B ady, Da id; Gao, Manjing; Gue a, Ch is ian; Kohle , Ul ich; Link, B uce
A icle — Published Ve sion
The long e m ela ionship be ween childhood Medicaid
expansions and se e e ch onic condi ions in adul hood
Social Policy & Adminis a ion
P o ided in Coope a ion wi h:
WZB Be lin Social Science Cen e
Sugges ed Ci a ion: B ady, Da id; Gao, Manjing; Gue a, Ch is ian; Kohle , Ul ich; Link, B uce (2024) :
The long e m ela ionship be ween childhood Medicaid expansions and se e e ch onic condi ions
in adul hood, Social Policy & Adminis a ion, ISSN 1467-9515, Wiley, Ox o d, Vol. 58, Iss. 1, pp.
39-60,
h ps://doi.o g/10.1111/spol.12942
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/313535
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 ps://c ea i ecommons.o g/licenses/by/4.0/
ORIGINAL ARTICLE
The long e m ela ionship be ween childhood
Medicaid expansions and se e e ch onic
condi ions in adul hood
Da id B ady
1
| Manjing Gao
2
| Ch is ian Gue a
2
|
Ul ich Kohle
3
| B uce Link
2
1
School o Public Policy, Uni e si y o
Cali o nia, Ri e side & WZB Be lin Social
Science Cen e , Ri e side, Cali o nia, USA
2
Depa men o Sociology, Uni e si y o
Cali o nia, Ri e side, Cali o nia, USA
3
Me hods o Empi ical Social Resea ch,
Uni e si y o Po sdam, Po sdam,
B andenbu g, Ge many
Co espondence
Da id B ady, School o Public Policy,
Uni e si y o Cali o nia, Ri e side & WZB
Be lin Social Science Cen e , INTS 4133,
900 Uni e si y A e., Ri e side, CA 92521,
USA.
Email: [email p o ec ed]
Funding in o ma ion
Na ional Ins i u e o Aging, G an /Awa d
Numbe s: 1R03AG062842-01A1,
5-R01AG040213-10; Na ional Ins i u es o
Heal h, G an /Awa d Numbe s: R01
HD069609, R01 AG040213; Na ional Science
Founda ion, G an /Awa d Numbe s:
1157698, 1623684; Eunice Kennedy Sh i e
Na ional Ins i u e o Child Heal h and Human
De elopmen , G an /Awa d Numbe s:
1-R03HD091871-01, 1-R03HD100924-01
[Co ec ions added on 18 July 2023, a e
i s online publica ion: Da id B ady's email
add ess has been upda ed in his e sion.]
Abs ac
We es whe he he expansions o child en's Medicaid
eligibili y in he 1980s–1990s esul ed in long- e m heal h
bene i s in e ms o se e e ch onic condi ions. S ill ela i ely
a e in he ield, we use p ospec i e indi idual-le el panel
da a om he Panel S udy o Income Dynamics (PSID) along
wi h he highe quali y income measu es om he C oss-
Na ional Equi alen File (adjus ing o axes, ans e s and
household size). We obse e se e e ch onic condi ions
(high blood p essu e/hea disease, cance , diabe es, o lung
disease) a ages 30–56 (a e age age 43.1) o 4670 espon-
den s who we e also p ospec i ely obse ed du ing child-
hood (i.e., a ages 0–17). Ou analysis exploi s wi hin- egion
empo al a ia ion in childhood Medicaid eligibili y and
adjus s o s a e- and indi idual-le el con ols. We uniquely
concen a e a en ion on adjus ing o childhood income. A
s anda d de ia ion g ea e childhood Medicaid eligibili y
signi ican ly educes he p obabili y o se e e ch onic con-
di ions in adul hood by 0.05 o 0.12 (16%–37.5% educ ion
om mean 0.32). Ac oss he ange o obse ed childhood
Medicaid eligibili y, he p obabili y is app oxima ely cu in
hal . G ea e childhood Medicaid eligibili y also subs an ially
educes childhood income dispa i ies in se e e ch onic
Recei ed: 5 Sep embe 2022 Re ised: 19 May 2023 Accep ed: 26 May 2023
DOI: 10.1111/spol.12942
This is an open access a icle unde he e ms o he C ea i e Commons A ibu ion License, which pe mi s use, dis ibu ion and
ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed.
© 2023 The Au ho s. Social Policy & Adminis a ion published by John Wiley & Sons L d.
Soc Policy Adm. 2024;58:39–60. wileyonlinelib a y.com/jou nal/spol 39
condi ions. A highe le els o childhood Medicaid eligibili y,
we ind no signi ican childhood income dispa i ies in adul
se e e ch onic condi ions.
KEYWORDS
ch onic condi ions, heal h dispa i ies, li e cou se, Medicaid, social
policy
In he la e 1970s, child en in he U.S. we e e y unlikely o be co e ed by Medicaid. By he mid-1990s, a much
g ea e sha e o child en could ecei e Medicaid o a mo e o hei childhoods. These expansions subs an ially
enla ged public heal h insu ance and heal hca e o low-income child en compa ed o wo decades ea lie (G ogan &
And ews, 2015; Kame man & Kahn, 2001; Ka z, 2001; Kouse , 2002). Al hough U.S. heal hca e emains much mo e
p i a ely-p o ided han in o he ich democ acies (B ady e al., 2016; Wend e al., 2009), Ame ican heal h policy
g ew ma kedly mo e inclusi e o child en –and especially low-income child en –o e he 1980s and 1990s.
Figu e 1shows his based on Mille and Whe y (Mille & Whe y., 2019) measu es s anda dised ac oss he
U.S. and applied o ou sample (see below). In 1979, he “a e age”Ame ican child was eligible o Medicaid o abou
wo- hi ds o one yea o hei en i e childhood. In he mos gene ous s a e, he a e age Ame ican child would ha e
been eligible o sligh ly mo e han 1.5 yea s. By he mid-1980s, he a e age Ame ican child would ha e been eligible
o abou 1.25 yea s and mo e han 2 yea s in some s a es. By he mid-1990s, he a e age Ame ican child was eligi-
ble o almos 2 yea s and abou 2.75 yea s in some s a es. Whe eas Medicaid eligibili y was almos uni o mly low
ac oss s a es in he 1970s, he e was a mo e a ia ion wi h a much highe a e age ac oss s a es by 1995.
FIGURE 1 Childhood medicaid eligibili y in yea s o ypical child ac oss s a es in he U.S., 1979–1995. [Colou
igu e can be iewed a wileyonlinelib a y.com]
40 BRADY ET AL.
This s udy builds on he ex ensi e and aluable li e a u e on he e ec s o Medicaid expansions. We speci ically
in es iga e he ela ionship be ween childhood Medicaid eligibili y (a ages 0–17) and adul se e e ch onic condi ions
(a ages 30–56). In a s ill ela i ely a e app oach, we use p ospec i e indi idual-le el panel da a om he Panel
S udy o Income Dynamics (PSID) me ged wi h highe quali y income da a om he C oss-Na ional Equi alen File
(adjus ing o axes, ans e s and household size). While ou s udy builds on ich p io esea ch, ou analysis o e s
wo ad an ages. Fi s , we use mo e comp ehensi e measu es o ch onic heal h condi ions and obse e hem a olde
ages. Second, ou models mo e sys ema ically adjus o p ospec i ely measu ed indi idual cha ac e is ics ac oss he
du a ion o childhood. As a esul , we can uniquely in o m how Medicaid expansions educed long- e m income dis-
pa i ies in heal h. By inno a ing in hese ways, we p o ide no el e idence by add essing limi a ions o pas esea ch.
In he p ocess, ou s udy also illumina es a ini ies be ween he Medicaid and compa a i e social policy li e a u es.
This includes bo h he measu es o social policy and o unde s anding how social policy migh explain c oss-na ional
di e ences in heal h.
1|PAST RESEARCH
The esea ch p og amme on he e ec s o Medicaid expansions has been e y p oduc i e. The ield p o ides many
igo ous s udies examining Medicaid's ini ial implemen a ion since he 1960s (e.g., Boud eaux e al., 2016;
Goodman-Bacon, 2018,2021), a ied expansions ac oss s a es in he 1980s and 1990s (e.g., Mille & Whe y., 2019;
O'B ien & Robe son, 2018), and he mos ecen expansion in he A o dable Ca e Ac (e.g., Kaes ne e al., 2017).
By now, he li e a u e al eady p o ides many excellen na a i es o he p og amme's his o ical e olu ion and he
p ecise policy changes in he 1980s-1990s (G ogan & And ews, 2015; Ka z, 2001). The li e a u e also al eady o e s
me iculous accoun s o po en ial mechanisms leading o a ious ou comes. Among o he mechanisms, Medicaid
access can in luence s ess, esou ce deple ion, heal h, and heal hy de elopmen , all o which could subsequen ly
in luence ch onic condi ions.
Many analyse how he 1980s-1990s expansions a ec ed sho e e m ou comes du ing childhood (e.g., Cu ie
e al., 2008; Cu ie & G ube , 1996a, Cu ie & G ube , 1996b; Jackson e al., 2021; Le ine & Schanzenbach, 2009).
O he s examine he longe e m e ec s o childhood Medicaid eligibili y on ou comes la e in adul hood (e.g., Le e e
e al., 2019). Rega ding he expansions in he 1980s and 1990s speci ically, schola s ha e linked Medicaid o many
bene icial ou comes (B own e al., 2020; Cohodes e al., 2016; Lip on e al., 2016; Mille & Whe y., 2019; O'B ien &
Robe son,2018; Thompson, 2017).
The p e ailing app oach o he la e li e adul consequences o Medicaid expansions in he 1980s-1990s is as
ollows (e.g., Cohodes e al., 2016; Cu ie & G ube , 1996a, Cu ie & G ube , 1996b; Cu ie e al., 2008; Eas
e al., 2023; Lip on e al., 2016). Schola s usually u ilise c oss-sec ional da a obse ed in adul hood ha con ains e -
ospec i e in o ma ion on he da e and loca ion o childbi h. F om his, indi iduals a e assigned he Medicaid eligi-
bili y ha exis ed in u e o o in childhood. This app oach is acili a ed by he well-es ablished measu es o Medicaid
eligibili y discussed below. In a compelling s udy, Mille and Whe y (2019) analyse he c oss-sec ional Na ional
Heal h In e iew Su ey and Cu en Popula ion Su ey a ages 19–36 in yea s 1998–2015. They hen e ospec-
i ely link ha da a o he Medicaid eligibili y whe e and when hose adul s we e in u e o. This app oach has indis-
pu ably been aluable. Howe e , because childhood is no p ospec i ely obse ed, one canno adjus o many o
he indi idual-le el di e ences be ween esponden s ha occu ed du ing childhood. Ins ead, hese s udies mainly
ely on ins umen al a iables o o se unobse ed con ounding.
I has been less common o use p ospec i e longi udinal da a on esponden s obse ed du ing bo h childhood
and adul hood. One ad an age o such p ospec i e longi udinal s udies (compa ed o he e ospec i e app oach) is
he capaci y o adjus o indi idual-le el di e ences du ing childhood. Using adminis a i e ax da a on 10 million
esponden s, B own e al. (2020) show childhood Medicaid eligibili y lead o g ea e access o Medicaid a ages
12–15, and his esul ed in g ea e college en olmen , wages and axes paid and lowe mo ali y and e ili y a ages
BRADY ET AL.41
19–28. A ew use p ospec i e indi idual-le el panel su eys like us (e.g., Jackson e al., 2021). These s udies obse e
a g ea e du a ion o childhood and mo e exhaus i ely adjus o di e ences in childhood cha ac e is ics. Using he
Na ional Longi udinal Su ey o You h 1979, Thompson (2017) inds childhood Medicaid eligibili y p edic s se e al
ou comes a ages 18–20 (sel - a ed heal h, limi ed in abili y o wo k/a end school, ch onic condi ions, and as hma
a acks). Using he PSID like us, bu analysing he ini ial ollou o Medicaid 1966–1982 (no he 1980s–1990s
expansions), Boud eaux e al. (2016) show Medicaid exposu e a ages 0–5p edic s heal h ou comes a ages 25–54.
2|INNOVATIONS BEYOND PAST RESEARCH
Ou analysis builds on his s ong esea ch p og amme by o e ing wo c i ical ad an ages. Fi s , we use mo e com-
p ehensi e measu es o ch onic heal h condi ions and obse e hem a olde ages (i.e., 30–56). Unlike p io esea ch,
we ocus mos on ou mo e salien and consequen ial and “se e e”ch onic condi ions –high blood p essu e/hea
disease, cance , diabe es, and lung disease (B ady e al., 2022; Link e al., 2008). Fu he analyses p ecisely es each
se e e and h ee less se e e condi ions, and obesi y (Appendix C). These ch onic condi ions a e salien gi en hey
p edic mo ali y, a e a sou ce o eno mous p i a e and public heal hca e cos s, cons ain employmen and p oduc-
i i y, and unde mine well-being.
Despi e clea con ibu ions, p io s udies use less comp ehensi e measu es o ch onic condi ions. Mille and
Whe y (2019) ind signi ican e ec s o a “ch onic disease index”o obesi y, diabe es, hea disease, and high blood
p essu e. Thompson (2017) does no ind obus ly signi ican e ec s o “any condi ion ha equi es equen medi-
cal a en ion, he egula use o medica ion, o he use o special equipmen ”o ha ing had an as hma a ack. Ana-
lysing he ini ial ollou o Medicaid 1966–1982, Boud eaux e al. (2016) analyse a ch onic condi ion index o high
blood p essu e, hea disease/hea a ack, adul onse diabe es, and obesi y. Unlike Mille and Whe y (2019) and
Boud eaux e al. (2016), we include cance and lung disease. Unlike Mille and Whe y (2019) and Boud eaux e al.
(2016), we analyse obesi y sepa a ely om ou main ou come (see Appendix C).
1
P e ious esea ch also shows i is essen ial o obse e esponden s a olde ages as ch onic condi ions usually
eme ge la e in li e (B ady e al., 2022). Some s udies analyse he mo ali y consequences o he ini ial ollou o
Medicaid and he e o e include olde esponden s (e.g., Goodman-Bacon, 2018,2021). Fo example, Boud eaux
e al. (2016) PSID sample includes 25–54 yea olds. Fo all he s udies o he 1980s–1990s expansions howe e , all
samples a e much younge han ou sample o 30–56 yea olds. (Mille & Whe y., 2019) sample is 19–36 yea s old,
Thompson (Thompson, 2017)is18–20 yea s old, B own e al. (2020)is19–28 yea s old, and Cohodes e al. (2016)is
22–29 yea s old. Thus, we examine he oldes –and a guably mos app op ia e –sample o any s udy o he 1980s–
1990s expansions.
2
Second, we adjus o p ospec i ely measu ed indi idual cha ac e is ics –and especially income –ac oss he
du a ion o childhood.
3
Doing so is jus i ied because o he as li e a u e showing childhood ci cums ances shape
long e m heal h ou comes (e.g., Duncan e al., 2012; Link e al., 2017). Using he PSID, Johnson and Schoeni (2011)
show amily income and po e y a ages 13–16 in luence sel - a ed heal h, as hma, hype ension, diabe es, s oke,
hea a ack, and hea disease a ages 39–56. Pe haps mos ele an is he la ge in e disciplina y li e a u e on he
“long a m o childhood”(e.g., Haywa d & Go man, 2004; Tu ne e al., 2016). Fo ins ance, using he PSID, B ady
e al. (2022) show ha p ospec i e high quali y income measu ed a ages 0–17 p edic s sel - a ed heal h, psychologi-
cal dis ess, hea a ack, s oke, and se e e ch onic condi ions a ages 40–65. Gi en ex ensi e esea ch shows child-
hood income a ec s adul heal h; omi ing childhood income om models could lead o biassed es ima es.
Mo eo e , bo h he ea ly sensi i e pe iod o childhood and adolescence a e plausibly in luen ial o subsequen
heal h. Hence, ou longe obse ed du a ion o childhood plausibly p o ides g ea e in o ma ion abou childhood
ci cums ances.
To he bes o ou knowledge, he li e a u e on Medicaid and adul ou comes mos ly omi s childhood income
om models. La gely, his is a by-p oduc o using con empo a y c oss-sec ional da a and e ospec i ely assigning
42 BRADY ET AL.
childhood Medicaid eligibili y. In he con empo a y c oss-sec ional su eys used (e.g., NHIS, ACS), e en e ospec i e
measu es o childhood ci cums ances a e no usually a ailable.
4
E en when p ospec i e childhood income measu es
a e a ailable howe e , some omi childhood income (e.g., Thompson, 2017). While no con olling o income du ing
childhood, Boud eaux e al. (2016) adjus o head's educa ion and o icial po e y du ing childhood and mo he 's
educa ion and ma i al s a us a bi h.
5
In he one s udy we could ind ha assesses obus ness when con olling o
income, B own e al.s’(2020) esul s a e qui e sensi i e o adjus ing o childhood income.
6
P io esea ch mos ly uses ins umen al a iables o a oid omi ed a iable bias om unobse ed cha ac e is ics
like childhood income. Following Cu ie & G ube , 1996a, Cu ie & G ube , 1996b, many use simula ed Medicaid eligi-
bili y o ins umen o ac ual Medicaid eligibili y. As well, when indi idual-le el da a a e used, po e y is o en used
o calcula e eligibili y (bu see en. 5). To he ex en he ins umen s a e exogenous; his app oach can es he causal
e ec o Medicaid eligibili y wi hou adjus ing o all ele an p edic o s. Howe e , a oiding omi ed a iable bias
wi h ins umen al a iables is only as c edible as he exclusion es ic ion.
I seems ai o aise ques ions abou whe he his s ong assump ion is c edible. The e a e ac ually se e al plau-
sible backdoo causal pa hways om simula ed childhood Medicaid eligibili y o childhood income. As one plausible
pa hway, childhood Medicaid eligibili y is a o m o wel a e s a e gene osi y (B ady e al., 2016; Reynolds &
A endano, 2018), and wel a e s a e gene osi y is endogenous o poli ical ins i u ions like powe esou ces and acial
egimes (e.g., G ogan & And ews, 2015; Hube & S ephens, 2001; Kame man & Kahn, 2001; Ka z, 2001;
Ko pi, 1989; Quadagno, 1994; Wend e al., 2009). In ac , in e s a e and empo al a ia ion in Medicaid expansions
in he 1980s and 1990s we e in luenced by such in e s a e and empo ally a ying ac o s (e.g., G ogan &
And ews, 2015; G ogan & Pa ashnik, 2003; Ka z, 2001; Kouse , 2002; Lan o d & Quadagno, 2022; Michene , 2018;
Quadagno, 2006). In e s a e and empo al a ia ion in such ac o s also in luences income and especially o he poo
(e.g., Bake , 2022; VanHeu elen & B ady, 2022). The e o e, such ac o s plausibly cause bo h Medicaid eligibili y and
income. In u n, he ins umen o childhood Medicaid eligibili y could be con ounded wi h childhood income. As a
complemen o p io analyses, i would be p oduc i e o es ima e models wi h a mo e comple e se o con ols.
Ano he –and pe haps e en s onge –mo i a ion o including and igo ously measu ing childhood income is
he capaci y o es o income dispa i ies. La gely, he li e a u e unde s andably p esumes Medicaid eligibili y should
ha e mo e powe ul e ec s on low-income child en. Indeed, many s udies es o he e ogeneous e ec s ac oss
income, po e y, pa en s' educa ion, o Medicaid eligibili y (e.g., Boud eaux e al., 2016; B own e al., 2020; Cu ie &
G ube , 1996a; Cu ie e al., 2008; Jackson e al., 2021; Mille & Whe y., 2019). The e o e, he e ogeneous e ec s
by income a e likely and wo h es ing. Howe e , hese p io es s o in e ac ions ha e been o ced o ely on less
igo ous measu es o amily backg ound han ou measu es o childhood income and amily backg ound.
7
As a esul ,
ou analyses can uniquely in e ac wi h and adjus o a mo e eliable and alid measu e o childhood income. The e-
o e, ou analyses can be e cla i y how Medicaid expansions in luence childhood income dispa i ies in long e m
heal h.
3|METHODS
We use he longi udinal, na ionally ep esen a i e Panel S udy o Income Dynamics (PSID), which has been ielded
annually 1968–1997 and biannually since. We also use he C oss-Na ional Equi alen File (CNEF), which p o ides a
supplemen a y se o highe quali y income a iables o he en i e sample o PSID esponden s (F ick e al., 2007).
The sample includes 4670 indi iduals who we e child en (0–17 yea s) in households in e iewed 1979–1995
and ollowed un il 30–56 yea s old. The 1979–1995 pe iod is de e mined by he a ailabili y o Mille and Whe y's
Medicaid measu e.
8
To obse e adul ch onic condi ions, we selec he las a ailable obse a ion o each espon-
den h ough he 2019 wa e. This cap u es esponden s a hei oldes age possible. 81.3% o obse a ions come
om 2019 and he mean age when ch onic condi ions a e obse ed is 43.1.
9
Appendix Ashows desc ip i e s a is ics
and Appendix Bshows bi a ia e co ela ions. We discuss all obus ness checks a he end o he esul s sec ion.
BRADY ET AL.43
3.1 |Adul ch onic condi ions
The PSID includes sel - epo s o whe he a doc o has diagnosed he esponden as ha ing any ch onic condi ions.
We ocus on a bina y measu e o any o ou se e e ch onic condi ions: high blood p essu e/hea disease, cance , dia-
be es, and lung disease (B ady e al., 2022; Link e al., 2008). 31.9% has a leas one se e e condi ion (see
Appendix A). Tha means 1399 o 4670 cases o he unweigh ed sample (see Appendix C). Appendix C epo s he
p e alence o each ch onic condi ion. The mos common is high blood p essu e/hea disease (23.9%), ollowed by
diabe es (8.6%), cance (4.4%), and lung disease (4.0%).
3.2 |Childhood Medicaid eligibili y
Ou key independen a iable is Mile and Whe y's (2019) syn he ic measu e o he pe cen age o child en eligible
o Medicaid (which upda es Cu ie & G ube , 1996a, Cu ie & G ube , 1996b). In each yea 1979–1995, hey d aw a
na ionally ep esen a i e sample om he Census Bu eau's Cu en Popula ion Su ey o 1000 child en in each o
ou age anges (1–4, 5–9, 10–14, and 15–18). They hen apply each s a e-yea 's eligibili y c i e ia and es ima e wha
sha e o ha na ional age g oup would be eligible o Medicaid i i esided in a gi en s a e. This simula es a
s anda dised pe cen o child en eligible as i all s a es had he same na ional demog aphic cha ac e is ics. Mille and
Whe y con e his in o ma ion o he numbe o yea s co e ed by Medicaid o each age g oup in a s a e-yea . We
ma ch each esponden o hei s a e-yea -age g oup. We hen a e age his measu e o e all obse ed yea s o child-
hood o measu e childhood Medicaid eligibili y (hence o h childhood Medicaid).
This measu e has se e al use ul ea u es. Fi s , i easonably cap u es in en o ea e ec s, which ealis ically
gauge how s a e-le el policy changes a ec indi iduals when ake-up and access a e always incomple e.
Policymake s ha e mo e con ol o e legisla ed eligibili y han he ac ual access o and en olmen in Medicaid
(Hein ich e al., 2022; He d & Moynihan, 2019). Second, because he measu e is simula ed based on a ixed na ion-
wide popula ion, his p e en s s a e-yea eligibili y con la ing popula ion needs wi h ac ual social policy gene osi y.
Thi d, a as compa a i e social policy li e a u e p o ides a p eceden o his app oach. Since a leas Ko pi (1989),
compa a i e social policy esea che s ha e cons uc ed measu es o wel a e gene osi y by indexing p og ammes o
“ ypical wo ke ”wages o “no ional household ypes”(Beck ield & Bamb a, 2016; Nelson e al., 2020; Sc uggs &
Ta oya, 2022). Simila ly, hese measu es assess how Medicaid eligibili y o he same na ionally-s anda dised “ ypical
child” a ies ac oss s a es and yea s.
Ideally, we would also measu e eceip o Medicaid a he indi idual-le el and/o ins umen o i wi h he eligi-
bili y measu e. Un o una ely, he PSID only began asking he necessa y ele an ques ions in 1999. See Appendix D
o a de ailed summa y o he PSID's Medicaid ques ions. The e o e, he PSID simply does no allow us o de e mine
whe he hese child en in he 1980s and 1990s we e ac ually en olled in Medicaid. Again, his means ou es ima es
a e in en - o- ea e ec s no ea men e ec s on he ea ed.
10
Because he childhoods o esponden s o di e en ages we e obse ed o di e en pe iods o ime
1979–1995, all models adjus o he numbe o yea s childhood Medicaid was obse ed (ne e s a is ically signi ican ).
Fo ins ance, a esponden who was 17 in 1980 could be obse ed o 2 yea s in ou da a (1979 and 1980) while a
esponden who was 17 in 1995 could be obse ed o 16 yea s (1979–1995).
3.3 |O he independen a iables
To op imise he measu emen o childhood income, we inco po a e leading s anda ds in in e na ional income mea-
su emen (B ady e al., 2018; Duncan e al., 2002; Jän i & Jenkins, 2015; Mazumde , 2016; Rainwa e &
Smeeding, 2003). We use he CNEF measu e o “pos - isc”equi alized household income, de ined as including
44 BRADY ET AL.
ma ke income, cash and nea cash ans e s (e.g., he Supplemen al Nu i ional Assis ance P og amme), and
sub ac ing axes, and adding ax c edi s (e.g., he Ea ned Income Tax C edi ). This measu e includes all household
membe s, and we equi alize household size by di iding by he squa e oo o HH membe s. To s anda dise income
o e ime, we con e income o ela i e ank pe cen iles in each yea . Fo una ely, he PSID-CNEF has a la ge
na ionally ep esen a i e sample each yea . Fo each esponden , we hen a e age hei ela i e ank pe cen iles o e
childhood (i.e., ages 0–17).
This measu e a ou pe o ms c ude measu es o income (o weal h, occupa ion o ea nings) as a p oxy o pe -
manen income (B ady e al., 2018). Inco po a ing hese s anda ds also esul s in highe es ima es o he in e -
gene a ional ansmission o income (Jän i & Jenkins, 2015; Mazumde , 2016), be e p edic s heal h and well-being
(B ady, Cu an, & Ca piano, 2023) and mo ali y (B ady, Kohle , & Zheng, 2023), and be e explains Black-Whi e
inequali ies and adul li e chances (B ady e al., 2020). Using he PSID, B ady e al. (2022) show ha his pa icula
measu e bes p edic s ma u e adul heal h, including ch onic condi ions. They also show i pe o ms a leas sligh ly
be e han se e al al e na i e p ospec i e o e ospec i e measu es o pa en s' occupa ion, social class, educa ion,
and absolu e income.
11
We include ou o he amily backg ound measu es a e aged o e all obse a ions du ing childhood. Ra he han
he PSID iden i ied “ e e ence pe son”, we de ine he household lead as he adul wi h he highes labou ma ke
ea nings in a gi en yea (B ady e al., 2018). Ties a e b oken by age (and andomly i age is ied). We use he lead o
calcula e he a e age pa en age in yea s and a e age pa en educa ion in yea s o schooling. We also adjus o child-
hood sibship size, which is he a e age numbe o o he child en in he household du ing childhood. Finally, we
include childhood single mo he hood as he p opo ion o yea s in a single mo he household du ing childhood. The
a e age esponden g ew up wi h a pa en wi h 12.8 yea s o schooling and an a e age age o 36.8 yea s, and a
sibship size o 1.5. On a e age, abou 15% o yea s du ing childhood we e in single mo he households.
Ma ching esponden s o hei s a e-yea , he models also adjus o se e al o he empo ally a ying s a e-le el
a iables a e aged o e childhood (VanHeu elen & B ady, 2022). Childhood s a e unemploymen ( a e) and childhood
g oss s a e p oduc pe capi a (GSP PC in la ion-adjus ed $) cap u e he business cycle and le el o economic de el-
opmen . Childhood s a e go e nmen spending (as % o GSP) and childhood s a e unionisa ion (% among nonag icul u al
wo ke s) cap u e s a e size and labou ma ke o ganisa ion as c i ical ins i u ions.
We also iden i y he modal s a e in which each child esided ac oss childhood. We hen code ha modal s a e
in o he 9 Census di isions, which we call nine smalle egions, and include ixed e ec s (FEs) o hose egions.
Un o una ely, he e a e oo ew esponden s pe s a e o include s a e FEs ins ead.
12
The ad an age o including
egion FEs is we can adjus o s able unobse ed di e ences be ween geog aphic a eas ac oss he U.S.
In he wa e when heal h ou comes a e measu ed, we adjus o he age o he esponden in yea s (mean 43.1).
The models con ol o bina y indica o s o 1970s and 1980s bi h coho s ( e e ence<1970 bi h). We also include
mu ually exclusi e bina y measu es o whe he esponden s a e Black (14%), La ino o O he Race
( e e ence =Whi e).
13
Finally, we con ol o a bina y indica o o Woman (50%).
3.4 |Analy ic app oach
All models a e linea p obabili y models. We clus e s anda d e o s based on he modal s a e o each esponden 's
childhood.
14
We i he ollowing eg ession models wi h he indi idual as he uni o analysis:
Yijk ¼β0þβMedicaidjþβXiþβZjþβWiþβRkþεijk
Se e e ch onic condi ions (Y) a e obse ed a age 30–56. Y a ies be ween indi iduals (i), who a e nes ed in
s a es (j) and nine egions (k). Yis p edic ed by s a e-le el childhood Medicaid (Medicaid
j
) and s a e-le el con ols (Z
j
),
bo h a e aged o e childhood. We include indi idual cha ac e is ics (X
i
) a e aged o e childhood o obse ed in
BRADY ET AL.45
adul hood. Because we use he oldes /mos ecen obse a ion o each esponden and hose obse a ions come
om a ious wa es, we include wa e FEs (W
i
) when he ou come was obse ed. Again, he models include FEs o
nine smalle egions (R
k
).
We exploi ha exposu e o childhood Medicaid a ied empo ally wi hin egions. Appendix GFigu e A1 con-
i ms and shows ha child en a a ious poin s 1979–1995 expe ienced subs an ially di e en Medicaid eligibili y.
While child en in 1979 expe ienced mo e uni o mly low le els, child en in 1987 and especially in 1995 expe ienced
a g ea e in e s a e he e ogenei y wi h a much highe mean and maximum. Thus, ou sample was exposed o e y
di e en Medicaid eligibili y depending on whe e and when hey we e child en.
The indi idual-le el con ols educe he con ounding wi h childhood and adul hood cha ac e is ics p edic ing
adul ch onic condi ions. By including he indi idual-le el con ols, we also adjus o be ween-s a e di e ences in
popula ion composi ion in he indi idual cha ac e is ics (e.g., ace and educa ion). By adjus ing o o he s a e-le el
a iables, we educe he unobse ed con ounding o childhood Medicaid wi h o he s a e-le el a iables. Including
he wa e and coho FEs means any e ec s o childhood Medicaid a e no simply due o na ion-wide imp o emen s
in popula ion heal h. Including nine smalle egion FEs, alongside he indi idual- and s a e-le el con ols, we assess
he e ec o Medicaid om wi hin- egion empo al a ia ion.
Ou iden i ying assump ion is ha o he unmeasu ed wi hin- egion changes in s a e-le el cha ac e is ics did no
co-occu wi h Medicaid policy changes. O cou se, his assump ion is p obably jus as s ong as he exclusion es ic-
ion o ins umen al a iables. We emphasise ha ou models simply p o ide di e en and complemen a y e i-
dence. We conjec u e ha he ield bene i s om e idence buil on a a ie y o modelling assump ions.
4|RESULTS
Table 1shows a ious models o se e e ch onic condi ions in adul hood on childhood Medicaid and o he inde-
penden a iables. Model 1 only includes childhood Medicaid, model 2 adds he indi idual-le el con ols, model
3 adds he egion FEs, and model 4 only includes he s a e-le el con ols and egion FEs. Model 5 includes all
a iables.
Ac oss models, childhood Medicaid is nega i e and s a is ically signi ican ly associa ed wi h se e e ch onic con-
di ions. Fo childhood Medicaid, we epo x-s anda dised coe icien s (i.e., co esponding o a one s anda d de ia-
ion change in he independen a iable). Fo a s anda d de ia ion inc ease in childhood Medicaid, he p obabili y o
a ch onic condi ion declines by 0.05–0.12. In model 5, he s anda dised coe icien is 0.05. Abou 32% has a se e e
ch onic condi ion. The e o e, a s anda d de ia ion highe childhood Medicaid is associa ed wi h abou a 16%–37.5%
educ ion om he mean p obabili y o ha ing a ch onic condi ion.
Using model 5, Figu e 2displays he declining p obabili y o a se e e ch onic condi ion ac oss he s anda dised
ange o childhood Medicaid. I he child expe ienced one s anda d de ia ion below a e age childhood Medicaid, he
p obabili y o a ch onic condi ion exceeds 0.4. Nea he middle o he dis ibu ion o childhood Medicaid, he p oba-
bili y o a ch onic condi ion is abou 0.3. A he high end, wo s anda d de ia ions abo e he a e age, he p obabili y
o a ch onic condi ion is only 0.2. Ac oss he ange o obse ed childhood Medicaid, he p obabili y o se e e ch onic
condi ions in adul hood is app oxima ely cu in hal .
While he childhood Medicaid coe icien is obus , no e he con as be ween models 1 and 2.
15
When
indi idual-le el con ols –including especially childhood income –a e added, he coe icien o childhood Medicaid
declines by abou 20%. Hence, omi ing he indi idual-le el a iables may upwa dly bias he childhood Medicaid
coe icien . In addi ion, childhood income has a obus ly signi ican nega i e associa ion wi h se e e ch onic condi-
ions be o e and a e adjus ing o egion FEs and s a e-le el con ols. In model 5, a s anda d de ia ion inc ease in
childhood income educes he p obabili y o a se e e ch onic condi ion by 0.04.
Beyond childhood Medicaid and childhood income, we no e pa en educa ion and being La ino a e obus ly
nega i ely associa ed wi h ha ing a se e e ch onic condi ion.
16
Among s a e-le el a iables, only GSP p.c. is
46 BRADY ET AL.
6
B own e al. (2020: 813) w i e: “Con olling o income a enua es he OLS es ima es, bu hey emain nega i e.”Thei
pe inen Online Appendix 12 only shows he esul s g aphically and only o axes paid (bu no o he ou comes). Thei
appendix ac ually shows a d ama ic decline in he size and signi icance o he Medicaid coe icien in bo h OLS and
educed o m models. They suspec (2020: pp. 798–799) –bu p o ide no e idence – ha his is because income a age
15 is pos ea men con ol o Medicaid eligibili y a younge ages.
7
Fo example, B own e al. (2020) in e ac wi h he o icial po e y measu e, which we explain is an in alid and un eliable
income measu e in en. 5 ( ecall also, hey only obse e childhood s a ing a ages 12–15 and impu e ages 0–11). Jackson
e al. (2021) in e ac wi h mo he 's educa ion. Cu ie & G ube , 1996a, Cu ie & G ube , 1996b in e ac wi h a c ude mea-
su e o non-equi alized p e- isc income in b acke ed ca ego ies (see en. 4). Boud eaux e al. (2016) in e ac wi h Medic-
aid eligibili y based on he o icial po e y measu e (see en. 5).
8
Ano he ad an age o ending he obse a ion o childhood in 1995 is ha he PSID becomes biannual soon a e in
1997. Hence, including yea s a e 1997 would make he obse a ion o childhood inconsis en .
9
This also educes sample a i ion as we only equi e one obse a ion a age 30+ ega dless o when. Fo ins ance, i
dea h occu s by 2019 (5% o sample), we can use he las obse a ion.
10
Al hough he PSID does no allow us o measu e and ins umen o ac ual eligibili y, he e is p eceden o using educed
o m models (Jackson e al., 2021). E e y s udy we ha e ead ha shows educed o m alongside ins umen al a iable
models inds nea ly iden ical es ima es (e.g. B own e al. 2020; Goodman-Bacon, 2018,2021; Mille & Whe y., 2019;
Thompson, 2017).
11
B ady e al. (2022) also es nonlinea income e ec s wi h polynomials and logs. They ind linea income i s as well o
be e han nonlinea income o all ma u e heal h ou comes.
12
Twel e s a es ha e ewe han en esponden s, and six s a es ha e ewe han i e esponden s. By con as , he nine
smalle egions con ain 154, 156, 455, 461, 468, 472, 527, 905, and 1072 esponden s. We ind la ge and mo e signi i-
can childhood Medicaid coe icien s when we use ou la ge egion FEs a he han he nine smalle egions. Fo p ece-
den , Mille and Whe y. (2019) also use egion FEs.
13
As is well-documen ed, long- e m PSID samples mainly include Black and Whi e indi iduals (B ady e al., 2020). We
ound consis en esul s when collapsing La inos in o O he Race.
14
We ind e en mo e s a is ically signi ican e ec s o childhood Medicaid eligibili y i we ollow mos PSID
esea ch (e.g., Duncan e al., 2012) and clus e s anda d e o s a he o iginal PSID household-le el (Abadie
e al., 2017).
15
I is unlikely ha he indi idual-le el con ols a e pos ea men con ol o childhood Medicaid. Race and sex a e
mos ly immu able o policy. Pa en s' educa ion and age a e es ablished mos ly be o e a child's exposu e o Medic-
aid. Childhood sibship size and single mo he hood a e a e ages ac oss childhood and heo e ically could change in
esponse o Medicaid. Howe e , hese wo a e ne e signi ican p edic o s so a e unlikely o be salien media o s.
Appendix Balso shows ha childhood Medicaid is weakly associa ed wi h o he a iables excep he dependen a -
iable, pa en s' educa ion, and age (which i should be co ela ed wi h as childhood Medicaid changed o e ime).
Also, childhood income is mo e likely a con ounde han a media o as ou income measu e does no mone ize
Medicaid.
16
O cou se, childhood income is likely pos ea men con ol o pa en 's educa ion and o he indi idual-le el con ols
(B ady e al., 2022). S ill, in model 4, he x-s anda dised coe icien o childhood income (0.04) is la ge han he
x-s anda dised coe icien o pa en educa ion (0.03), and he coe icien o being La ino (0.01).
17
Fo compa ison, we ind 32% o ou 30–56 yea olds ha e a se e e ch onic condi ion, while Thompson (2017) ind
5.8% o 18–20 yea olds ha e any ch onic condi ion, and Mille and Whe y. (2019) ind 24.6% o 19–36 yea olds
ha e any ch onic condi ion. I is di icul o compa e e ec magni udes ac oss s udies. Howe e , Thompson does
no ind obus ly signi ican e ec s on any ch onic condi ion. Medicaid eligibili y is no signi ican in OLS, educed
o m o ins umen al a iable models (Thompson, 2017;Table2). Thompson inds a signi ican nega i e e ec o
eligibili y a ages 0–5and6–11, bu no 12–18 (Thompson, 2017;Table 4). His obus ness checks show a signi ican
nega i e e ec in only wo o eigh models (Thompson, 2017;Table 5). Mille and Whe y ind signi ican nega i e
e ec s o p ena al and age 5–9 eligibili y o any ch onic condi ions. They ha e mo e han 10 imes as many cases
wi h he Na ional Heal h In e iew Su ey, which may accoun o why hey ind signi ican nega i e e ec s o
diabe es and high blood p essu e whe eas we only ind nea signi ican nega i e e ec s (Appendix C). Analysing
he ini ial ollou o Medicaid, Boud eaux e al. (2016) ind a signi ican nega i e e ec o hei ch onic condi ion
i
ndexamong helowincomebu no hemode a eincome(Boud eauxe al.,2016;Table2). Boud eaux e al. also
ind a signi ican nega i e in e ac ion o exposu e o and p edic ed pa icipa ion (Boud eaux e al., 2016;Table4).
BRADY ET AL.53
REFERENCES
Abadie, A., A hey, S., Imbens, G. W., & Woold idge, J. (2017). When should you adjus s anda d e o s o clus e ing? NBER
Wo king Pape No. 24003. Na ional Bu eau o Economic Resea ch.
Bake , R. S. (2022). The his o ical acial egime and acial inequali y in po e y in he Ame ican sou h. Ame ican Jou nal o
Sociology,127, 1721–1781.
Beck ield, J., & Bamb a, C. (2016). Sho e li es in s ingie s a es: Social policy sho comings help explain he U.S Mo ali y
Disad an age. Social Science & Medicine,171,30–38.
Boud eaux, M. H., Golde s ein, E., & McAlpine, D. D. (2016). The long- e m impac s o Medicaid exposu e in Ea ly child-
hood: E idence om he P og am's o igin. Jou nal o Heal h Economics,45, 161–175.
B ady, D., Cu an, M., & Ca piano, R. M. (2023). A es o he p edic i e alidi y o ela i e e sus absolu e income o sel -
epo ed heal h and well-being in he Uni ed S a es. Demog aphic Resea ch,48, 775–808.
B ady, D., Finnigan, R., Kohle , U., & Legewie, J. (2020). The inhe i ance o ace e isi ed: Childhood weal h and income, and
black-whi e disad an ages in adul li e chances. Sociological Science,7, 599–627.
B ady, D., Giesselman, M., Kohle , U., & Radenacke , A. (2018). How o measu e and p oxy pe manen income: E idence
om Ge many and he U.S. Jou nal o Economic Inequali y,16, 321–345.
B ady, D., Gue a, C., Kohle , U., & Link, B. (2022). The long a m o p ospec i e childhood income ank o ma u e adul
heal h in he U.S. Jou nal o Heal h and Social Beha io ,63, 543–559.
B ady, D., Kohle , U., & Zheng, H. (2023). No el es ima es o mo ali y associa ed wi h po e y in he US. JAMA In e nal
Medicine,17, 2023.
B ady, D., Ma qua d , S., Gaucha , G., & Reynolds, M. M. (2016). Pa h dependency and he poli ics o socialized heal hca e.
Jou nal o Heal h Poli ics, Policy and Law,41, 355–392.
B own, D. W., Kowalski, A. E., & Lu ie, I. Z. (2020). Long- e m impac s o childhood Medicaid expansions on ou comes in
adul hood. Re iew o Economic S udies,87, 792–821.
Callaway B, Goodman-Bacon A, San Anna PHC. 2021. Di e ence-in-di e ences wi h a con inuous ea men . Wo king pape :
h ps://a xi .o g/pd /2107.02637.pd .
Cohodes, S. R., G ossman, D. S., Kleine , S. A., & Lo enheim, M. F. (2016). The e ec o child heal h insu ance access on
schooling: E idence om public insu ance expansions. Jou nal o Human Resou ces,51, 727–759.
Cu ie, J., & G ube , J. (1996a). Heal h insu ance eligibili y, u iliza ion o medical ca e, and child heal h. Qua e ly Jou nal o
Economics,111, 431–466.
Cu ie, J., & G ube , J. (1996b). Sa ing babies: The e icacy and cos o ecen expansions o Medicaid eligibili y o p egnan
women. Jou nal o Poli ical Economy,104, 1263–1296.
Cu ie, J., Decke , S., & Lin, W. (2008). Has public heal h Insu ance o Olde Child en Reduced Dispa i ies in access o ca e
and heal h ou comes. Jou nal o Heal h Economics,27, 1567–1581.
Duncan, G. J., Daly, M. C., McDonough, P., & Williams, D. R. (2002). Op imal indica o s o socioeconomic s a us o Heal h
Resea ch. Ame ican Jou nal o Public Heal h,92, 1151–1157.
Duncan, G. J., Magnuson, K., Kalil, A., & Ziol-Gues , K. M. (2012). The impo ance o Ea ly childhood po e y. Social Indica-
o s Resea ch,108,87–98.
Eas , C. N., Mille , S., Page, M., & Whe y, L. R. (2023). Mul i-gene a ional impac s on childhood access o he sa e y ne :
Ea ly li e exposu e o Medicaid and he nex Gene a ion's heal h. Ame ican Economic Re iew,113,98–135.
F ick, J. R., Jenkins, S. P., Lilla d, D. R., Lipps, O., & Wooden, M. (2007). The c oss-na ional equi alen ile (CNEF) and i s
membe coun y household panel s udies. Jou nal o Con ex ual Economics –Schmolle s Jah buch,127, 627–654.
Goodman-Bacon, A. (2018). Public insu ance and mo ali y: E idence om Medicaid implemen a ion. Jou nal o Poli ical
Economy,126, 216–262.
Goodman-Bacon, A. (2021). The long- un e ec s o childhood insu ance co e age: Medicaid implemen a ion, adul heal h,
and labo ma ke ou comes. Ame ican Economic Re iew,111, 2550–2593.
G ogan, C., & Pa ashnik, E. (2003). Be ween wel a e medicine and mains eam en i lemen : Medicaid a he poli ical c oss-
oads. Jou nal o Heal h Poli ics, Policy and Law,28, 821–858.
G ogan, C. M., & And ews, C. M. (2015). Medicaid. In D. Beland, C. Howa d, & K. J. Mo gan (Eds.), The Ox o d handbook o
U.S. social policy (pp. 337–354). Ox o d Uni e si y P ess.
Haywa d, M. D., & Go man, B. K. (2004). The long a m o childhood: The in luence o Ea ly-li e social condi ions on Men's
mo ali y. Demog aphy,41,8
7–107.
Hein ich, C. J., Camacho, S., Hende son, S. C., He nandez, M., & Joshi, E. (2022). Consequences o adminis a i e bu den o
social sa e y ne s ha suppo he heal hy de elopmen o child en. Jou nal o Policy Analysis and Managemen ,41,11–44.
He d, P., & Moynihan, D. P. (2019). Adminis a i e Bu den. Russell Sage Founda ion.
Hube , E., & S ephens, J. D. (2001). De elopmen and c isis o he wel a e s a e. Uni e si y o Chicago P ess.
Jackson, M., Agbai, C., & Rausche , E. (2021). The e ec s o s a e-le el Medicaid co e age on amily weal h. RSF The Russell
Sage Founda ion Jou nal o he Social Sciences,7, 216–234.
54 BRADY ET AL.
Jakiela P. 2021. Simple diagnos ics o wo-way ixed e ec s. Wo king Pape : h ps://a xi .o g/abs/2103.13229.
Jän i, M., & Jenkins, S. P. (2015). Income mobili y. In Handbook o income dis ibu ion (Vol. 2, pp. 807–935). Else ie .
Johnson, R. C., & Schoeni, R. F. (2011). Ea ly-li e o igins o adul disease: Na ional Longi udinal Popula ion-based S udy o
he Uni ed S a es. Ame ican Jou nal o Public Heal h,101, 2317–2324.
Kaes ne , R., Ga e , B., Chen, J., & Gangopadhyaya, A. (2017). E ec s o ACA Medicaid expansions on heal h insu ance
co e age and labo supply. Jou nal o Policy Analysis & Managemen ,36, 608–642.
Kame man, S. B., & Kahn, A. J. (2001). Child and amily policy in he Uni ed S a es a he opening o he wen y- i s cen u y.
Social Policy & Adminis a ion,35,69–84.
Ka z, M. B. (2001). The P ice o ci izenship. Hol .
Ko pi, W. (1989). Powe , poli ics and s a e au onomy in he de elopmen o social ci izenship: Social igh s du ing sickness
in eigh een OECD coun ies since 1930. Ame ican Sociological Re iew,54, 309–328.
Kouse , T. (2002). The poli ics o disc e iona y Medicaid spending, 1980-1993. Jou nal o Heal h Poli ics, Policy and Law,27,
639–671.
Lan o d, D., & Quadagno, J. (2022). Iden i ying he undese ing poo : The e ec o acial, e hnic, and an i-immig an sen i-
men on s a e Medicaid eligibili y. The Sociological Qua e ly,63,1–20.
Le e e, M., O zol, S., Leininge , L., & Ea ly, N. (2019). Con empo aneous and long- e m e ec s o Child en's public heal h
insu ance expansions on supplemen al secu i y income pa icipa ion. Jou nal o Heal h Economics,64,80–92.
Le ine, P. B., & Schanzenbach, D. (2009). The impac o Child en's public heal h insu ance expansions on educa ional ou -
comes. F on ie s in Heal h Policy Resea ch,12,1–26.
Link, B. G., Phelan, J. C., Miech, R., & Wes in, E. L. (2008). The esou ces ha ma e . Jou nal o Heal h and Social Beha io ,
49,72–91.
Link, B. G., Susse , E. S., Fac o -Li ak, P., Ma ch, D., Kezios, K. L., Lo asi, G. S., Rundle, A. G., Suglia, S. F., Fade , K. M.,
And ews, H. F., Johnson, E., Ci illo, P. M., & Cohn, B. A. (2017). Dispa i ies in sel - a ed heal h ac oss gene a ions and
h ough he li e cou se. Social Science & Medicine,174,17–25.
Lip on, B. J., Whe y, L. R., Mille , S., Kenney, G. M., & Decke , S. (2016). P e ious Medicaid expansion may ha e had las ing
posi i e e ec s on O al heal h o non-Hispanic black child en. Heal h A ai s,35, 2249–2258.
Mazumde , B. (2016). Es ima ing he in e gene a ional elas ici y and ank Associa ion in he Uni ed S a es: O e coming he
cu en limi a ions o ax da a. In S. Polachek & K. Ta si amos (Eds.), Inequali y: Causes and consequences, esea ch in labo
economics (pp. 83–129). Eme ald G oup Publishing Limi ed.
Michene , J. (2018). F agmen ed democ acy. Camb idge Uni e si y P ess.
Mille , S., & Whe y, L. (2019). The long- e m e ec s o Ea ly li e Medicaid co e age. Jou nal o Human Resou ces,54,
785–824.
Nelson, K., F ed iksson, D., Ko pi, T., Ko pi, W., Palme, J., & Sjöbe g, O. (2020). The social policy indica o s (SPIN) da abase.
In e na ional Jou nal o Social Wel a e,29, 285–289.
O'B ien, R., & Robe son, C. L. (2018). Ea ly-li e Medicaid co e age and in e gene a ional economic mobili y. Jou nal o
Heal h and Social Beha io ,59, 300–315.
Quadagno, J. (1994). The colo o wel a e: How acism unde mined he wa on po e y. Ox o d Uni e si y P ess.
Quadagno, J. (2006). One na ion, uninsu ed: Why he U.S. In Has No Na ional Heal h Insu ance. Ox o d Uni e si y P ess.
Rainwa e , L., & Smeeding, T. M. (2003). Poo kids in a ich coun y New Yo k. Russell Sage Founda ion.
Reynolds, M. M., & A endano, M. (2018). Social policy expendi u es and li e expec ancy in high-income coun ies. Ame ican
Jou nal o P e en i e Medicine,54,72–29.
Sc uggs, L. A., & Ta oya, G. (2022). Fi y yea s o wel a e s a e gene osi y. Social Policy Adminis a ion, 56, 791–807.
Thompson, O. (2017). The long- e m heal h impac s o Medicaid and CHIP. Jou nal o Heal h Economics,51,26–40.
Tu ne , R. J., Thomas, C. S., & B own, T. H. (2016). Childhood ad e si y and adul heal h: E alua ing in e ening mechanisms.
Social Science & Medicine,156, 114–124.
VanHeu elen, T., & B ady, D. (2022). Labo Unions and Ame ican Po e y ILR Re iew. Sage Jou nal,75, 891–917.
Wend , C., F isina, L., & Ro hgang, H. (2009). Heal hca e sys em ypes: A concep ual amewo k o compa ison. Social Pol-
icy & Adminis a ion,43,70–90.
How o ci e his a icle: B ady, D., Gao, M., Gue a, C., Kohle , U., & Link, B. (2024). The long e m
ela ionship be ween childhood Medicaid expansions and se e e ch onic condi ions in adul hood. Social
Policy & Adminis a ion,58(1), 39–60. h ps://doi.o g/10.1111/spol.12942
BRADY ET AL.55
APPENDIX A: DESCRIPTIVE STATISTICS (N=4670)
Mean SD
(1) Se e e ch onic condi ion 0.32 0.47
(2) Childhood income (a e age %) 45.65 22.92
(3) Pa en s' educa ion (yea s) 12.81 2.47
(4) Pa en s' age (yea s) 36.80 6.69
(5) Childhood sibship size (#) 1.48 0.99
(6) Childhood single mo he hood (p opo ion) 0.15 0.26
(7) Age (yea s) 43.14 7.97
(8) Black 0.14 0.34
(9) La ino 0.00 0.03
(10) O he ace 0.04 0.19
(11) Woman 0.50 0.50
(12) 1970s coho 0.36 0.48
(13) 1980s coho 0.29 0.45
(14) Childhood medicaid eligibili y (yea s) 1.22 0.52
(15) Childhood S a e unemploymen (a e age a e) 6.88 1.37
(16) Childhood S a e GSP PC ( eal $) 42109.68 6924.11
(17) Childhood S a e Go . spending (a e age %) 10.00 1.60
(18) Childhood S a e unionisa ion (%) 20.48 8.47
56 BRADY ET AL.
APPENDIX B: CORRELATION MATRIX (N=4670)
Va iables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
(1) Se e e Ch onic
Condi ion
1.00
(2) Childhood Income 0.09 1.00
(3) Pa en s' Educa ion 0.16 0.60 1.00
(4) Pa en s' Age 0.01 0.46 0.16 1.00
(5) Childhood Sibship
Size
0.06 0.35 0.28 0.12 1.00
(6) Childhood Single
Mo he hood
0.03 0.51 0.24 0.38 0.08 1.00
(7) Age 0.26 0.01 0.24 0.11 0.23 0.03 1.00
(8) Black 0.08 0.51 0.36 0.27 0.26 0.45 0.01 1.00
(9) La ino 0.01 0.01 0.02 0.00 0.00 0.01 0.01 0.02 1.00
(10) O he Race 0.01 0.03 0.05 0.01 0.01 0.01 0.00 0.15 0.00 1.00
(11) Woman 0.01 0.05 0.06 0.03 0.04 0.10 0.06 0.05 0.02 0.00 1.00
(12) 1970s Coho 0.03 0.00 0.04 0.12 0.11 0.05 0.12 0.08 0.01 0.02 0.01 1.00
(13) 1980s Coho 0.18 0.03 0.17 0.06 0.11 0.07 0.74 0.08 0.02 0.01 0.01 0.61 1.00
(14) Childhood Medicaid
Eligibili y
0.24 0.05 0.28 0.11 0.20 0.03 0.79 0.01 0.03 0.01 0.04 0.14 0.71 1.00
(15) Childhood S a e
Unemploymen
0.11 0.03 0.12 0.05 0.08 0.05 0.38 0.04 0.01 0.03 0.03 0.21 0.42 0.40 1.00
(16) Childhood S a e GSP
PC
0.18 0.17 0.25 0.01 0.14 0.00 0.50 0.07 0.01 0.03 0.04 0.17 0.50 0.65 0.33 1.00
(17) Childhood S a e
Go . Spending
0.08 0.06 0.03 0.03 0.03 0.04 0.32 0.06 0.03 0.02 0.01 0.12 0.32 0.36 0.13 0.04 1.00
(18) Childhood S a e
Unionisa ion
0.02 0.22 0.07 0.05 0.03 0.08 0.35 0.29 0.01 0.01 0.00 0.05 0.32 0.16 0.46 0.12 0.00 1.00
No e: The co ela ions a e based on s anda dised alues o s a e-le el a iables.
BRADY ET AL.57
APPENDIX C: OLS REGRESSION MODELS OF SPECIFIC CHRONIC CONDITIONS ON CHILDHOOD MEDICAID ELIGIBILITY AND CONTROLS
Se e e ch onic
condi ions
High blood p essu e/
Hea disease Cance Diabe es
Lung
disease
Less se e e ch onic
condi ions A h i is As hma
O he ch onic
condi ion Obesi y
Childhood 0.050* 0.022 0.013 0.012 0.023** 0.055* 0.025+0.020 0.007 0.02
Medicaid
eligibili y
(0.02) (0.02) (0.01) (0.01) (0.01) (0.02) (0.01) (0.02) (0.02) (0.02)
R
2
0.109 0.110 0.030 0.066 0.039 0.061 0.077 0.025 0.026 0.068
N 4670 4670 4669 4670 4670 4602 4669 4669 4589 4587
% Posi i e o Ch onic Condi ions
Weigh ed % 31.86 23.91 4.43 8.57 4.01 33.42 13.09 13.21 16.16 33.49
Unweigh ed
%
29.96 23.02 3.47 7.11 4.09 29.79 10.62 13.11 12.99 36.87
No e: Models include all a iables and ixed e ec s om model 5 o Table 1. Robus clus e ed s anda d e o s in pa en heses. ***p<0.001, **p< 0.01, *p< 0.05.
58 BRADY ET AL.
APPENDIX D: ROBUSTNESS CHECKS: REGRESSION MODELS OF SEVERE CHRONIC CONDITIONS ON
CHILDHOOD MEDICAID ELIGIBILITY AND CONTROLS
APPENDIX E: ROBUSTNESS CHECKS INSPIRED BY RECENT ECONOMETRICS OF TWO-WAY FE MODELS
Only
e e ence
pe sons Woman Man
Ages
40+Logi
Childhood
incomein
1s Obs .
Childhood
income Ages
0–5
Childhood 0.051 0.058* 0.044 0.066* 0.312* 0.56* 0.060*
Medicaid
eligibili y
(0.03) (0.03) (0.03) (0.03) (0.14) (0.02) (0.02)
R2 0.126 0.119 0.125 0.095 0.111 0.092
N3306 2540 2130 2385 4668 4614 4111
No e: Models include all a iables and ixed e ec s om model 5 o Table 1. Robus clus e ed s anda d e o s in
pa en heses. ***p< 0.001, **p< 0.01, *p< 0.05.
Only
2019
wa e
Omi ing below
mean
esidualized
ea men cases
Omi ing
nega i e
esidualized
ea men
cases
Residualized
ou come on
esidualized
ea men *
childhood
Medicaid
eligibili y
D op
below
age 38
D op egions
wi h low
esidualized
ea men
Childhood
medicaid
eligibili y
0.074**
(0.02)
0.066*(0.03) 0.057*(0.02) 0.013**
(0.004)
0.055
(0.03)
0.064*
(0.03)
Residualized
ea men
0.068***
(0.014)
Childhood
medicaid
eligibili y*
esidualized
ea men
0.032***
(0.008)
R
2
0.111 0.079 0.111 0.149 0.094 0.100
N3612 1595 3723 4670 2819 2842
No e: Models include all a iables and ixed e ec s om model 5 o Table 1. Robus clus e ed s anda d e o s in
pa en heses.
*** p<0.001,** p< 0.01,* p< 0.05.
BRADY ET AL.59
APPENDIX F: PSID QUESTIONS AND DATA AVAILABILITY ON MEDICAID
In 1999, he PSID began asking: (a) whe he anyone in he household is co e ed by heal h insu ance; (b) wha kind
o heal h insu ance each membe has; (c) wi h mul iple op ions o each membe ; (d) wi h Medicaid as a po en ial
answe ; and o heads (e) whe he he head applied o Medicaid; and ( ) easons o being denied Medicaid. F om
1986 o 1997, esponden s we e asked i hey o any o he amily membe s ha e “Medicaid/Medi-Cal/Medical
Assis ance/Wel a e/Medical Se ices”. In 1979–1981 and 1983–1984, esponden s we e asked whe he anyone in
he amily ecei ed medical ca e paid o by Medica e o Medicaid. In 1977–1978, esponden s we e asked i anyone
in he amily ecei ed heal hca e paid o by Medicaid. F om 1969 o 1972, esponden s we e asked i hey could
ge “ ee medical ca e as a e e an, h ough Medicaid, o any o he way.
APPENDIX G
FIGURE G1 Dis ibu ion o Childhood Medicaid Eligibili y in Yea s in analy ical sample ac oss pe iods.
60 BRADY ET AL.
