Does Public Expenditure on Education Improve Well-Being? International Evidence

Kyklos, 2025; 0:1–17
h ps://doi.o g/10.1111/kykl.12466
1 o 17
Kyklos
ORIGINAL ARTICLE OPEN ACCESS
Does Public Expendi u e on Educa ion Imp o e Well- Being?
In e na ional E idence
RubénRubio-O iz | Da idPa iño | F anciscoGómez-Ga cía
Depa amen o de Economía e His o ia Económica, Facul ad de Ciencias Económicas y Emp esa iales, Uni e sidad o Se illa, Se illa, Andalusia,Spain
Co espondence: Da id Pa iño (pa [email protected])
Recei ed: 5 Sep embe 2024 | Re ised: 17 Feb ua y 2025 | Accep ed: 4 May 2025
Keywo ds: economics o educa ion| public expendi u e| subjec i e well- being
ABSTRACT
The aim o his a icle is o quan i y he ela ionship be ween public expendi u e on educa ion and indi idual subjec i e well-
being, p o iding empi ical e idence o he social e u n on his in es men . We use mic oda a om he Eu opean Social Su ey
(ESS) me ged wi h mac oeconomic a iables om o icial sou ces. Econome ic es ima ions a e ca ied ou using mul ile el
models. Ou esul s show a posi i e associa ion be ween public expendi u e on educa ion and indi idual well- being. Howe e ,
his e ec is no homogeneous ac oss educa ional le els, as obus e idence o a posi i e con ibu ion is only ound o e ia y
educa ion. Fu he mo e, we explo e whe he his ela ionship is con ingen on indi iduals' ideological p e e ences. Ou indings
indica e ha indi iduals who espouse a conse a i e ideology exhibi a weake e ec compa ed o hose wi h a p og essi e mind-
se . Ne e heless, he posi i e co ela ion be ween educa ion expendi u e and well- being pe sis s o he la e g oup. To assess
he obus ness o ou esul s, we ha e eplica ed he calcula ions using a di e en su ey, speci ically he Eu oba ome e , and
conduc ed es ima ions wi h al e na i e me hodologies, which con i m hei consis ency.
JEL Classi ica ion: H52, I26, I38, I31, I22
1 | In oduc ion
Viewing educa ion as an essen ial good o people's de elop-
men and well- being (Sen1999) as well as o economic g ow h
(Becke  1964; Schul z 1961; Solow 1957) makes in es men
in such a good o na ional in e es and one o he cen al as-
pec s o he policy o any coun y, as shown by i s inclusion in
he Sus ainable De elopmen Goals o he UN Agenda 2030.
Recei ing an adequa e s ock o educa ion is a undamen al as-
pec o any pe son's li e and i s p o ision is a cen al elemen
in any coun y's policy. Consequen ly, alloca ion o go e nmen
expendi u e owa ds educa ion is signi ican , accoun ing o
app oxima ely 10.1% o o al public expendi u e in Eu opean
Union coun ies o 2019 (Eu os a 2023). Gi en he signi icance
o his expendi u e, in es iga ing he e u ns on his in es men
is wo hwhile.
Re u n on his in es men , mainly om a inancial pe spec-
i e, has been ex ensi ely s udied om he classical app oach o
human capi al heo ies—pa icula ly h ough he educa ional
e u n a e (Ang is and K uege 1999; Ca d1999; Mince 1974;
Psacha opoulos and Pa inos2018). Howe e , calcula ing edu-
ca ional e u n a es p o es complex and has by no means been
exemp om me hodological conce ns. This complexi y a ises,
among o he easons, om he di icul ies in ol ed in quan i-
ying ce ain nonmone a y bene i s o educa ion in mone a y
e ms. As a esul , hese bene i s a e omi ed due o he inhe -
en di icul y o measu ing hem in mone a y e ms, among
o he limi a ions associa ed wi h adi ional mone a y mea-
su es (S igli z e al.2009). F om ano he pe spec i e, he p es-
ence o posi i e ex e nali ies a all le els also makes i di icul
o quan i y he ull bene i s o educa ion. Thus, he ela ionship
be ween educa ion and subjec i e well- being can be analyzed
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.
© 2025 The Au ho (s). Kyklos published by John Wiley & Sons L d.
2 o 17 Kyklos, 2025
unde wo app oaches (Lei e e al.2024). A he mic oeconomic
le el, his pape inds a posi i e bu weak ela ionship. Howe e ,
a he mac oeconomic le el, he ela ionship is also posi i e, bu
e y s ong. In his con ex , ou wo k, in he exe cise o e al-
ua ing he e ec o public educa ional spending on subjec i e
well- being, allows us o cap u e he posi i e ex e nali y ha i
gene a es and ha can explain his ela ionship, which is no
possible o cap u e in s udies ha a e limi ed o he indi id-
ual le el.
The subjec i e well- being app oach o e s a good al e na i e o
add ess hese and o he s limi a ions. This app oach p o ides a
amewo k o es ima e indi idual's o e all u ili y ac oss a ious
dimensions om a holis ic pe spec i e, aking in o accoun ac-
o s such as income and weal h, inequali y in i s dis ibu ion,
en i onmen al conce ns, leisu e, and he sociocul u al con ex ,
among o he s. This allows us o iden i y ac ions, con ex s, and
policies ha help o enhance people's well- being and makes
i easible o es ima e a e u n a e o educa ion ha includes
all i s e ec s. This is achie ed by moni o ing indi iduals' sel -
epo ed li e sa is ac ion o u ili y. By collec ing such epo s
om a di e se ange o indi iduals—a di e en poin s in ime
and ac oss a ious adminis a i e uni s—i is possible o in e
ela ionships be ween con ex ual ci cums ances and he a y-
ing deg ees o u ili y expe ienced by indi iduals (F ey and
S u ze 2002; Ode ma and S u ze 2017).
In o de o cla i y he concep ual amewo k unde pinning his
esea ch, i is impo an o conside he dis inc ion be ween he
h ee dimensions o he key independen a iable o his pape ,
subjec i e well- being: e alua i e, hedonic, and eudaimonic
(G aham and Nikolo a2015). Fi s , e alua i e well- being (li e
sa is ac ion) e e s o a esponden 's assessmen o he deg ee
o which hey a e sa is ied wi h hei li e as a whole. Secondly,
hedonic well- being e e s o he emo ional and a ec i e compo-
nen o subjec i e well- being. I e lec s how people expe ience
hei li es as opposed o how hey alue i gene ally (Kahneman
and K uege 2006). Finally, eudaimonic well- being e lec s he
A is o elian no ion o happiness as a li e pu pose and ocuses on
he lou ishing and ul illmen o human po en ial (O ganisa ion
o Economic Co- ope a ion and De elopmen  2013). On he
o he hand, Becche i e al.(2023) ind a ela ionship be ween
eudaimonic well- being, which hey associa e wi h he sense
o li e, and subjec i e su i al p obabili y, which hey use as a
p oxy o sel - assessed li e expec ancy. In his way, hey link
his subjec i e measu e o well- being wi h su i al p obabili y,
which becomes pa icula ly ele an in indi iduals' decisions e-
ga ding aspec s such as consump ion, sa ings, and e i emen .
Gi en ha ull li e sa is ac ion is one o he objec i es o be
achie ed o indi iduals and, by ex ension, go e nmen s—gi en
hei commi men o he well- being o hei ci izens (Oishi and
Diene 2014; Pe ei a e al.2023; Schube 2012)—explo ing in
dep h he policies and ac ions ha help achie e his objec i e
p o es c ucial. In addi ion, being public inancial esou ces lim-
i ed, i is impo an o alloca e spending app op ia ely among
di e en policies o spending p og ams and se ing p io i ies
in de ining he policies o be implemen ed. This pape seeks o
accomplish such a goal. We e alua e he e u ns o public edu-
ca ion; howe e , in con as o he p e ious li e a u e, we assess
i as h ough he subjec i e well- being app oach. The esul s ob-
ained o e in e es in hemsel es as hey allow o a di e en ,
and also en iching, way o assessing public policies, and he
indings complemen he adi ional assessmen de i ed om
classical human capi al heo ies.
The main objec i e o his pape is o assess he e ec o pub-
lic spending on educa ion. To do so, we p opose o quan i y i
h ough he well- being gene a ed by he expending. Using e-
po ed li e sa is ac ion collec ed h ough su eys ha we em-
ploy as a p oxy o well- being. Howe e , ou a p io i app oach is
ha public spending on educa ion imp o es people's li es and,
he e o e, has a posi i e e u n in e ms o well- being. Thus, ou
pape aims o empi ically es he posi i e e ec o public ed-
uca ion on indi iduals' sel - epo ed well- being and, he e o e,
gene a es imp o emen s in he well- being o socie ies. Howe e ,
we also conside a p io i ha , al hough posi i e, he e u n o
public spending on educa ion on subjec i e well- being is no ho-
mogeneous ac oss educa ional le els (p ima y, seconda y, and
e ia y), in line wi h i s con ibu ion o inancial pe o mance.
Ou empi ical wo k also aims o quan i y his e ec o he di -
e en educa ional le els.
Finally, we also conside he ex en o which people's poli ical
p e e ences modi y he e ec o public spending on he ex en o
which public educa ion p oduces he e ec we hypo hesize on
subjec i e well- being, as i is possible ha his e ec is media ed
by his aspec . To he bes o ou knowledge, ou s udy con ib-
u es o he li e a u e by es ing in a no el way whe he he e ec s
o educa ion spending a e simila ac oss educa ional le els and
whe he poli ical p e e ences modi y he e ec s on indi idual
wel a e. Ou s udy also aises conce ns abou he obus ness o
ou esul s. We alida e ou indings using wo di e en su eys
ha use di e en esponse scales o he endogenous a iable.
In addi ion o he main objec i e, he pape also has seconda y
objec i es ha ep esen ou s anding con ibu ions o he p e-
ious li e a u e. In pa icula , he pape conduc s a li e a u e
e iew on he ela ionship be ween subjec i e well- being and
educa ion, dis inguishing be ween he indi idual and social ap-
p oaches. Ou pape ocuses p ima ily on he social app oach.
Fo his app oach, he e is a small li e a u e e iew. While he
majo i y o p io s udies ha e p edominan ly adop ed an indi-
idual app oach o examining he ela ionship be ween educa-
ion and subjec i e well- being (among o he s, A aki2022; Cla k
and Oswald1996; F ey and S u ze 2000; K is o e sen2018),
ou pape adop s a social app oach by combining indi idual
a iables om su ey mic oda a wi h agg ega e a iables. We
ocus on public expendi u e in educa ion, as an in es men , and
use i as he main s udy a iable o es i s ela ionship wi h well-
being. In his way, we emphasize he mac o aspec o he policy
bu e alua e i in e ms o he e ec s i has on indi iduals.
The esul s con i m he ini ial hypo heses. Public expendi u e
on educa ion has a posi i e a e o e u n in e ms o subjec i e
well- being, and his e u n is no uni o m ac oss le els o edu-
ca ion. Speci ically, he e is s ong e idence o a posi i e e u n
only o public expendi u e on e ia y educa ion. Fu he mo e,
we also ind ha some indi idual cha ac e is ics play a modi y-
ing ole. I unde lines he ac ha indi iduals who display con-
se a i e poli ical p e e ences exhibi a smalle e ec o public
in es men in educa ion compa ed o o he subse s o he popu-
la ion wi h mo e p og essi e mindse .
14676435, 0, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/kykl.12466 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [23/05/2025]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
3 o 17
The pape is s uc u ed as ollows. In he nex chap e , we p o-
ide a b ie li e a u e e iew, ollowed by he p esen a ion o he
heo e ical amewo k and he da a used in he empi ical s udy.
The esul s a e hen p esen ed and discussed. Finally, we d aw
conclusions and sugges p omising a enues o u u e esea ch.
In addi ion, we include he bibliog aphy used and appendices o
supplemen a y in o ma ion.
2 | Li e a u e Re iew
The link be ween subjec i e well- being and educa ion has
mainly been s udied om an indi idual app oach. Nume ous
a icles ha e explo ed wha e ec indi iduals' le el o educa-
ion—some imes measu ed as yea s o schooling and on o he
occasions in e ms o he maximum le el achie ed—has on
subjec i e well- being (see Bücke e al.2018; Tan e al.2020;
Wi e e al.1984). Many o hese s udies con i m he posi i e
ela ionship ha he known educa ion bene i s an icipa e (see,
among o he s, Blanch lowe and Oswald2004; Eas e lin2001;
Fe e - i- Ca bonell2005; F ey and S u ze 2000). Howe e , as ye
no consensus has been eached, because se e al a icles epo a
nega i e o nonsigni ican ela ionship.
One common explana ion o his nega i e ela ionship in-
ol es he un ul illed expec a ions o indi iduals wi h highe
educa ion (Cla k and Oswald 1996; K is o e sen 2018).
O he possible causes ha e also been iden i ied, such as
s ess and ime cons ain s esul ing om highly skilled jobs
(Nikolae 2018), he need o spend e e - inc easing amoun s
o ime and money on o mal educa ion in a mo e compe i-
i e socie y (Veenho en and Be g2013), he oppo uni y cos
o educa ion and i s pe cep ibili y (Fe an e2009), o a en-
dency among unhappy indi iduals o pu sue highe le els
o educa ion (Veenho en2010). Howe e , o he s udies seek
o explain hese esul s wi hou calling in o ques ion he
posi i e con ibu ion made by educa ion. Ra he han only
es ima ing he di ec e ec —which hey assume o be null—
hey a emp o es ima e he indi ec e ec , which becomes
blu ed among o he a iables. O he s udies in ol e he use
o es ima ion echniques ha allow a nonlinea ela ionship
o be s udied (Cuñado and Pé ez de G acia2012; Ha og and
Oos e beek1998; Helliwell2003; Powd ha ee e al.2013).
The e is abundan amoun o li e a u e add essing he
indi idual- le el app oach, especially due o he inclusion o ed-
uca ion le el as a con ol a iable in mos subjec i e well- being
models. In con as , e y ew s udies ha e explo ed he ela-
ionship be ween agg ega e educa ion a iables and indi idual
subjec i e well- being using a monog aphic app oach. Despi e
he use ulness o such an app oach o policy design, ew s ud-
ies ha e linked people's well- being wi h he a iables ha de-
ine educa ion policies and mainly wi h spending on educa ion.
O e all, he esul s ound in he li e a u e allow us o build ou
s a ing hypo hesis ha public educa ion can be conside ed a
powe ul ins umen o aise he wel a e le els o indi iduals.
Among he ew exis ing s udies, Bukenya e al.(2003) include
public expendi u e on educa ion in 1988 o ce ain egions
o Wes Vi ginia (Uni ed S a es) as a a iable o explaining
li e sa is ac ion. The au ho s ind i s impac o be posi i e and
signi ican , ye also ind local public expendi u e on educa ion o
be nonsigni ican . Simila ly, explo ing a b oade dimension be-
yond public spending on educa ion, Acuña- Dua e e al.(2024)
ind ha p e e ences o e o ming he Chilean Cons i u ion
can be d i en by g ea e discon en wi h he p o ision o pub-
lic goods.
Hessami (2010) and Ho and Ng (2016) o e mo e closely e-
la ed e e ences because hey link subjec i e well- being wi h
size o go e nmen . The o me s udy employs ce ain econo-
me ic models o 12 Eu opean Union coun ies, using public
expendi u e on educa ion as an exogenous a iable in many o
hem. The au ho 's indings sugges an in e se U- shaped ela-
ionship, indica ing an op imal le el o public expendi u e ha
maximizes subjec i e well- being. Building upon his no ion, Ho
and Ng(2016) explo e he op imal le el o public expendi u e
in e ms o maximizing subjec i e well- being using panel da a
om 78 coun ies. The au ho s iden i y op imal go e nmen size
as anging be ween 35.3% and 37.8% o g oss domes ic p oduc
(GDP), wi h he op imal le el o public expendi u e on educa ion
being be ween 3.96% and 4.25% o GDP.
Cheung and Chan(2011) conduc an econome ic analysis using
agg ega ed and a e aged da a a coun y le el o es ima e wha
impac public expendi u e on educa ion—measu ed as a pe -
cen age o GDP—has on happiness. Using wo- ime e e ences
o public expendi u e on educa ion o explain a e age happiness
be ween 2000 and 2008, hey ound ha public expendi u e on
educa ion in 2004 had a posi i e e ec on a e age li e sa is ac-
ion. Howe e , hey ound no s a is ically signi ican ela ion-
ship o 2001. Also using a e age coun y- le el da a, Kim and
Kim(2012) ind a posi i e e ec o spending in educa ion on
subjec i e well- being.
Using Wo ld Values Su ey (WVS) mic oda a o a panel o 59
coun ies be ween 1981 and 2013, Jun(2015) applies mul ile el
eg essions wi h obus s anda d e o s wi h andom in e -
cep and coe icien s ha include indi idual- le el and agg e-
ga ed a iables. The exogenous a iables include a composi e
a iable—gene a ed by p incipal componen analysis—which
includes public expendi u e on educa ion, public social expendi-
u e, he Gini index, and o al seconda y educa ion en ollmen .
The au ho inds an ambiguous ela ionship be ween his com-
posi e a iable and subjec i e well- being and concludes ha no
clea e ec could be de e mined.
Using indi idual- le el da a o 109 coun ies om 1998 o 2008,
O ega Gil(2021) applies ixed- e ec s models o es ima e a pos-
i i e ela ionship be ween public expendi u e on educa ion and
li e sa is ac ion. Qasim and G imes(2022) examine changes in
li e sa is ac ion be ween wo ounds. They use a dynamic OLS
model, whe e hey in oduce mac oeconomic a iables e e -
enced o he i s ound o explain changes in li e sa is ac ion in
he pe iod be ween he wo ounds. Using public expendi u e on
educa ion as one o he exogenous a iables, hey ind no signi -
ican e ec on li e sa is ac ion.
Finally, O uza e al.(2021) employ single- coun y da a wi h
a egional app oach o quan i y he e ec o public expendi u e
in heal h on subjec i e well- being. Thei model includes public
expendi u e on educa ion as a con ol a iable and e idences a
14676435, 0, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/kykl.12466 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [23/05/2025]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
4 o 17 Kyklos, 2025
posi i e and signi ican coe icien . The s udy is no able o i s
inclusion o indi iduals' poli ical p e e ences and because i ex-
amines hei in e ac ion wi h he heal h expendi u e a iable.
The indings highligh ha he impac o his a iable on li e
sa is ac ion is con ingen upon poli ical ideology.
3 | Theo e ical F amewo k and Econome ic
S a egy
This pape is g ounded on he p emise ha li e sa is ac ion e-
ealed by indi iduals h ough su eys se es as an accu a e p oxy
a iable o indi idual u ili y le els. The jus i ica ion o employ-
ing his me hod es s on he assump ion ha indi iduals possess
he mos accu a e knowledge o hei own u ili y, he eby making
di ec inqui y he op imal s a egy (F ey and S u ze 2002).
This app oach signals a e u n o classical u ili a ian ideas o
u ili y measu abili y pu o wa d by au ho s such as Ben ham
o Edgewo h (Ansa- Eceiza and Gómez- Ga cía 2019). This
heo e ical co pus was elega ed by he e ealed p e e ences
app oach, which has he ad an age o only equi ing p e -
e ences o be o de ed a he han equi ing a measu e o o
wha ex en one choice is p e e ed o e ano he . Howe e ,
he e ealed p e e ences app oach makes un ealis ic assump-
ions such as he ex eme a ionali y o indi iduals, which has
been subjec o conside able deba e (Kahneman e al.1997;
Kahneman and Sugden2005). Gi en hese limi a ions, beha -
io al economis s ha e de eloped a obus heo e ical ame-
wo k ha p o ides a solid ounda ion o empi ical s udies on
subjec i e well- being.
The model employed in his s udy is based on he li e sa is ac ion
app oach (F ey e al.2010), which enables he ne e ec o di e -
en go e nmen policies on indi idual well- being o be es ima ed.
This app oach has he ad an age o a oiding some adi ional
limi a ions such as s a egic esponse beha io because—by ask-
ing abou li e sa is ac ion in a wide sense—i p e en s esponses
di ec ly linked o policy objec i es (Kahneman and Sugden2005;
Ode ma and S u ze 2017). Mo eo e , he comp ehensi e and in-
eg a ed pe spec i e o he app oach allows policy consequences
ha a e o en o e looked by o he me hodologies o be explo ed.
Speci ically, in addi ion o he di ec e ec s, he es ima ed e ec s
encompass ex e nali ies, unobse ed e ec s, o indi ec channels
o ansmission. This comp ehensi e assessmen is possible be-
cause indi iduals alue hei li es as a whole, conside ing u u e
p ojec ions, pas expe iences, and ela i e ci cums ances in com-
pa ison o o he s (F ey and S u ze 2002; And iani and Ashy o
2022). F om a b oad pe spec i e, indi iduals' le el o well- being
o eal u ili y (
Uij
) depends on ce ain indi idual cha ac e is ics
(
Xij
) as well as agg ega e, social, o con ex ual a iables (
Yj
), one
o which is ou main exogenous a iable—public expendi u e on
educa ion (
EDUj
):
whe e he subsc ip s ep esen he indi idual (i), he coun-
y whe e hey li e (j), and he ime ( ). This u ili y unc ion
can be app oxima ed wi h he li e sa is ac ion a iable, as a
p oxy, including an e o e m which—in addi ion o andom
pe u ba ion— e lec s indi iduals' inhe en inabili y o ac-
cu a ely epo hei ue li e sa is ac ion le el, among o he
ci cums ances (Blanch lowe and Oswald 2004; O uza
e al.2021):
Simila o nume ous p io s udies (such as Di Tella e al.2003;
Pa iño e al.2022), we employ indi idual- le el da a om su -
ey mic oda a, oge he wi h agg ega ed coun y- le el da a.
Howe e , unlike p e ious s udies ha g oup indi idual da a
using a e ages, which allows hem o make es ima ions using
simple econome ic echniques, we exploi he ichness o
wo king wi h indi idual mic oda a o ope a e wi h wo le -
els. As poin ed ou by Hox e al.(2010), elying solely on indi-
idual da a a e ages may lead o a loss o accu acy and yield
di e en esul s compa ed o es ima es based on mic oda a.
By using a nes ed da a s uc u e, we gain lexibili y, be e
con ol o e he e ogenei y h ough indi idual a iables, and
he abili y o employ mo e echniques such as a iable in e -
ac ions. Fu he mo e, his app oach acili a es a be e unde -
s anding o he ela ionships be ween a iables a di e en
le els. None heless, he nes ed da a s uc u e does pose one
challenge because i iola es he assump ion o independence,
he eby possibly esul ing in biased es ima o s and excessi e
Type I e o , ha is, i has a high endency o es ima e alse
posi i es. To add ess his issue, we employ mul ile el models
(Gelman e al.2012).
We es ima e mul ile el models based on his heo e ical ame-
wo k. This ype o model, which is also known as hie a chical,
mixed, o nes ed, is a combina ion o ixed- e ec s and andom-
e ec s models and is e y use ul in con ex s whe e obse a ions
a e nes ed in clus e s. This me hod allows in e ac ions ac oss
le els o be explo ed, he eby enabling analysis o how mac o-
economic policies and con ex s in luence indi idual a i udes,
hough s, and beha io s (Fai b o he 2014).
Ou choice o his s a egy di e ges om common al e na i es
in he li e a u e ha ha e p e iously explo ed his esea ch
ques ion. One equen ly employed echnique is OLS eg es-
sion wi h ixed e ec s, whe e dummy a iables a e employed
o con ol o spa ial o empo al he e ogenei y. Howe e , his
app oach uns he isk o dummies abso bing a subs an ial
amoun o le el a iabili y, possibly leading o mul icollinea -
i y issues wi h exogenous a iables a he coun y–yea le el
(Qasim and G imes2022). Mul ile el es ima ion a oids his
si ua ion.
Choice o g oup is c ucial in mul ile el models. In ou case,
da a a e clea ly g ouped by coun ies. Ne e heless, because
we d aw on di e en su ey ounds, hese ounds may possess
ime- ela ed aspec s ha a e sha ed ac oss yea ly obse a-
ions. This g ouping highligh s he ac ha om ins i u ional-
s uc u al aspec s, de e mined by he coun y in which he
indi idual esponden esides, o economic aspec s, es ablished
by he yea o which he esponse co esponds, all obse a ions
ha co espond o he same coun y–yea clus e ha e common
ela ionships ha co ela e hem. Some s udies ha e empha-
sized ha he minimum numbe o clus e s o es ima e e o s
(1)
Uij (
X
ij
,Y
j
,EDU
j ),
(2)
Li esa ij
=U
ij (
X
ij
,Y
j
,EDU
j )
+𝜀
i.
14676435, 0, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/kykl.12466 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [23/05/2025]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
5 o 17
accu a ely mus be a leas be ween 30 and 50 and ha he
numbe o clus e s is mo e impo an han he numbe o obse -
a ions in each clus e (Maas and Hox2005; Paccagnella2011).
In iew o his, we conside obse a ions clus e ed by coun-
y–yea g oups. This clus e s uc u e inc eases he numbe o
highe le el uni s while allowing us o cap u e unobse ed he -
e ogenei y ela ed o spa ial and empo al dimensions wi hou
he di icul ies caused by a less p ecise me hod such as simply
including ixed e ec s. Thus, ou mul ile el model is s uc u ed
in wo. Ou second le el (o highe le el) comp ises a iables
a he coun y–yea g oup le el, while he i s le el (o lowe
le el) consis s o indi idual obse a ions, which, s a is ically
eed om he s uc u al and conjunc u e e ec , can ade-
qua ely de e mine he e ec o public educa ion, he a iable
we in end o measu e.
As ega ds he mixed models employed, we ea he in e cep as
andom while conside ing he coe icien s as ixed. The g oup
in e cep se es as he s a ing poin o indi idual eg essions
wi hin each coun y–yea g oup. This modeling app oach en-
hances in e p e abili y, as he coe icien s ob ained indica e he
e ec o exogenous a iables on he o ma ion o he g oup in e -
cep . Speci ically, a posi i e coe icien signi ies ha indi iduals
belonging o he same coun y–yea g oup ha e a g ea e like-
lihood o a aining highe le els o li e sa is ac ion because o
a ia ions in he le el o he a iable s udied.
The gene al equa ion o he wo- le el mixed model es ima ed is
as ollows:
Le el 1: indi iduals in he su ey
Le el 2: coun y–yea g oups
Combined:
whe e ij ep esen s an indi idual i wi hin coun y–yea g oup j,
LSij
deno es hei li e sa is ac ion le el,
Xij
ep esen s he ec o
o indi idual cha ac e is ics ha a y wi hin he coun y–yea
g oup, and
Mj
is a ec o o mac oeconomic a iables ha e-
main cons an wi hin each g oup bu a y ac oss g oups. The
g oup andom e o e m (
uj
) and indi idual e o e m (
𝜀ij
) com-
ple e he gene al equa ion.
Gi en he o dinal na u e o he endogenous a iable—li e sa -
is ac ion—i is common p ac ice o es ima e disc e e choice
models such as he o dinal logi model. Howe e , p e ious li -
e a u e sugges s ha linea models may yield simila esul s
o disc e e choice models when he disc e e li e sa is ac ion
a iable employs a high- esponse scale (Fe e - i- Ca bonell and
F ij e s2004; K is o e sen2018). The ca dinali y assump ion
is gene ally accep ed in such cases (K is o e sen2017). In ou
main s udy, gi en ha i p o ides a high scale o esponses o he
li e sa is ac ion a iable, we employ linea models ollowing his
logic, as hey acili a e he in e p e a ion o coe icien s, which is
al eady complex in he case o mul ile el models. Addi ionally,
his app oach helps educe he compu a ional complexi y e-
qui ed by he algo i hm, which is pa icula ly demanding o o -
dinal models. We es ima e o dinal models as a obus ness check.
D awing on he Eu opean Social Su ey (ESS) da ase —which
p o ides a high- esponse scale o he li e sa is ac ion a iable—
we employ bo h linea mixed models and o dinal logi mixed
models as a obus ness check. Fo he Eu oba ome e da ase —
which con ains a li e sa is ac ion a iable wi h only ou ca e-
go ical le els—we only employ he o dinal mixed logi model.
We es ima e models using R, Ve sion 4.2.2. Speci ically, we use
he lme4 package o linea mixed eg essions (Ba es e al.2015)
and he o dinal package o o dinal logi mixed eg ession
(Ch is ensen2022).
4 | Da a
We use he mic oda a om he ESS (ESS ERIC2020), which
con ains all he indi idual da a we equi e. Desc ip i e s a is-
ics o indi idual con ol a iables a e con ained in Table 1.
Simila ly, he desc ip i e s a is ics o mac oeconomic a iables,
(3)
LSij
=
𝛾i
+
𝛽ij
∗
Xij
+
𝜀ij.
(4)
𝛾i
=
𝛾0
+
𝛽j
∗
Mj
+
uj.
(5)
LSij
=
𝛾
0+
𝛽ij
∗
Xij
+
𝛽j
∗
Mj
+
uj
+
𝜀ij,
TABLE 1 | Desc ip i e s a is ics o he Eu opean Social Su ey a iables.
Va iable N. Obs. Mean S d. De . Min. Max.
Li e sa is ac ion 336,521 7.02 2.2 010
Age 336,521 48.43 18.26 14 114
Poli ical p e e ences 336,521 1.97 0.75 1 3
Ma i al s a us 336,521 2.56 1.78 1 5
Unemployed 336,521 0.05 0.22 0 1
Educa ion le el 336,521 1.96 0.65 1 3
Indi idual heal h (au o e alua ion) 336,521 2.19 0.91 1 5
Sex 336,521 1.52 0.5 1 2
Immig an 336,521 1.09 0.29 1 2
Sou ce: Own elabo a ion based on da a used (ESS ERIC2020).
14676435, 0, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/kykl.12466 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [23/05/2025]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License

6 o 17 Kyklos, 2025
indica ing a iabili y, a e p esen ed in Table 2. The speci ic
ounds, yea s, and coun ies employed in he su ey a e de ailed
in TableA1. Mac oeconomic a iables a e me ged wi h indi id-
ual obse a ions using coun y and yea as e e ence, which
some imes, o a numbe o easons de ailed below, di e s om
su ey ound edi ion yea .
We u ilize da a om he nine ounds ha ha e been con-
duc ed, spanning he pe iod 2002–2020. The pe iod co e ed
is su icien ly long o cap u e a ia ions in public educa ion,
which cons i u es a s uc u al policy in Eu opean coun ies
and, he e o e, exhibi s e y limi ed a iabili y. Addi ionally,
his pe iod is o pa icula in e es as i includes nume ous
modi ica ions and signi ican his o ical changes ha make
i a aluable case s udy. Speci ically, i encompasses impo -
an s uc u al ans o ma ions, pa icula ly in some coun ies
in Eas e n Eu ope, which ha e unde gone p o ound changes
in hei si ua ion. I also includes a ious economic conjunc-
u es ha ha e signi ican ly a ec ed he p o ision o public
se ices, and educa ion in pa icula , p o iding a signi ican
deg ee o a iabili y, in addi ion o ha gene a ed by he di -
e en na ional models employed. In pa icula , i co e s he
s ong economic g ow h o he ea ly 21s cen u y, he G ea
Recession o 2008–2012, and he subsequen eco e y. The
ecession ep esen ed an in e es ing asymme ic shock ha
pa icula ly a ec ed Sou he n Eu opean coun ies.
We ha e ocused ou wo k on he ESS, disca ding o he sou ces
such as he Eu oba ome e because i is impo an o no e ha
he measu emen o li e sa is ac ion—which cons i u es he en-
dogenous a iable—di e s be ween he wo su eys, making
i di icul o compa e he esul s di ec ly. Speci ically, he ESS
measu es li e sa is ac ion h ough a ca ego ical a iable anging
om 0 (ex emely unsa is ied) o 10 (ex emely sa is ied), while
he Eu oba ome e limi s he possible alues wi h ou le els
( e y unsa is ied, unsa is ied, sa is ied, and e y sa is ied). In
his ega d, al hough he la e da ase also con ained in o ma-
ion on he a iables we equi ed, we decided o exclude i due o
he challenges in compa ing bo h esul s. We limi ed he use o
his sou ce o conduc ing a obus ness check on he quali a i e
esul s.
The es ima ed models inco po a e indi idual con ol a iables
commonly used in p e ious s udies, such as age, educa ional
le el, gende , household size, occupa ion, and poli ical p e e -
ences (Radcli 2013). Con inuous a iables ha e been no mal-
ized and cen e ed using he mean— ollowing he app oach o
Bell e al.(2018). This no maliza ion acili a es he calcula ion
in he likelihood unc ion op imiza ion p ocess.
Simila ly, he mac oeconomic a iables employed align wi h
hose used in p e ious s udies wi hin his ield (see, among o h-
e s, Hessami2010; O ega Gil2021; O uza e al.2021). F om
i , we in end o con ol o mac oeconomic aspec s and ins i u-
ional cha ac e is ics, which ha e o be cap u ed in a speci ic
way. The mac oeconomic a iables included in he models do
no ha e a high co ela ion be ween hem, and he e is no in-
di idual co ela ion g ea e han 0.7. In addi ion, he VIF es
is well below 5, such ha mul icollinea i y is unlikely. Finally,
o p o e ha he mixed model is a sui able echnique, we add
a base linea mixed model o ob ain he ICC me ic. The in a-
class co ela ion coe icien (ICC) es cap u es he pe cen age o
a iance ha can be explained by di e ences be ween uni s a
highe le el(s), and i is anged be ween 0 and 1. I he ICC es
is close o 0, a mul ile el model is no necessa y. As he alue
inc eases, he e is a g ea e iola ion o he assump ion o inde-
pendence, se ing as an indica o o he need o use a mul ile el
model. In social esea ch, i is common o he ICC o ange be-
ween 0.05 and 0.20 (Peugh2010).
Public expendi u e on educa ion a iables is aken om he Wo ld
Bank, wi h he o iginal sou ce being UNESCO. Addi ionally,
TABLE 2 | Desc ip i e s a is ics o he mac oeconomic a iables employed in he analysis.
Eu opean Social Su ey
N. Obs. Mean S d. De . Min. Max
To al public expendi u e on educa ion, % GDP 364 5.37 1.15 3.3 8.58
Public expendi u e on p ima y educa ion pe s uden , % GDP pe
capi a
217 20.97 4.01 10.79 32.35
Public expendi u e on seconda y educa ion pe s uden , % GDP pe
capi a
217 24.89 4.83 14.33 38.33
Public expendi u e on e ia y educa ion pe s uden , % GDP pe
capi a
217 31.39 10.66 13.29 71.97
Unemploymen a e 364 7.60 3.59 2.02 26.09
In la ion a e 364 2.22 2.27 −4.45 15.88
G oss na ional income pe capi a 364 36,117.83 21,149.14 1540.0 105,070.0
Public expendi u e on gene al goods and se ices, % GDP 364 6.17 1.92 2.67 13.04
Public expendi u e on heal h, % GDP 364 6.17 1.65 1.63 8.87
Public expendi u e on cul u e, % GDP 364 1.28 0.46 0.3 3.46
Sou ce: In e na ional Mone a y Fund(2024) and Wo ld Bank(2024).
14676435, 0, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/kykl.12466 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [23/05/2025]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
7 o 17
mac oeconomic a iables such as in la ion a e, unemploymen
a e, and income pe capi a a e ob ained om he Wo ld Bank.
Da a ela ed o o he public expendi u es a e sou ced om he
COFOG da abase o he In e na ional Mone a y Fund, speci i-
cally, public expendi u e on gene al goods and se ices (GF01),
heal h (GF09), and cul u e (GF08), all exp essed as a pe cen age
o GDP.
To al public expendi u e on educa ion is ela i ized as a pe cen -
age o GDP. Con e sely, public expendi u e on educa ion disag-
g ega ed by educa ion le els (p ima y, seconda y, and e ia y) is
di ided by he numbe o s uden s en olled in each le el and sub-
sequen ly exp essed as a pe cen age o GDP pe capi a. The eason
o no di iding o al public expendi u e on educa ion by he num-
be o s uden s en olled is o p e en loss o obse a ions due o he
una ailabili y o en ollmen da a o ce ain le els, which would
hinde ob aining he o al numbe o s uden s en olled. This ans-
o ma ion me hod allows o a mo e accu a e measu emen o he
coun y's in es men e o by aking in o accoun i s weal h.
5 | Resul s and Discussion
This sec ion p esen s he esul s ha ha e been ob ained om
he di e en eg essions pe o med. The main esul s a e p e-
sen ed in he subsequen subsec ions based on he esea ch
ques ion es ed and clea ly show he posi i e impac o public
spending on educa ion on li e sa is ac ion. In gene al, he di -
e en eg essions ha e es ima ed coe icien s showing he ex-
pec ed signs in line wi h p e ious s udies (Dolan e al. 2008;
Jun 2015; Vene oklis 2019), while hey a e also he subjec o
commen a y in he ex . The coe icien s o he indi idual and
mac oeconomic con ol a iables used a e de ailed in TableA2.
5.1 | The Impac o To al Public Expendi u e in
Educa ion on Well- Being
The esul s ega ding he impac o public expendi u e in educa-
ion—exp essed as a pe cen age o GDP—on li e sa is ac ion a e
p esen ed in Table3. The in o ma ion is o ganized in columns
co esponding o he ype o model used. Conside ing he qual-
i a i e na u e o he endogenous a iables, we employed mixed
o dinal logi eg essions. Howe e , because he a iable o li e sa -
is ac ion consis s o 11 esponse le els, we also use a linea mixed
model. This p o ides an al e na i e in e p e a ion based on ca -
dinali y and se es as a obus ness check o he esul s ob ained.
As shown in he able, he esul s demons a e a clea posi i e
ela ionship be ween public expendi u e on educa ion and li e
sa is ac ion, showing ha public spending on educa ion appea s
o be a use ul ool o aising indi iduals' li e sa is ac ion. As
shown, he ela ionship is s a is ically signi ican in all es ima ed
models. The obus ness o he indings is ema kably high, indi-
ca ing a consis en ela ionship ac oss di e en pe spec i es o
he endogenous a iable. In pa icula , he s a is ically signi i-
can ela ionship is ound in he eg essions pe o med holds o
bo h es ima ion me hods, ca dinal and o dinal. All es ima ed
models ind he es ima o o be s a is ically signi ican wi h a
posi i e sign, p o iding compelling e idence ha in es ing in ed-
uca ion is an e ec i e means o enhancing people's well- being.
This con i ms ou main esea ch ques ion ha public spending
on educa ion has a posi i e impac on li e sa is ac ion and is a
use ul ins umen o inc ease i . These esul s a e in consis en
wi h p e ious s udies (Hessami2010; Kim and Kim2012; O ega
Gil2021; O uza e al.2021) ha had indi ec ly ound his e-
la ionship. We add he base linea mixed model, as Model 1, o
p o e ha he s a is ician echnique is sui able o his da a.
TABLE 3 | E ec o public expendi u e on educa ion on subjec i e well- being in Eu opean coun ies (2002–2020).
Endogenous a iable: Li e sa is ac ion (0–10)
Linea mixed model O dinal logi mixed model
(1) (2) (5)
Indi idual con ol a iables No Yes Yes
Mac oeconomic con ol a iables No Yes Yes
Public expendi u e on educa ion (% GDP) 0.082** 0.108***
(0.034) (0.032)
S d. De . (g oup) 0.906 0.464 0.430
S d. De . (obse a ions) 2.113 1.866
Num. Obs. 428,508 336,521 336,521
Num. G oups 392 364 364
AIC 1,859,195.6 1,376,253.5 1,284,524.3
BIC 1,859,228.5 1,376,543.1 1,284,899.8
ICC 0.16
No e: S anda d e o s in pa en heses, obse a ions nes ed a coun y–yea le el.
Sou ce: Own elabo a ion based on da a used (ESS ERIC2020; In e na ional Mone a y Fund2024; Wo ld Bank2024).
**p < 0.05.
***p < 0.01.
14676435, 0, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/kykl.12466 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [23/05/2025]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
8 o 17 Kyklos, 2025
In addi ion, we can es ima e he quan i a i e e ec using he
coe icien alues. Quan i a i e in e p e a ion should be aken
wi h cau ion, gi en he na u e o he li e sa is ac ion a iable
which, e en in he su ey ha allows o a la ge numbe o
esponses, is always mainly o dinal in na u e. Speci ically, he
coe icien alue indica es he magni ude o he change in wel-
a e ha occu s when public expendi u e on educa ion a ies
by one s anda d uni o GDP, aking in o accoun he p e ious
s anda diza ion. The linea model e eals ha a one- s anda d-
uni inc ease in public expendi u e on educa ion co esponds
o a a ia ion o 0.082 poin s in li e sa is ac ion among espon-
den s. In e p e ed in e ms o a ia ion aw GDP poin s (no
s anda d uni s), his esul ansla es in o an inc ease in li e sa -
is ac ion o be ween 0.015 and 0.149 o each pe cen age poin
inc ease in educa ion spending, wi h a 95% con idence in e al.
O dinal logi mixed model coe icien s ep esen he change in
he log- odds o being in a highe ca ego y o li e sa is ac ion o
a one s anda d de ia ion inc ease in he public expendi u e on
educa ion a iable, holding all o he a iables cons an . When
exponen ia ed, his co esponds o an odds a io o 1.1135, indi-
ca ing ha a one s anda d de ia ion inc ease in public educa ion
expendi u e is associa ed wi h an 11.35% inc ease in he cumu-
la i e odds o epo ing highe li e sa is ac ion.
In e ms o ma ginal e ec s, a one s anda d uni inc ease in
educa ional expendi u e consis en ly dec ease he p obabili y
o epo ing low and medium sa is ac ion le els (0–7), while
inc easing he p obabili y o epo ing high sa is ac ion le els
(8–10). Speci ically, when e alua ed a he mean, his inc ease
educes he p obabili y o epo ing medium sa is ac ion (Le el
5) by 0.81 pe cen age poin s, while inc easing he p obabili y o
epo ing high sa is ac ion le els by 0.99 and 1.02 pe cen age
poin s o Le els 8 and 9, espec i ely, wi h a 0.68 pe cen age
poin inc ease o he maximum le el.
5.2 | E ec s o Public Expendi u e on Educa ion
on Well- Being by Le el o Educa ion
The nex ques ion we analyze conce ns he quan i ica ion o
he impac o public expendi u e on educa ion, dis inguishing
among he le els o educa ion a which he money is alloca ed
(p ima y, seconda y, and e ia y educa ion). Thus, we quan i y
he ex en o which in es men s in public educa ion a e e ec i e
in gene a ing wel a e. In o de o disagg ega e public educa ion
expendi u e by le el, and gi en he absence o published da a,
public expendi u e on educa ion has been adjus ed ela i e o
he numbe o s uden s en olled in each educa ional le el and
exp essed as a pe cen age o GDP pe capi a. Resul s om he es-
ima ions a e p esen ed in Table4. As in he p e ious subsec ion,
we o e sepa a e in o ma ion based on he di e en es ima ion
echniques employed. The econome ic models used in his anal-
ysis a e simila o hose p e iously desc ibed, and we epo only
he coe icien s o key a iables. The de ails o he o he esul s,
he coe icien s o he con ol a iables, a e p o ided in TableA2.
These indings show ha while public expendi u e on educa-
ion has a posi i e e ec on subjec i e well- being, his e ec is
no gene a ed ac oss educa ional le els. In pa icula , he only
le el ha consis en ly p oduces signi ican and posi i e esul s
is e ia y educa ion. We can he e o e conclude ha in es -
men in highe educa ion is he only ype o in es men ha has
a posi i e e ec on well- being. In con as , he o he wo le els
o educa ion conside ed ei he ha e no e ec o e en educe i .
Gi en he lack o p e ious s udies add essing his speci ic ques-
ion ha could se e as a basis o compa ison, i would be in-
e es ing o compa e hese indings wi h es ima es o socie al
and indi idual e u n a es. In gene al, es ima ed e u n a es
end o be mo e uni o m and e en dec ease wi h highe le els o
educa ion (Psacha opoulos and Pa inos2018). Con e sely, he
si ua ion is di e en when assessing he impac on well- being,
as only highe le els o educa ion exhibi a clea in luence.
In he case o public expendi u e on p ima y and seconda y edu-
ca ion, he coe icien s a e no signi ican in any model, showing
ha his le el o educa ion has no e ec on well- being. In he
ligh o he esul s ob ained, i can be seen ha he posi i e e ec
o public educa ion on wel a e is limi ed o highe educa ion,
which seems o be he eal ins umen ha p oduces his esul .
On he o he hand, he o he le els o educa ion ei he ha e no
e ec a all. Ob iously, such a coun e in ui i e esul is bo h su -
p ising and complica ed o in e p e .
TABLE 4 | Public expendi u e on educa ion di e encing by
educa ion le els e ec on subjec i e well- being.
Endogenous a iable:
Li e sa is ac ion (0–10)
Linea
mixed
model
O dinal logi
mixed model
(3) (6)
Indi idual con ol a iables Yes Yes
Mac oeconomic con ol
a iables
Yes Yes
P ima y, public expendi u e
on educa ion by s uden (%
GDP pe capi a)
0.018 0.042
(0.046) (0.042)
Seconda y, public
expendi u e on educa ion by
s uden (% GDP pe capi a)
−0.060 −0.072
(0.049) (0.044)
Te ia y, public expendi u e
on educa ion by s uden (%
GDP pe capi a)
0.159*** 0.180***
(0.053) (0.048)
S d. De . (g oup) 0.516 0.467
S d. De . (obse a ions) 1.871
Num. Obs. 211,320 211,320
Num. G oups 217 217
AIC 865,402.3 807,681.8
BIC 865,699.8 808,061.4
No e: S anda d e o s in pa en heses, obse a ions nes ed a coun y–yea le el.
Sou ce: Own elabo a ion based on da a used (ESS ERIC2020; In e na ional
Mone a y Fund2024; Wo ld Bank2024).
***p < 0.01.
14676435, 0, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/kykl.12466 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [23/05/2025]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
9 o 17
On he one hand, a possible explana ion o his esul could be he
sample o coun ies o which we es ic ed he wo k. In pa icula ,
gi en ha Eu opean coun ies a e he egion in which he highes
sha es o public expendi u e on educa ion ha e been ad anced, he
impac o public expendi u e on educa ion a p ima y and second-
a y le els may indica e a sa u a ion o public expendi u e. Wi h
p ima y and seconda y educa ion ha ing been uni e salized wi h
ela i ely high s anda ds o quali y in Eu opean coun ies, one ad-
di ional uni o expendi u e may no ha e a signi ican e ec and
can e en become nega i e i sec ions o he popula ion p e e o he
o ms o educa ion. Agains his backg ound, he da a show ha
in es men in highe educa ion is qui e p oduc i e in gene a ing
wel a e, possibly because he e a e s ill widesp ead sho ages and
de iciencies and only pa o he popula ion can a o d his ype o
educa ional se ice. In addi ion, highe educa ion can gene a e im-
po an social e u ns in e ms o ex e nali ies, h ough he labo
ma ke o h ough R&D&I, knowledge ans e , and spillo e e -
ec s p oduced by uni e si ies. The esul may he e o e e lec his
si ua ion. This implies ha in socie ies whe e mos people ha e
p ima y o seconda y educa ion, di e en ia ion is mainly h ough
highe educa ion. I is easonable o assume ha enhancing he
quali y o inc easing he quan i y o highe educa ion could ha e
a g ea e impac on indi idual well- being, pa icula ly when he
supply is emains ela i ely low compa ed o o al demand. In such
cases, public in es men may be pe cei ed as highly cos - e ec i e in
e ms o wel a e.
On he o he hand, he limi ed con ibu ion o public expendi u e
on basic and seconda y educa ion o subjec i e well- being could be
in e p e ed as indi iduals no ully accoun ing o he signi ican
in e empo al e ec s o public educa ion spending. In eali y, in-
c eases in public unding o educa ion equi e conside able ime
o ma e ialize in o angible educa ional bene i s ha indi iduals
can pe cei e. In socie ies such as hose in Eu ope, whe e public ed-
uca ion a lowe le els is ully es ablished, his expendi u e may be
aken o g an ed, and any inc eases a e likely aimed a enhancing
se ice quali y. Howe e , such imp o emen s may no lead o im-
media e well- being gains, as hei bene i s o en become appa en
only when he ecipien s each adul hood and in eg a e in o he
labo ma ke and socie y as a whole. Consequen ly, hese expendi-
u es may no gene a e sho - e m well- being imp o emen s and
could e en explain he nega i e sign i hei unding has ad e sely
a ec ed o he policies o ce ain g oups.
Bo h in e p e a ions a e no mu ually exclusi e and can be seen as
complemen a y explana ions o he coe icien s we ha e es ima ed.
5.3 | How he E ec o Public Expendi u e on
Educa ion Changes When Conside ing Poli ical
P e e ences
The hi d exe cise ha we ca y ou is o analyze whe he indi-
idual poli ical p e e ences shape wha e ec public expendi u e
in educa ion has on subjec i e well- being. In his way, we explic-
i ly ake in o accoun he possibili y o signi ican he e ogenei y
be ween indi iduals when in e p e ing he e ec s o public ed-
uca ion and he amoun s spen on i . The conc e e way in which
we app oach his analysis is by in oducing a a iable ob ained
by mul iplying he main exogenous a iable, public expendi-
u e on educa ion, by ano he a iable de ining he ideological
p e e ences o indi iduals. The cha ac e iza ion o indi iduals
acco ding o hei ideological p e e ence a iable is de i ed om
he su eys hemsel es. Speci ically, om a ques ion in he su -
ey, whe e indi iduals a e asked o posi ion hemsel es on a dis-
c e e scale o poli ical ideology. In he ESS, i anges om 0 o 10,
wi h he lowe end ep esen ing he poli ical le and he highe
end ep esen ing he poli ical igh .
In o de o ensu e compa abili y be ween su eys and o acili a e
in e p e a ion and calcula ions, we ha e chosen o con e he
ini ial esponses in o a h ee- le el a iable: le , cen e , and igh ,
which allow indi iduals o be cha ac e ized. Speci ically, we adop
he ollowing g ouping: le i he indi idual posi ions hemsel es
wi hin he esponse ange o (0–4), cen e (5–6), and igh (7–10).
Following he s uc u e o he p e ious subsec ions, Table 5
p esen s he coe icien s o key a iables o add ess he esea ch
ques ion. By in oducing an in e ac ion be ween he ca ego ical
a iable o ideological p e e ences—wi h he cen e ideology as
he base le el—and public expendi u e on educa ion, we can
calcula e he e ec o public educa ion on wel a e when judged
by a pe son wi h di e en ideological p e e ences.
In he ligh o he esul s ob ained, we can conclude ha al-
hough he e ec o public educa ion as an ins umen o
TABLE 5 | E ec o indi idual poli ical p e e ences on sa is ac ion
gene a ed by public expendi u e on educa ion.
Endogenous a iable:
Li e sa is ac ion (0–10)
Linea mixed
model
O dinal logi
mixed model
(4) (7)
Indi idual con ol
a iables
Yes Yes
Mac oeconomic con ol
a iables
Yes Yes
Public expendi u e on
educa ion (% GDP)
0.103*** 0.130***
(0.034) (0.032)
Pub. Exp. Edu. * Le 0.006 0.009
(0.008) (0.007)
Pub. Exp. Edu. * Righ −0.081*** −0.067***
(0.008) (0.007)
S d. De . (g oup) 0.465 0.435
S d. De . (obse a ions) 1.865
Num. Obs. 336,521 336,521
Num. G oups 364 364
AIC 1,376,129.6 1,284,437.2
BIC 1,376,440.7 1,284,834.1
No e: S anda d e o s in pa en heses, obse a ions nes ed a coun y–yea le el.
Sou ce: Own elabo a ion based on da a used (ESS ERIC2020; In e na ional
Mone a y Fund2024; Wo ld Bank2024).
***p < 0.01.
14676435, 0, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/kykl.12466 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [23/05/2025]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License
16 o 17 Kyklos, 2025
(1) (2) (3) (4) (5) (6) (7)
Ma i al s a us—Di o ced −0.563*** −0.579*** −0.564*** −0.517*** −0.523*** −0.517***
(0.012) (0.015) (0.012) (0.011) (0.014) (0.011)
Ma i al s a us—Widowed −0.508*** −0.493*** −0.507*** −0.492*** −0.472*** −0.492***
(0.013) (0.017) (0.013) (0.013) (0.016) (0.013)
Ma i al s a us—Ne e ma ied −0.401*** −0.406*** −0.402*** −0.418*** −0.423*** −0.419***
(0.010) (0.012) (0.010) (0.009) (0.012) (0.009)
Unemployed −0.989*** −0.974*** −0.989*** −0.856*** −0.832*** −0.856***
(0.015) (0.019) (0.015) (0.014) (0.018) (0.014)
Educa ion le el—Seconda y 0.136*** 0.137*** 0.139*** 0.099*** 0.106*** 0.100***
(0.014) (0.018) (0.014) (0.014) (0.017) (0.013)
Educa ion le el—Te ia y 0.386*** 0.397*** 0.388*** 0.294*** 0.296*** 0.294***
(0.015) (0.020) (0.015) (0.015) (0.019) (0.015)
Heal h—Good −0.507*** −0.511*** −0.508*** −0.591*** −0.591*** −0.591***
(0.009) (0.011) (0.009) (0.008) (0.010) (0.008)
Heal h—A e age −1.177*** −1.179*** −1.178*** −1.234*** −1.226*** −1.235***
(0.010) (0.013) (0.010) (0.010) (0.012) (0.010)
Heal h—Bad −2.119*** −2.109*** −2.120*** −2.047*** −2.025*** −2.048***
(0.015) (0.019) (0.015) (0.015) (0.019) (0.015)
Heal h—Ve y bad −3.084*** −3.103*** −3.084*** −2.858*** −2.866*** −2.859***
(0.030) (0.038) (0.030) (0.031) (0.039) (0.031)
Sex—Woman 0.109*** 0.100*** 0.108*** 0.124*** 0.118*** 0.123***
(0.007) (0.008) (0.007) (0.006) (0.008) (0.006)
Immig an −0.168*** −0.191*** −0.169*** −0.169*** −0.191*** −0.170***
(0.012) (0.015) (0.012) (0.011) (0.014) (0.011)
Poli ical p e e ences—Le −0.184*** −0.188*** −0.187*** −0.173*** −0.179*** −0.175***
(0.008) (0.010) (0.008) (0.007) (0.009) (0.007)
Poli ical p e e ences—Righ 0.309*** 0.316*** 0.318*** 0.321*** 0.325*** 0.328***
(0.008) (0.010) (0.008) (0.008) (0.010) (0.008)
No e: S anda d e o s in pa en heses, obse a ions nes ed a coun y–yea le el.
Sou ce: Own elabo a ion based on da a used (ESS ERIC2020; In e na ional Mone a y Fund2024; Wo ld Bank2024).
*p < 0.1.
**p < 0.05.
***p < 0.01.
TABLE A2 | (Con inued)
14676435, 0, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/kykl.12466 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [23/05/2025]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License

17 o 17
TABLE A3 | Es ima ion esul s o he con ol a iables om he
Eu oba ome e .
(8) (9) (10)
Th eshold poin s: 1|2 −3.402*** −3.394*** −3.402***
(0.035) (0.046) (0.034)
2|3 −1.467*** −1.377*** −1.467***
(0.034) (0.045) (0.034)
3|4 1.725*** 1.828*** 1.725***
(0.034) (0.045) (0.034)
Mac oeconomic con ol a iables
Unemploymen a e −0.206*** −0.064 −0.206***
(0.037) (0.047) (0.035)
In la ion a e −0.098*** −0.145*** −0.098***
(0.033) (0.041) (0.033)
GDP pe capi a 0.536*** 0.646*** 0.536***
(0.034) (0.045) (0.034)
Public expendi u e
on gene al se ices (%
GDP)
−0.111*** −0.160*** −0.111***
(0.037) (0.049) (0.036)
Public expendi u e on
heal h (% GDP)
−0.027 −0.063*−0.027
(0.029) (0.038) (0.030)
Public expendi u e on
cul u e (% GDP)
0.016 0.118*** 0.016
(0.029) (0.032) (0.028)
Indi idual con ol a iables
Age −0.120*** −0.135*** −0.120***
(0.006) (0.008) (0.006)
Squa ed age 0.212*** 0.237*** 0.212***
(0.004) (0.006) (0.004)
Sex—Man −0.103*** −0.077*** −0.103***
(0.007) (0.009) (0.007)
Poli ical
p e e ences—Le
−0.147*** −0.145*** −0.148***
(0.008) (0.010) (0.008)
Poli ical
p e e ences—Righ
0.136*** 0.164*** 0.137***
(0.008) (0.011) (0.008)
Household size—2 0.130*** 0.114*** 0.130***
(0.012) (0.016) (0.011)
Household size—3 0.086*** 0.086*** 0.086***
(0.013) (0.017) (0.013)
Household size—4 o
mo e
0.124*** 0.125*** 0.124***
(0.013) (0.017) (0.013)
Ma i al s a us—Li ing
wi h pa ne
−0.205*** −0.238*** −0.205***
(0.012) (0.016) (0.012)
(Con inues)
(8) (9) (10)
Ma i al s a us—Single −0.417*** −0.439*** −0.418***
(0.012) (0.016) (0.012)
Ma i al s a us—
Sepa a ed o di o ced
−0.653*** −0.653*** −0.653***
(0.014) (0.018) (0.014)
Ma i al
s a us—Widowed
−0.493*** −0.491*** −0.493***
(0.015) (0.020) (0.015)
Age
educa ion—16–19 yea s
0.203*** 0.240*** 0.204***
(0.010) (0.013) (0.010)
Age educa ion—Mo e
han 20 yea s
0.513*** 0.550*** 0.514***
(0.011) (0.015) (0.011)
Age educa ion—S ill
s udying
0.034 0.024 0.035
(0.074) (0.107) (0.080)
Occupa ion—Manage s 0.196*** 0.157*** 0.196***
(0.015) (0.021) (0.015)
Occupa ion—O he
whi e colla s
−0.062*** −0.114*** −0.062***
(0.015) (0.021) (0.015)
Occupa ion—Manual
wo ke s
−0.236*** −0.290*** −0.236***
(0.014) (0.019) (0.013)
Occupa ion—House
pe son
−0.251*** −0.307*** −0.251***
(0.017) (0.025) (0.017)
Occupa ion—
Unemployed
−1.140*** −1.141*** −1.141***
(0.017) (0.024) (0.017)
Occupa ion—Re i ed −0.266*** −0.318*** −0.267***
(0.015) (0.021) (0.015)
Occupa ion—S uden 0.429*** 0.478*** 0.428***
(0.075) (0.109) (0.080)
No e: S anda d e o s in pa en heses, obse a ions nes ed a coun y–yea le el.
Sou ce: Own elabo a ion based on Eu opean Commission(2020), In e na ional
Mone a y Fund(2024), and Wo ld Bank(2024).
*p < 0.1.
**p < 0.05.
***p < 0.01.
TABLE A3 | (Con inued)
14676435, 0, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/kykl.12466 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [23/05/2025]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License