%, whe eas declines a e mo e p og essi e
in couple-only households. Women’s body weigh dec eases signi ican ly in bo h
g oups, bu he e ec is almos wice as la ge in households wi h child en. O e all,
he e ec s o sepa a ion on women a e mo e p onounced when child en a e p esen ,
al hough he di e ence is s a is ically signi ican only o income.5
Households wi h child en a e mo e likely o all wi hin he bo om 30% o he
p e-sepa a ion income dis ibu ion han couple-only households. When examining
ea men e ec he e ogenei y by amily ype wi hin income g oups, I ind ha he
4 S a is ical signi icance is assessed using an in e ac ion model ha includes ei he e en ime dummies
o a single pos -sepa a ion dummy, in e ac ed wi h a dummy o households in he op 30% o he p e-
sepa a ion pe capi a income dis ibu ion. A di e ence is conside ed s a is ically signi ican i he in e ac-
ion e m is signi ican a he 5% le el.
5 S a is ical signi icance is assessed using an in e ac ion model ha includes ei he e en ime dummies
o a single pos -sepa a ion dummy, in e ac ed wi h a dummy o amily composi ion g oups. A di e ence
is conside ed s a is ically signi ican i he in e ac ion e m is signi ican a he 5% le el.
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 17 o 56 26
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
Couple only households
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
Households wi h child en
A. Income
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
Couple only households
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
Households wi h child en
B. Food quan i ies pu chased
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
Couple only households
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
Households wi h child en
C. Spouse's body weigh index
Di e ence ela i e o -1
Yea s since b eak-up
Couple only households
Di e ence ela i e o -1
Yea s since b eak-up
Households wi h child en
D. Sha e o unheal hy ood pu chased
Fig. 3 T end in ou come a iables a ound sepa a ion, by amily composi ion (p esence o child en). The igu e
shows e en ime coe icien s o he ea men g oup o households whe e he male pa ne lea es he emale
pa ne , ela i e o he con ol g oup o households whe e no sepa a ion occu s and ela i e o he yea jus
be o e sepa a ion (
j
=−
1
). The le -hand panels p esen esul s o couple-only households, while he igh -
hand panels show esul s o households wi h child en. The dependen a iables a e he loga i hm o income,
ood quan i ies pu chased, emaining pa ne ’s BMI, and he sha e o unheal hy ood p oduc s pu chased by
household i in yea . In addi ion o household and yea ixed e ec s and household size, all eg essions include
dummies o bo h pa ne s’ ages. The da a co e he pe iod om 2005 o 2015. The 95% con idence in e als
a e based on s anda d e o s clus e ed a he le el o he household
J.Mink
26 Page 18 o 56
impac s o sepa a ion emain gene ally la ge in households wi h child en han in
couple-only households. Howe e , he es ima es a e imp ecise, and he di e ences
a e no s a is ically signi ican . See Fig.16 in he Appendix.
Some o he obse ed di e ences in ood pu chases could e lec cus odial
a angemen s. Child en may be egis e ed in he household e en i hey do no
eside he e ull- ime. In F ance, ollowing sepa a ion, 76% o child en a e p ima -
ily ca ed o by hei mo he , 9% by hei a he , and 15% li e in al e na ing esi-
dence a angemen s. E en when no he p ima y ca egi e , he non-cus odial pa en
usually has egula isi a ion, o en including weekends (Bonne e al. 2015). The
a e age numbe o meals ea en a home inc eases sligh ly a e sepa a ion (see col-
umn 1 in Table9 and he sec ion on mechanisms below), bu his may be an impe -
ec p oxy o child en’s p esence and consump ion. While his may in oduce some
measu emen e o , he s onge e ec on women’s body weigh in households wi h
child en sugges s a genuine educ ion in o e all calo ie in ake, consis en wi h a
decline in ood pu chases ha exceeds changes in household needs. Taken oge he ,
hese esul s poin o heigh ened ulne abili y among single-pa en households ol-
lowing sepa a ion.
3.3.3 He e ogenei y o  ea men e ec s byp e‑sepa a ion employmen s a us
Figu e17 in he Appendix shows he e ec s o sepa a ion by he woman’s p e-sep-
a a ion employmen s a us. The le -hand panels show esul s o women who we e
employed be o e sepa a ion, and he igh -hand panels o hose who we e no .
The decline in income is g ea e among employed women, while he d op in ood
pu chases is mo e p onounced among women who we e no employed be o e sepa-
a ion. This pa e n is consis en wi h dual-ea ne households expe iencing a sha pe
loss in income due o he depa u e o a high-ea ning pa ne , assuming simila pa -
ne ea nings, while inac i e women may ace g ea e di icul y adjus ing, as e lec ed
in deepe cu s o ood consump ion.
The e is li le di e ence in BMI loss ac oss he wo g oups. The sha e o
unheal hy ood pu chases declines among households whe e he woman was
employed, while i emains la gely unchanged in households whe e she was inac i e.
These di e ences should be in e p e ed wi h cau ion, as none a e s a is ically sig-
ni ican .6 Labou ma ke pa icipa ion is also likely o change in esponse o sepa-
a ion and is discussed below as a po en ial mechanism o mi iga ing declines in
li ing s anda ds.
3.3.4 He e ogenei y in ea men e ec s due ounobse ed ac o s
T ea men e ec s may a y no only along obse able cha ac e is ics bu also in
unobse ed ways. I his is he case, he main es ima es e lec a weigh ed a e age
6 S a is ical signi icance is assessed using an in e ac ion model ha includes ei he e en ime dummies
o a single pos -sepa a ion dummy, in e ac ed wi h a dummy o p e-sepa a ion employmen s a us. A
di e ence is conside ed s a is ically signi ican i he in e ac ion e m is signi ican a he 5% le el.
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 19 o 56 26
o he unde lying e ec s, whe e he weigh s may be unin ui i e o undesi able. In
pa icula , some weigh s may be nega i e, po en ially biasing he a e age ea -
men e ec (Sun and Ab aham 2021; De Chaisema in and d’Haul oeuille 2020).
To add ess his conce n, I apply he app oach p oposed by De Chaisema in and
d’Haul oeuille (2020), which uses no -ye - ea ed uni s and a pa allel ends
assump ion o es ima e ea men e ec s by g oup be o e a e aging hem. This
me hod yields esul s ha a e quali a i ely iden ical o hose o he main analysis, as
shown in Fig.18 in he Appendix.
3.3.5 Mechanisms: ea ing ou  e susea ing a home, labou ma ke pa icipa ion,
and epa ne ing
Se e al mechanisms may help explain he obse ed changes in li ing s anda ds ol-
lowing sepa a ion. One conce n is ha he decline in ood pu chases migh e lec a
shi om ea ing a home o ea ing ou , a he han a ue d op in ood consump ion.
Al hough ood pu chased ou side he home is no eco ded in he da a, I obse e he
a e age numbe o meals ea en a home pe week, which se es as a p oxy o meals
consumed a home e sus ou side he home. As shown in column 1 o Table9, his
numbe inc eases a e sepa a ion. This sugges s ha he decline in ood pu chases
is no d i en by subs i u ion owa d ood ou side he home. I any hing, ea ing ou
appea s o dec ease, possibly due o i s highe cos and lowe appeal when dining
alone. The concu en d op in BMI u he suppo s he in e p e a ion ha o al ood
in ake declines a e sepa a ion.
Labou ma ke pa icipa ion may also se e as an adjus men mechanism. As
shown in column 2 o Table 9, he p obabili y ha he woman emains inac i e
alls a e sepa a ion. Abo e, I showed ha being employed p io o sepa a ion does
no p o ec women om declines in income o ood consump ion. I now examine
whe he en e ing he labou ma ke a e sepa a ion helps mi iga e hese e ec s. Fig-
u e19 shows ou come ajec o ies o households in which he woman en e s he
labou ma ke ollowing sepa a ion. While es ima es a e less p ecise due o smalle
sample sizes, he di ec ion o e ec s sugges s ha labou ma ke en y does no
ully o se he economic consequences o sepa a ion. Household income declines
by app oxima ely 40–50%, and BMI dec eases by 2.5% up o 5yea s a e sepa a-
ion. Food pu chases appea o decline by a ound 25%, al hough hese coe icien s
a e imp ecisely es ima ed. These esul s should be in e p e ed wi h cau ion, as hey
likely e lec selec ion. Women who become economically ac i e may be hose mos
nega i ely a ec ed, which sugges s ha emaining inac i e would ha e led o e en
poo e ou comes. As such, he es ima es should be iewed as desc ip i e e idence
a he han causal e ec s.
Repa ne ing is ano he po en ial adjus men mechanism. Column 3 o Table9
shows ha he p obabili y o en e ing a new ela ionship inc eases in yea s 2 and 3
a e sepa a ion. To assess whe he epa ne ing helps mi iga e he nega i e e ec s
o sepa a ion, I es ima e ea men e ec s sepa a ely o households whe e a new
pa ne joins. Figu e20 shows ha in households whe e a new couple o ms, he
nega i e e ec s o sepa a ion a e less pe sis en . Income s ill declines by abou 20%
in he i s 2 yea s a e sepa a ion bu e u ns o p e-sepa a ion le els he ea e .
J.Mink
26 Page 20 o 56
Food pu chases also appea o all in epa ne ed households, bu es ima es a e no
s a is ically signi ican ly di e en om 0 a e yea 3. While en e ing he labou
ma ke does no appea o mi iga e he decline in li ing s anda ds, o ming a new
pa ne ship is associa ed wi h a quicke eco e y in income. This inding is con-
sis en wi h p e ious esea ch, such as Page and S e ens (2004), which shows ha
ema iage educes he long- e m economic cos s o sepa a ion. Howe e , as wi h
labou ma ke pa icipa ion, epa ne ing is likely subjec o selec ion, and hese
esul s should be in e p e ed as desc ip i e e idence only.
3.4 Discussion
This s udy inds ha sepa a ion leads o pe sis en declines in he li ing s anda ds o
women in F ance. Disposable income alls by a ound 30%, ood pu chases d op by
20% o 40%, and body weigh declines, consis en wi h educed calo ic in ake. These
changes pe sis o se e al yea s and a e no d i en by composi ional e ec s. Reduc-
ions a e pa icula ly p onounced in low-income and single-pa en households,
which also show a highe p obabili y o unde weigh s a us pos -sepa a ion. Labou
ma ke pa icipa ion o e s limi ed p o ec ion. Women who we e employed be o e
sepa a ion expe ience smalle educ ions in ood pu chases bu s ill ace losses.
The e is sugges i e e idence ha epa ne ing may help mi iga e some e ec s, bu
es ima es a e imp ecise due o he small numbe o obse ed cases. The e idence
sugges s ha bo h pa ne income loss and loss o economies o scale play a ole,
and ha nei he ans e s no own ea nings a e su icien o p e en he decline in
li ing s anda ds pos -sepa a ion.
These indings a e b oadly consis en wi h exis ing li e a u e. In he Uni ed S a es,
Page and S e ens (2004) documen a 50% d op in disposable income and a 35%
decline in ood spending a e sepa a ion. A eason he es ima ed income e ec s
may appea mo e modes in his s udy is measu emen e o . Income is epo ed in
bins, which likely in oduces a enua ion bias and pushes es ima es owa d ze o. As
a esul , he es ima ed 30% d op in income may be a lowe bound. This also cau ions
agains using my esul s o in e an income elas ici y o ood demand. A 40% d op
in ood pu chases om a 30% income all would imply an elas ici y g ea e han
one, which is inconsis en wi h he empi ical li e a u e, whe e elas ici ies ypically
all be ween ze o and one. Ano he possible eason is ha he e ec s a e o e all
smalle in F ance due o mo e gene ous public ans e s. Bonne e al. (2021) epo
a disposable income decline o 18.5% o women, while Cos emalle (2017) inds
declines o 14–25%, depending on ma i al s a us.
The e ec s o sepa a ion in F ance appea o las longe han in he US. I ind
li le e idence o eco e y o e ime, whe eas Page and S e ens (2004) epo ha
ood pu chases pa ially eco e wi hin 6 yea s. Page and S e ens (2004) a ib-
u e his eco e y mainly o ema iage. I a ely obse e a new pa ne joining he
household, which may explain he lack o agg ega e eco e y in my sample. Fo
households ha do epa ne , I also ind ha income e e s o p e-sepa a ion le els
4yea s a e sepa a ion.

B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 21 o 56 26
The decline in body weigh is consis en wi h indings om some p e ious s ud-
ies (Lee e al. 2004; Eng e al. 2005), hough o he s epo weigh gain pos -sepa a-
ion (Ma a e al. 2018). I ind no compa able es ima es o changes in he nu i ional
quali y o pu chases. A ew s udies ha e explo ed pos -sepa a ion die a y changes,
hough ypically ocusing on speci ic i ems (Lee e al. 2004; Vin he e al. 2016).
Al hough he main esul s e lec ou comes o women and child en, since mos
pos -sepa a ion households in he da a a e headed by women, I also ind modes
e ec s o men. Income alls by a ound 8% and ood pu chases decline by a ound
18%. This is consis en wi h a b oad li e a u e showing mo e se e e and consis -
en pos -sepa a ion impac s o women, as in (Ho man 1977; Duncan and Ho man
1985b; Bianchi and McA hu 1991; Holden and Smock 1991; McLanahan and San-
de u 1994; Pe e son 1996; Gala neau and S u ock 1997; McKee e and Wol in-
ge 2001; A ella and Smock 2005; Tach and Eads 2015), wi h mo e he e ogeneous
and gene ally weake e ec s o men (Smock 1994; Gala neau and S u ock 1997;
McManus and DiP e e 2001; Bonne e al. 2021).
A key ques ion is whe he he esul s can be in e p e ed as causal e ec s o sep-
a a ion. The s udy design suppo s his in e p e a ion in se e al ways. Household
ixed e ec s accoun o all ime-in a ian di e ences, and ime ixed e ec s con-
ol o common mac oeconomic shocks. Robus ness checks show ha he indings
a e no d i en by con ol g oup selec ion, ime windows, o sample composi ion.
Placebo es s using andomly assigned sepa a ion da es e eal no simila pa e ns,
sugges ing he obse ed e ec s a e unlikely o be due o chance. I also ind no sim-
ila declines when ano he household membe , a he han he pa ne , lea es he
household. Some bias om unobse ed, ime- a ying shocks canno be ully uled
ou . Fo example, sudden heal h e en s could a ec income, die , and ela ionship
s abili y simul aneously. Howe e , he absence o p e- ends, he sha p iming and
pe sis ence o he changes poin o sepa a ion as he d i e o he esul s.
Ano he cen al ques ion is whe he he obse ed e ec s a e ele an om a
policy pe spec i e. One in e p e a ion is ha he decline in li ing s anda ds ollow-
ing sepa a ion is so se e e ha many households canno main ain minimum con-
sump ion le els o mee hei nu i ional needs. This is e lec ed in sha p educ ions
in ood pu chases and measu able weigh loss. These indings may seem su p is-
ing. One migh expec ha in a coun y wi h a well-de eloped wel a e sys em like
F ance, public and p i a e ans e s would a leas p e en households om alling
below subsis ence h esholds. I ha is no he case, policy in e en ion may be
jus i ied bo h on e hical g ounds and o p e en b oade socie al cos s a ising om
unme basic needs.
An al e na i e explana ion is ha some women may in en ionally lose weigh
a e sepa a ion, o example o imp o e hei chances o inding a new pa ne . This
could help explain he decline in he p obabili y o being o e weigh and he ise
in he p obabili y o being o no mal weigh . Howe e , i does no accoun o he
inc ease in unde weigh s a us, excep in a e cases such as ea ing diso de s. In addi-
ion, he s onges educ ions in ood pu chases and body weigh a e obse ed among
low-income and single-pa en households, which a e a guably less able o smoo h
consump ion a e an income shock. In hese g oups, he inc ease in unde weigh
p e alence is no ma ched by simila pa e ns among highe -income households,
J.Mink
26 Page 22 o 56
whe e weigh changes appea o e lec a shi om o e weigh o no mal weigh .
These pa e ns sugges ha , a leas o some low-income households, he educ-
ions in consump ion and body weigh a e d i en by inancial cons ain s a he han
p e e ence.
Financial p essu e may also in luence beha iou indi ec ly. Fo women in eco-
nomically insecu e households, weigh loss may be seen less as a pe sonal goal and
mo e as a s a egy o imp o e p ospec s on he da ing ma ke , especially i epa ne -
ing is iewed as he only ealis ic way o es o e li ing s anda ds. P io s udies ha e
ound ha households o en egain hei p e-sepa a ion s anda d o li ing only a e
epa ne ing (Page and S e ens 2004), and I show sugges i e e idence o he same.
Women om highe -income households may ace less u gency o e-en e he da ing
ma ke , gi en g ea e inancial secu i y. These dynamics sugges ha some beha -
iou al changes a e sepa a ion may e lec economic necessi y a he han pe sonal
p e e ence.
Men al heal h could also play a ole. Sepa a ion may lead o dep ession o ela ed
challenges ha educe appe i e and ood in ake. I men al heal h e ec s a e mo e
se e e in inancially s ained households, his would ein o ce he in e p e a ion ha
economic ulne abili y is cen al o he obse ed ou comes.
Taken oge he , hese indings indica e ha he s onge e ec s obse ed in low-
income households a e mo e likely o e lec educed li ing s anda ds han olun a y
li es yle changes. The ac ha body weigh loss eaches unheal hy le els in hese
households, alongside a ise in unde weigh p e alence, poin s o unme basic needs.
I he decision o seek a new pa ne is shaped by inancial cons ain s, his aises
conce ns abou cons ained au onomy and economic dependence a e sepa a ion.
Some indi iduals may eel unable o lea e unsa is ac o y ela ionships o may en e
new ones p ima ily o economic su i al. Simila ly, i men al heal h e ec s like
dep ession a e con ibu ing o lowe ood in ake, and a e hemsel es exace ba ed by
economic s ain, his p o ides an addi ional eason o policy in e en ion.
The indings aise conce ns abou he capaci y o wel a e sys ems in high-income
coun ies o p o ec households om alling below basic li ing s anda ds a e sepa-
a ion. They also show he impo ance o looking beyond income when assessing
he e ec s o sepa a ion. Changes in ood consump ion, body weigh , and die qual-
i y o e mo e di ec indica o s o households’ abili y o mee essen ial needs. The
esul s sugges ha some sepa a ed households, especially hose wi h low incomes
o headed by single pa en s, s uggle o main ain adequa e nu i ion, wi h po en ially
se ious implica ions o heal h and social policy.
Policy in e en ion may be jus i ied no only on e hical g ounds bu also o eco-
nomic and public heal h easons. The obse ed ise in unde weigh p e alence and
he inc ease in he sha e o unheal hy ood p oduc s pu chased poin o a de e io a-
ion in die quali y, which may lead o poo e heal h and lowe p oduc i i y o e he
long e m. While some indi iduals shi om o e weigh in o he no mal weigh
ange, his canno be seen as a heal h gain when i e lec s in olun a y weigh loss
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 23 o 56 26
and wo sening nu i ion. I would be misguided o iew educed pu chasing powe
as bene icial simply because i lowe s calo ie in ake.
A policy p io i y could be o inc ease suppo o single-pa en households. Chil-
d en, hough no esponsible o hei amily si ua ion, a e o en pa o he mos
se e ely a ec ed households. Wi hou adequa e suppo , hey isk las ing disad an-
ages in heal h, educa ion, and well-being, wi h long- e m cos s o bo h indi iduals
and socie y. Mo e b oadly, s eng hening he social sa e y ne would help ensu e ha
all households can mee basic needs a e sepa a ion.
4 Conclusion
This s udy p o ides new e idence on he long- e m e ec s o couple sepa a ion on
women’s li ing s anda ds in F ance, wi h a pa icula ocus on ood consump ion
and nu i ional well-being. Using de ailed panel da a and an e en -s udy design, I
show ha sepa a ion leads o sha p and pe sis en declines in disposable income,
ood pu chases, and women’s body weigh . These e ec s a e especially p onounced
in low-income and single-pa en households, whe e he isk o unde weigh s a us
inc eases.
The indings challenge he iew ha wel a e sys ems in high-income coun-
ies ully shield households om ma e ial ha dship. In F ance, despi e a gene ous
sa e y ne , some households s uggle o main ain adequa e nu i ion a e sepa a-
ion. Labou ma ke pa icipa ion o e s limi ed p o ec ion, and while epa ne ing
is associa ed wi h pa ial eco e y, i may also e lec cons ained choices shaped by
economic need.
Mo e b oadly, he esul s highligh he impo ance o looking beyond income
when assessing he impac o sepa a ion. Using indica o s such as ood quan i ies
pu chased, die quali y, and body weigh p o ides e idence ha some households
all below subsis ence h esholds ollowing sepa a ion, wi h po en ial long- e m
consequences o heal h and well-being.
Ensu ing ha households can mee hei basic needs a e sepa a ion is a ma -
e o public heal h, social jus ice, and economic e iciency. S eng hening suppo
o low-income, single-pa en amilies and imp o ing en o cemen o exis ing sup-
po mechanisms could help p e en nu i ional ha dship and p omo e mo e esilien
household ajec o ies ollowing amily dissolu ion.
Appendix
See Tables1, 2, 3, 4, 5, 6, 7, 8, 9 and Figs.4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16,
17, 18, 19 and 20
J.Mink
26 Page 24 o 56
Table 1 Summa y s a is ics, pooled household-yea obse a ions
Summa y s a is ics o pooled household-yea obse a ions using all a ailable household obse a ions,
compa ed o he sample es ic ed o households obse ed o a leas 7 consecu i e yea s. The las col-
umn shows mean di e ences wi h s anda d e o s in pa en heses.
Hh. obse ed
≥
7yea s All households Mean di . (s d. e .)
Mean (s d. de .) n Mean (s d. de .) n
Household income (EUR) 2667.16 80,764 2646.61 183,870 20.56
(1413.96) (1419.32) (5.985)
Food quan i y (kg) 764.84 80,764 624.82 183,870 140.0
(430.38) (435.28) (1.831)
BMI o emale pa ne 25.09 76,876 24.92 173,027 0.172
(4.74) (4.85) (0.0209)
Sha e o unheal hy ood
pu chases
0.19 80,764 0.21 183,870 −0.0202
(0.11) (0.12) (0.000502)
Household size 2.51 80,764 2.61 183,870 −0.0997
(1.32) (1.37) (0.00572)
Age pa ne 1 53.04 80,764 47.66 183,870 5.377
(14.85) (15.33) (0.0641)
Age pa ne 2 54.48 80,764 49.1 183,870 5.380
(15.02) (15.47) (0.0647)
Pa ne 1 inac i e = 1 0.41 80,764 0.32 183,870 0.0870
(0.49) (0.47) (0.00200)
Pa ne 2 inac i e = 1 0.37 80,764 0.27 183,870 0.0978
(0.48) (0.44) (0.00192)
Household calo ie needs 3417.47 77,600 3508.97 174,641 −91.50
(1797.38) (1801.73) (7.767)
Meals a home pe day 2.15 79,975 2.17 175,762 −0.0180
(1.12) (1.17) (0.00492)
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 31 o 56 26
Table 7 E olu ion o he ou come a iables pe capi a, pe consump ion uni and in le els a ound he ime o sepa a ion
(1) (2) (3) (4) (5) (6) (7)
Pe capi a
income
Pe capi a
ood pu chases
Pe cons. uni
income
Pe cons. uni
ood pu chases
Income Food pu chases BMI
E en ime –7 −0.0315 −0.0325 −0.0313 −0.0323 −60.19 20.62 0.147
(0.030) (0.048) (0.031) (0.050) (81.801) (37.087) (0.158)
E en ime –6 −0.00160 −0.0628 −0.0000452 −0.0613 −12.37 −1.627 0.0271
(0.019) (0.041) (0.019) (0.041) (50.562) (24.755) (0.104)
E en ime –5 0.00833 −0.0408 0.0127 −0.0364 16.02 16.53 0.0473
(0.017) (0.041) (0.016) (0.041) (42.330) (22.447) (0.090)
E en ime –4 −0.00339 −0.0434 −0.000629 −0.0407 −8.836 −3.417 0.0440
(0.015) (0.033) (0.014) (0.033) (41.073) (19.468) (0.093)
E en ime –3 −0.00348 −0.0211 0.000145 −0.0175 5.510 15.47 −0.0142
(0.014) (0.028) (0.014) (0.028) (35.718) (15.230) (0.068)
E en ime –2 0.00809 −0.0445 0.00943 −0.0431 21.48 0.158 0.0396
(0.008) (0.026) (0.008) (0.026) (20.076) (12.855) (0.048)
E en ime –1 0 0 0 0 0 0 0
(.) (.) (.) (.) (.) (.) (.)
E en ime 0 −0.0719*** 0.0289 −0.0727*** 0.0281 −156.0*** −33.70** −0.121**
(0.010) (0.022) (0.010) (0.022) (24.885) (10.780) (0.040)
E en ime 1 −0.0883*** −0.103*** −0.105*** −0.120*** −482.7*** −113.2*** −0.345***
(0.016) (0.031) (0.015) (0.031) (37.774) (13.723) (0.058)
E en ime 2 −0.157*** −0.266*** −0.181*** −0.290*** −648.6*** −170.3*** −0.447***
(0.018) (0.039) (0.017) (0.039) (44.212) (15.618) (0.074)
E en ime 3 −0.169*** −0.258*** −0.196*** −0.285*** −682.4*** −165.8*** −0.322***
(0.021) (0.042) (0.021) (0.042) (53.712) (17.229) (0.091)

J.Mink
26 Page 32 o 56
Table 7 (con inued)
(1) (2) (3) (4) (5) (6) (7)
Pe capi a
income
Pe capi a
ood pu chases
Pe cons. uni
income
Pe cons. uni
ood pu chases
Income Food pu chases BMI
E en ime 4 −0.153*** −0.282*** −0.180*** −0.309*** −597.2*** −153.0*** −0.0344
(0.027) (0.059) (0.027) (0.059) (83.288) (23.354) (0.128)
E en ime 5 −0.158*** −0.359*** −0.182*** −0.384*** −600.1*** −190.8*** −0.0340
(0.037) (0.079) (0.037) (0.079) (122.775) (29.289) (0.176)
E en ime 6 −0.135*** −0.185** −0.163*** −0.214*** −581.6*** −153.2*** 0.102
(0.038) (0.061) (0.038) (0.061) (119.209) (30.407) (0.210)
E en ime 7 −0.102* −0.0931 −0.135** −0.127 −477.0** −108.0** −0.0958
(0.050) (0.074) (0.051) (0.075) (164.296) (38.282) (0.244)
E en ime 8 −0.157* −0.0957 −0.197** −0.136 −644.7*** −141.5* −0.338
(0.076) (0.079) (0.073) (0.082) (156.443) (57.626) (0.345)
Household size −0.295*** −0.198*** −0.227*** −0.129*** 145.5*** 114.3*** 0.0269
(0.004) (0.006) (0.004) (0.006) (11.383) (4.023) (0.020)
Obse a ions 80,860 80,860 80,860 80,860 80,860 80,860 77,111
R2
0.391 0.070 0.292 0.058 0.130 0.121 0.029
Yea FE Yes Yes Yes Yes Yes Yes Yes
Household FE Yes Yes Yes Yes Yes Yes Yes
Age dummies Yes Yes Yes Yes Yes Yes Yes
Robus s anda d e o s clus e ed a he le el o he household in pa en heses. The able shows e en ime coe icien s o he ea men g oup o households whe e he
male pa ne lea es he emale pa ne , ela i e o he con ol g oup o households whe e no sepa a ion occu s and ela i e o he yea jus be o e sepa a ion (
j
=−
1
). The
dependen a iables a e pe capi al o pe consump ion uni disposable household income, ood quan i ies pu chased, and disposable household income, household ood
quan i ies pu chased, he woman’s BMI measu ed in le els. In addi ion o household and yea ixed e ec s and household size, all eg essions include dummies o bo h
pa ne s’ ages. The da a co e he pe iod om 2005 o 2015.
*
p
<
0.05
, **
p
<
0.01
, ***
p
<
0.001
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 33 o 56 26
Table 8 E olu ion o ou comes a ound he ime o sepa a ion - obus ness o di e en model speci ica ions
Log income Log ood Log BMI Sha e unheal hy ood
E en ime -7 −0.00759 −0.00793 −0.00770 −0.0112 0.00830 0.00831 −0.0174** −0.0173**
(0.032) (0.032) (0.050) (0.051) (0.006) (0.006) (0.006) (0.006)
E en ime -6 0.0153 0.0148 −0.0474 −0.0508 0.00322 0.00322 −0.0186*** −0.0185***
(0.019) (0.019) (0.042) (0.043) (0.004) (0.004) (0.005) (0.005)
E en ime -5 0.0201 0.0208 −0.0270 −0.0272 0.00297 0.00300 −0.0168*** −0.0166***
(0.016) (0.016) (0.041) (0.041) (0.003) (0.003) (0.005) (0.005)
E en ime -4 0.00342 0.00405 −0.0352 −0.0347 0.00159 0.00161 −0.0159*** −0.0158***
(0.015) (0.015) (0.033) (0.033) (0.003) (0.003) (0.004) (0.004)
E en ime -3 −0.00431 −0.00299 −0.0180 −0.0156 −0.000761 −0.000717 −0.0110*** −0.0109***
(0.014) (0.014) (0.029) (0.029) (0.002) (0.002) (0.003) (0.003)
E en ime -2 0.00623 0.00679 −0.0446 −0.0444 0.00133 0.00136 −0.00685** −0.00676**
(0.008) (0.008) (0.026) (0.026) (0.002) (0.002) (0.002) (0.002)
E en ime -1 0 0 0 0 0 0 0 0
(.) (.) (.) (.) (.) (.) (.) (.)
E en ime 0 −0.0643*** −0.0651*** 0.0351 0.0348 −0.00480** −0.00484** 0.00179 0.00166
(0.010) (0.010) (0.022) (0.022) (0.002) (0.002) (0.002) (0.002)
E en ime 1 −0.188*** −0.180*** −0.189*** −0.188*** −0.0115*** −0.0110*** 0.00805* 0.00959**
(0.014) (0.014) (0.031) (0.031) (0.002) (0.002) (0.003) (0.003)
E en ime 2 −0.261*** −0.254*** −0.356*** −0.358*** −0.0154*** −0.0150*** 0.00540 0.00681
(0.016) (0.016) (0.038) (0.038) (0.003) (0.003) (0.004) (0.004)
E en ime 3 −0.272*** −0.266*** −0.352*** −0.356*** −0.0108** −0.0104** 0.0000122 0.00131
(0.020) (0.020) (0.041) (0.041) (0.003) (0.003) (0.005) (0.005)
E en ime 4 −0.259*** −0.252*** −0.370*** −0.373*** −0.000617 −0.000210 0.00647 0.00774
(0.027) (0.027) (0.060) (0.060) (0.005) (0.005) (0.006) (0.006)
J.Mink
26 Page 34 o 56
Table 8 (con inued)
Log income Log ood Log BMI Sha e unheal hy ood
E en ime 5 −0.276*** −0.268*** −0.476*** −0.474*** −0.00150 −0.00108 0.00598 0.00728
(0.039) (0.039) (0.078) (0.078) (0.007) (0.007) (0.008) (0.008)
E en ime 6 −0.271*** −0.263*** −0.332*** −0.329*** 0.00492 0.00536 −0.00502 −0.00365
(0.040) (0.040) (0.059) (0.059) (0.008) (0.008) (0.008) (0.008)
E en ime 7 −0.244*** −0.236*** −0.226** −0.224** −0.00206 −0.00163 −0.00728 −0.00595
(0.051) (0.051) (0.076) (0.076) (0.009) (0.009) (0.012) (0.012)
E en ime 8 −0.321*** −0.313*** −0.242** −0.242** −0.00901 −0.00855 −0.00988 −0.00843
(0.068) (0.067) (0.082) (0.084) (0.012) (0.012) (0.012) (0.012)
Household size 0.0697*** 0.180*** 0.00137* 0.00413***
(0.004) (0.006) (0.001) (0.001)
Consump ion uni 0.0113*** 0.0263*** 0.000268* 0.000816***
(0.001) (0.001) (0.000) (0.000)
Cons an 7.660*** 7.611*** 12.79*** 12.73*** 3.210*** 3.208*** 0.184*** 0.178***
(0.009) (0.011) (0.017) (0.020) (0.002) (0.002) (0.002) (0.002)
Obse a ions 80,860 80,860 80,860 80,860 77,111 77,111 80,860 80,860
R2
0.131 0.133 0.081 0.078 0.021 0.021 0.035 0.035
Yea FE Yes Yes Yes Yes Yes Yes Yes Yes
Household FE Yes Yes Yes Yes Yes Yes Yes Yes
Robus s anda d e o s clus e ed a he le el o he household in pa en heses. The able shows e en ime coe icien s o he ea men g oup o households whe e he
male pa ne lea es he emale pa ne , ela i e o he con ol g oup o households whe e no sepa a ion occu s and ela i e o he yea jus be o e sepa a ion (
j
=−
1
). The
dependen a iables a e he loga i hm o household disposable income, he loga i hm o ood quan i ies pu chased, he loga i hm o he woman’s BMI and he sha e o
unheal hy ood pu chases in o al ood pu chases. All eg essions include household and yea ixed e ec s. The da a co e he pe iod om 2005 o 2015.
*
p
<
0.05
, **
p
<
0.01
, ***
p
<
0.001
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 35 o 56 26
Table 9 E olu ion o meals ea en a home, emale pa ne ’s labo ma ke s a us, and epa ne ing s a us
a ound he ime o sepa a ion
Robus s anda d e o s clus e ed a he le el o he household in pa en heses. The able shows e en ime
coe icien s o he ea men g oup o households whe e he male pa ne lea es he emale pa ne , ela-
i e o he con ol g oup o households whe e no sepa a ion occu s and ela i e o he yea jus be o e
sepa a ion (
j
=−
1
). The dependen a iables a e he a e age numbe o meals ea en a home in a ypical
week and he labou ma ke s a us o he emale pa ne who emains in he household a e sepa a ion,
measu ed as a dummy a iable equal o one i she is inac i e.
Nb. meals Women is inac i e = 1 Women epa ne ed = 1
E en ime −7 0.0834 −0.00227 0.00210
(0.056) (0.024) (0.002)
E en ime −6 0.0175 −0.00466 0.00467**
(0.030) (0.023) (0.002)
E en ime −5 0.0678* 0.00164 0.00400**
(0.029) (0.018) (0.001)
E en ime −4 0.0306 −0.00663 0.00358**
(0.024) (0.014) (0.001)
E en ime −3 0.00975 −0.0105 0.000606
(0.014) (0.011) (0.001)
E en ime −2 0.0249 −0.000990 −0.0000905
(0.013) (0.006) (0.001)
E en ime −1 0 0 0
(.) (.) (.)
E en ime 0 −0.0767*** −0.0155** −0.0000642
(0.011) (0.005) (0.000)
E en ime 1 0.225*** −0.0320*** 0.0000181
(0.019) (0.007) (0.000)
E en ime 2 0.0816*** −0.0428*** 0.0506***
(0.019) (0.009) (0.009)
E en ime 3 0.0571** −0.0299** 0.0296**
(0.020) (0.012) (0.009)
E en ime 4 0.117* −0.0474*** 0.0281*
(0.055) (0.014) (0.013)
E en ime 5 0.146 −0.0210 0.0104
(0.096) (0.026) (0.016)
E en ime 6 −0.0116 −0.0339 0.00684
(0.048) (0.029) (0.015)
E en ime 7 −0.0172 −0.0258 −0.00855
(0.060) (0.041) (0.007)
E en ime 8 −0.0782 −0.0753 −0.000133
(0.124) (0.039) (0.008)
Cons an 0.901*** 0.445*** −0.000192
(0.004) (0.003) (0.000)
Obse a ions 80,027 80,860 80,860
R2
0.010 0.023 0.032
Yea ixed e ec s Yes Yes Yes
Household ixed e ec s Yes Yes Yes
J.Mink
26 Page 36 o 56
Fig. 4 T ends in ou come a iables a ound sepa a ion o a all households, ega dless o he sex o he
pa ne who emains, and b households in which he male pa ne emains a e sepa a ion. The igu e
shows e en ime coe icien s ela i e o he con ol g oup o households whe e no sepa a ion occu s and
ela i e o he yea jus be o e sepa a ion (
j
=−
1
). a Resul s o all households, while b shows esul s
o households whe e he emaining pa ne is male. The dependen a iables a e he loga i hm o
income, ood quan i ies pu chased, emaining pa ne ’s BMI, and he sha e o unheal hy ood p oduc s
pu chased by household i in yea . In addi ion o household and yea ixed e ec s and household size, all
eg essions include dummies o bo h pa ne s’ ages. The da a co e he pe iod om 2005 o 2015. The
95% con idence in e als a e based on s anda d e o s clus e ed a he le el o he household
▸

B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 37 o 56 26
J.Mink
26 Page 38 o 56
Fig. 5 T ends in he p obabili y o being unde weigh , no mal weigh , o e weigh o obese a ound sepa-
a ion. The igu e shows e en ime coe icien s o he ea men g oup o households whe e he male
pa ne lea es he emale pa ne , ela i e o he con ol g oup o households whe e no sepa a ion occu s
and ela i e o he yea jus be o e sepa a ion (
j
=−
1
). The dependen a iables a e dummy a iables
indica ing whe he he emale pa ne in household i in yea is unde weigh ( a iable equals 1 o a BMI
< 18.5 and 0 o he wise), o no mal weigh ( a iable equals 1 o a BMI
≥
18.5 and < 25 and 0 o he -
wise), o e weigh ( a iable equals 1 o a BMI
≥
25 and < 30 and 0 o he wise) o obese ( a iable equals
1 o a BMI > 30 and 0 o he wise). In addi ion o household and yea ixed e ec s and household size,
all eg essions include dummies o bo h pa ne s’ ages. The da a co e he pe iod om 2005 o 2015.
The 95% con idence in e als a e based on s anda d e o s clus e ed a he le el o he household
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 39 o 56 26
Fig. 6 T end in ou come a iables a ound sepa a ion using pe capi a o pe consump ion uni income
and ood pu chases. The igu e shows e en ime coe icien s o he ea men g oup o households
whe e he male pa ne lea es he emale pa ne , ela i e o he con ol g oup o households whe e no
sepa a ion occu s and ela i e o he yea jus be o e sepa a ion (
j
=−
1
). The dependen a iables a e
he pe capi a o pe consump ion uni income, ood quan i ies pu chased by household i in yea j. Con-
sump ion uni s a e cons uc ed by assigning a weigh o 1 o he i s adul , 0.7 o each addi ional pe son
aged 15 o olde , and 0.5 o each child unde 15, o e lec age-speci ic needs and economies o scale. In
addi ion o household and yea ixed e ec s and household size, all eg essions include dummies o bo h
pa ne s’ ages. The da a co e he pe iod om 2005 o 2015. The 95% con idence in e als a e based on
s anda d e o s clus e ed a he le el o he household
J.Mink
26 Page 40 o 56
Fig. 7 T end in ou come a iables a ound sepa a ion using he o iginal non- ans o med dependen a i-
ables. The igu e shows e en ime coe icien s o he ea men g oup o households whe e he male
pa ne lea es he emale pa ne , ela i e o he con ol g oup o households whe e no sepa a ion occu s,
and ela i e o he yea jus be o e sepa a ion (
j
=−
1
). The dependen a iables a e he income, ood
quan i ies pu chased, emaining pa ne ’s BMI, he sha e o unheal hy ood p oduc s pu chased in o al
quan i ies pu chased by household i in yea . In addi ion o household and yea ixed e ec s and house-
hold size, all eg essions include dummies o bo h pa ne s’ ages. The da a co e he pe iod om 2005
o 2015. The 95% con idence in e als a e based on s anda d e o s clus e ed a he le el o he house-
hold
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 47 o 56 26
Fig. 14 T end in ou come a iables a ound placebo da es o sepa a ion. The igu e shows he mean and
he 95 h pe cen ile o he coe icien dis ibu ions o placebo exe cises, ha is, e en ime coe icien s
es ima ed ela i e o he placebo con ol g oup o households and ela i e o he yea jus be o e he pla-
cebo da e o sepa a ion (
j
=−
1
). The uppe ou g aphs show he e ec s o assigning andom sepa a-
ion da es in households ha do no sepa a e, while he lowe ou g aphs show he e ec s o andomly
eassigning sepa a ion da es in he households ha sepa a e. The dependen a iables a e he loga i hm
o income, ood quan i ies pu chased, emale pa ne ’s BMI, and he sha e o unheal hy ood p oduc s
pu chased by household i in yea . In addi ion o household and yea ixed e ec s and household size, all
eg essions include dummies o bo h pa ne s’ ages. The da a co e he pe iod om 2005 o 2015. The
95% con idence in e als a e based on s anda d e o s clus e ed a he le el o he household

J.Mink
26 Page 48 o 56
Fig. 15 T end in he p obabili y o being unde weigh , no mal weigh o o e weigh o obese, by house-
hold income. The igu e shows e en ime coe icien s o he ea men g oup o households whe e he
male pa ne lea es he emale pa ne , ela i e o he con ol g oup o households whe e no sepa a ion
occu s and ela i e o he yea jus be o e sepa a ion (
j
=−
1
). The le -hand panels p esen esul s o
he 30% o households wi h he lowes p e-sepa a ion pe capi a incomes, while he igh -hand panels
show esul s o he 30% o households wi h he highes incomes. The dependen a iables a e dummy
a iables indica ing whe he he emale pa ne in household i in yea is unde weigh ( a iable equals 1
o a BMI < 18.5 and 0 o he wise), o no mal weigh ( a iable equals 1 o a BMI
≥
18.5 and < 25 and 0
o he wise), o e weigh ( a iable equals 1 o a BMI
≥
25 and < 30 and 0 o he wise), o obese ( a iable
equals 1 o a BMI > 30 and 0 o he wise). In addi ion o household and yea ixed e ec s and household
size, all eg essions include dummies o bo h pa ne s’ ages. The da a co e he pe iod om 2005 o
2015. The 95% con idence in e als a e based on s anda d e o s clus e ed a he le el o he household
▸
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 49 o 56 26
Di e ence ela i e o j -1
Yea s since b eak-up
30% poo es households
Di e ence ela i e o j -1
Yea s since b eak-up
30% iches households
A. Unde weigh
Di e ence ela i e o j -1
Yea s since b eak-up
30% poo es households
Di e ence ela i e o j -1
Yea s since b eak-up
30% iches households
B. No mal weigh
Di e ence ela i e o j -1
Yea s since b eak-up
30% poo es households
Di e ence ela i e o j -1
Yea s since b eak-up
30% iches households
C. O e weigh o obese
J.Mink
26 Page 50 o 56
Fig. 17 T end in ou come a iables a ound sepa a ion, by p e-sepa a ion employmen s a us o he
emale pa ne . The igu e shows e en ime coe icien s o he ea men g oup o households whe e he
male pa ne lea es he emale pa ne , ela i e o he con ol g oup o households whe e no sepa a ion
occu s and ela i e o he yea jus be o e sepa a ion (
j
=−
1
). The le -hand panel p esen s esul s o
households whe e he woman is employed and he igh -hand panel whe e he woman is no employed.
The dependen a iables a e he loga i hm o income, ood quan i ies pu chased, emale pa ne ’s BMI,
and he sha e o unheal hy ood p oduc s pu chased by household i in yea . In addi ion o household and
yea ixed e ec s and household size, all eg essions include dummies o bo h pa ne s’ ages. The da a
co e he pe iod om 2005 o 2015. The 95% con idence in e als a e based on s anda d e o s clus e ed
a he le el o he household
▸
Di e ence ela i e o -1
Yea s since b eak-up
30% poo es households, couples only households
Di e ence ela i e o -1
Yea s since b eak-up
30% iches households, couples only households
Di e ence ela i e o -1
Yea s since b eak-up
30% poo es households, households wi h child en
Di e ence ela i e o -1
Yea s since b eak-up
30% iches households, households wi h child e
n
A. Income
Di e ence ela i e o -1
Yea s since b eak-up
30% poo es households, couples only households
Di e ence ela i e o -1
Yea s since b eak-up
30% iches households, couples only households
Di e ence ela i e o -1
Yea s since b eak-up
30% poo es households, households wi h child en
Di e ence ela i e o -1
Yea s since b eak-up
30% iches households, households wi h child en
B. Food quan i ies pu chased
Di e ence ela i e o -1
Yea s since b eak-up
30% poo es households, couples only households
Di e ence ela i e o -1
Yea s since b eak-up
30% iches households, couples only households
Di e ence ela i e o -1
Yea s since b eak-up
30% poo es households, households wi h child en
Di e ence ela i e o -1
Yea s since b eak-up
30% iches households, households wi h child en
C. Spouse's body weigh index
Di e ence ela i e o -1
Yea s since b eak-u
p
30% poo es households, couples only households
Di e ence ela i e o -1
Yea s since b eak-u
p
30% iches households, couples only households
Di e ence ela i e o -1
Yea s since b eak-u
p
30% poo es households, households wi h child en
Di e ence ela i e o -1
Yea s since b eak-u
p
30% iches households, households wi h child en
D. Sha e o unheal hy ood pu chased
Fig. 16 T end in ou come a iables a ound sepa a ion, by amily composi ion and p e-sepa a ion income.
The igu e shows e en ime coe icien s o he ea men g oup o households whe e he male pa ne
lea es he emale pa ne , ela i e o he con ol g oup o households whe e no sepa a ion occu s and
ela i e o he yea jus be o e sepa a ion (
j
=−
1
). The panels show he esul s o sepa a e eg essions
o household g oups de ined by p e-sepa a ion household income and amily composi ion (couple-only
households o households wi h child en unde 18 a he ime o he sepa a ion). The dependen a iables
a e he loga i hm o income, ood quan i ies pu chased, emale pa ne ’s BMI, and he sha e o unheal hy
ood p oduc s in o al ood pu chased by household i in yea . In addi ion o household and yea ixed
e ec s and household size, all eg essions include dummies o bo h pa ne s’ ages. The da a co e he
pe iod om 2005 o 2015. The 95% con idence in e als a e based on s anda d e o s clus e ed a he
le el o he household
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 51 o 56 26
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
Spouse is ac i e when sepa a ing
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
Spouse is inac i e when sepa a ing
A. Income
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
Spouse is ac i e when sepa a ing
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
Spouse is inac i e when sepa a ing
B. Food quan i ies pu chased
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
Spouse is ac i e when sepa a ing
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
Spouse is inac i e when sepa a ing
C. Spouse's body weigh index
Di e ence ela i e o -1
Yea s since b eak-up
Spouse is ac i e when sepa a ing
Di e ence ela i e o -1
Yea s since b eak-up
Spouse is inac i e when sepa a ing
D. Sha e o unheal hy ood pu chased
J.Mink
26 Page 52 o 56
Fig. 18 T end in ou come a iables a ound sepa a ion, obus ness o unobse ed e ec he e ogene-
i y. The igu e shows coe icien s es ima ed using he app oach p oposed by De Chaisema in and
d’Haul oeuille (2020) o es ima es ha a e obus o unobse ed e ec he e ogenei y (implemen ed ia
he
did_mul ipleg
package). The app oach elies on no -ye - ea ed uni s and he pa allel ends assump-
ion o iden i y ea men e ec s o each ea ed uni ype, which can hen be a e aged oge he . The
dependen a iables a e he loga i hm o income, ood quan i ies pu chased, emale pa ne ’s BMI, and
he sha e o unheal hy ood p oduc s pu chased by household i in yea . In addi ion o household and
yea ixed e ec s and household size, all eg essions include dummies o bo h pa ne s’ ages. The da a
co e he pe iod om 2005 o 2015. The 95% con idence in e als a e based on s anda d e o s clus e ed
a he le el o he household

B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 53 o 56 26
Fig. 19 T end in ou come a iables a ound sepa a ion o he subg oup o households in which he
woman becomes ac i e in he labou ma ke a e sepa a ion. The igu e shows e en ime coe icien s
o he ea men g oup o households whe e he male pa ne lea es he emale pa ne , ela i e o he
con ol g oup o households whe e no sepa a ion occu s and ela i e o he yea jus be o e sepa a ion
(
j
=−
1
). The igu e p esen s esul s o he subg oup o households in which he emaining pa ne ( he
woman) becomes employed a e he sepa a ion. The dependen a iables a e he loga i hm o income,
ood quan i ies pu chased, he woman’s BMI, and he sha e o unheal hy ood p oduc s in o al ood pu -
chased by household i in yea . In addi ion o household and yea ixed e ec s and household size, all
eg essions include dummies o bo h pa ne s’ ages. The da a co e he pe iod om 2005 o 2015. The
95% con idence in e als a e based on s anda d e o s clus e ed a he le el o he household
J.Mink
26 Page 54 o 56
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
A
. Income
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
B. Foo
d
quan i ies pu c
h
ase
d
Pe cen age di e ence ela i e o -1
Yea s since b eak-up
C. Spouse's body weigh index
Di e ence ela i e o -1
Yea s since b eak-up
D. Sha e o unheal hy ood pu chased
Fig. 20 T end in ou come a iables a ound sepa a ion o he subg oup o households in which he
woman en e s a new ela ionship a e he sepa a ion. The igu e shows e en ime coe icien s o he
ea men g oup o households whe e he male pa ne lea es he emale pa ne , ela i e o he con ol
g oup o households whe e no sepa a ion occu s and ela i e o he yea jus be o e sepa a ion (
j
=−
1
).
The igu e shows esul s o he subg oup o households in which he women en e s a new ela ionship
a e he sepa a ion. The dependen a iables a e he loga i hm o income, ood quan i ies pu chased,
woman’s BMI, and he sha e o unheal hy ood p oduc s pu chased by household i in yea . In addi-
ion o household and yea ixed e ec s and household size, all eg essions include dummies o bo h
pa ne s’ ages. The da a co e he pe iod om 2005 o 2015. The 95% con idence in e als a e based on
s anda d e o s clus e ed a he le el o he household
Acknowledgemen This esea ch is suppo ed by he S a egic Resea ch Ini ia i e Nu iPe so om Uni-
e si y Pa is-Saclay and he p ojec AlimaSSen [ANR-14-CE20-0003-01] om he F ench Na ional
Resea ch Agency (ANR). Suppo by he Ge man Resea ch Founda ion (DFG) h ough CRC TR 224
(P ojec B07) and h ough Ge many’s Excellence S a egy (EXC 2126/1-390838866) is g a e ully
acknowledged.
Au ho con ibu ions JM made subs an ial con ibu ions o he concep ion o he wo k, he analysis and
in e p e a ion o da a; JM d a ed he wo k; JM app o ed he e sion o be published and ag ees o be
accoun able o all aspec s o he wo k in ensu ing ha ques ions ela ed o he accu acy o in eg i y o
any pa o he wo k a e app op ia ely in es iga ed and esol ed.
Funding Open Access unding enabled and o ganized by P ojek DEAL.
B oken homes andemp y pan ies: heimpac o couple sepa a ion… Page 55 o 56 26
Da a a ailabili y The da a canno be made publicly a ailable by he au ho as his s udy uses da a om a
p op ie a y consume panel which he au ho is no au ho ised o sha e. The au ho will p o ide in o ma-
ion on how o ob ain he da a om Kan a Wo ldpanel upon eques .
Decla a ions
Con lic o in e es The au ho s decla e no con lic o in e es .
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
A ella S, Smock PJ (2005) The economic consequences o he dissolu ion o cohabi ing unions. J Ma -
iage Fam 67(2):315–327
Bianchi SM, McA hu E (1991) Family dis up ion and economic ha dship: he sho - un pic u e o chil-
d en. US Bu eau o Census, Cu en Popula ion Repo Se ies, p 70
Bonne C, Ga bin i B, Solaz A, e al (2015) Les condi ions de ie des en an s ap ãs le di o ce. Technical
epo
Bonne C, Ga bin i B, Solaz A (2021) The lip side o ma i al specializa ion: he gende ed e ec o
di o ce on li ing s anda ds and labo supply. J Popul Econ 34:515–573
Cos emalle V (2017) Fo ma ions e up u es d’unions: quelles son les spéci ici és des unions lib es?
INSEE é é ences-F ance po ai social, pp 95–110
De Chaisema in C, d’Haul oeuille X (2020) Two-way ixed e ec s es ima o s wi h he e ogeneous ea -
men e ec s. Am Econ Re 110(9):2964–2996
de Sain Pol T (2006) Co ps e appa enance sociale: la co pulence en eu ope. Données sociales 2006-La
socié é ançaise, pp 649–656
De Vaus D, G ay M, Qu L, S an on D (2014) The economic consequences o di o ce in Aus alia. In J
Law Policy Fam 28(1):26–47
Duncan GJ, Ho man SD (1985a) Economic consequences o ma i al ins abili y. In: Ho izon al equi y,
unce ain y, and economic well-being, Uni e si y o Chicago P ess, pp 427–470
Duncan GJ, Ho man SD (1985b) A econside a ion o he economic consequences o ma i al dissolu-
ion. Demog aphy 22(4):485–497
Eng PM, Kawachi I, Fi zmau ice G, Rimm EB (2005) E ec s o ma i al ansi ions on changes in die-
a y and o he heal h beha iou s in us male heal h p o essionals. J Epidemiol Commun Heal h
59(1):56–62
Fishe H, Low H (2009) Sha ing li es, di iding asse s: an in e disciplina y s udy. Ha Publishing
Fishe H, Low H (2016) Reco e y om di o ce: compa ing high and low income couples. In J Law
Policy Fam 30(3):338–371
Gala neau D, S u ock J (1997) Family income a e sepa a ion. S a is ics Canada
Ho man S (1977) Ma i al ins abili y and he economic s a us o women. Demog aphy 14(1):67–76
Holden KC, Smock PJ (1991) The economic cos s o ma i al dissolu ion: Why do women bea a disp o-
po iona e cos ? Ann Re Sociol 17(1):51–78
INED (2018) Mo ali y in F ance. h ps:// www. ined. / / ou - sa oi - popul a ion/ chi es/ ance/ mo a li e-
cause- deces/ able- mo a li e
INSEE (2015) Couples e amilies. h ps:// www. insee. / / s a i s iqu es/ 20175 10? somma i e= 20175 28
INSEE (2016) 123 500 di o ces en 2014
J.Mink
26 Page 56 o 56
Lee S, Cho E, G ods ein F, Kawachi I, Hu FB, Coldi z GA (2004) E ec s o ma i al ansi ions on
changes in die a y and o he heal h beha iou s in us women. In J Epidemiol 34(1):69–78
Ma a J, Rich e D, Schneide T, He wig R (2018) How cohabi a ion, ma iage, sepa a ion, and di o ce
in luence BMI: a p ospec i e panel s udy. Heal h Psychol 37(10):948
McKee e M, Wol inge NH (2001) Reexamining he economic cos s o ma i al dis up ion o women.
Soc Sci Q 82(1):202–217
McLanahan S, Sande u G (1994) G owing up wi h a single pa en . Wha Hu s, Wha Helps. ERIC
McLanahan S, Tach L, Schneide D (2013) The causal e ec s o a he absence. Ann Re Sociol
39:399–427
McManus PA, DiP e e TA (2001) Lose s and winne s: he inancial consequences o sepa a ion and
di o ce o men. Am Sociol Re 66(2):246
Mille DL (2023) An in oduc o y guide o e en s udy models. J Econ Pe spec 37(2):203–230
OECD (2011) Doing be e o amilies
OECD (2017) Public spending on amily bene i s. h ps:// www. oecd. o g/ els/ amily/ da ab ase. h m
Page ME, S e ens AH (2004) The economic consequences o absen pa en s. J Hum Resou 39(1):80–107
Pe e son RR (1996) A e-e alua ion o he economic consequences o di o ce. Am Sociol Re
61:528–536
Smock PJ (1994) Gende and he sho - un economic consequences o ma i al dis up ion. Soc Fo ces
73(1):243–262
Sun L, Ab aham S (2021) Es ima ing dynamic ea men e ec s in e en s udies wi h he e ogeneous ea -
men e ec s. J Econom 225(2):175–199
Tach LM, Eads A (2015) T ends in he economic consequences o ma i al and cohabi a ion dissolu ion in
he uni ed s a es. Demog aphy 52(2):401–432
Vin he JL, Conklin AI, Wa eham NJ, Monsi ais P (2016) Ma i al ansi ions and associa ed changes in
ui and ege able in ake: indings om he popula ion-based p ospec i e epic-no olk coho , uk.
Soc Sci Med 157:120–126
Weiss RS (1984) The impac o ma i al dissolu ion on income and consump ion in single-pa en house-
holds. J Ma iage Fam 46:115–127
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.