Unveiling the mutual dynamics: Institutions, education and economic growth over 138 years in OECD countries

Rö hel, Tim; Leschke, Ma in
A icle — Published Ve sion
Un eiling he mu ual dynamics: Ins i u ions, educa ion
and economic g ow h o e 138 yea s in OECD coun ies
Kyklos
P o ided in Coope a ion wi h:
John Wiley & Sons
Sugges ed Ci a ion: Rö hel, Tim; Leschke, Ma in (2024) : Un eiling he mu ual dynamics: Ins i u ions,
educa ion and economic g ow h o e 138 yea s in OECD coun ies, Kyklos, ISSN 1467-6435, Wiley
Pe iodicals, Inc., Hoboken, NJ, Vol. 78, Iss. 1, pp. 243-270,
h ps://doi.o g/10.1111/kykl.12416
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/313689
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h p://c ea i ecommons.o g/licenses/by/4.0/
ORIGINAL ARTICLE
Un eiling he mu ual dynamics: Ins i u ions,
educa ion and economic g ow h o e 138 yea s in
OECD coun ies
Tim Rö hel
1
| Ma in Leschke
2
1
Ge man Ins i u e o De elopmen and
Sus ainabili y (IDOS), Bonn, Ge many
2
Depa men o Law, Business, and
Economics, Uni e si y o Bay eu h, Bay eu h,
Ge many
Co espondence
Tim Rö hel, Ge man Ins i u e o De elopmen
and Sus ainabili y (IDOS), Tulpen eld 6, 53113
Bonn, Ge many,
Email: im. oe hel@idos- esea ch.de
Abs ac
Many a icles examine he e ec o ins i u ions and
educa ion on economic g ow h in g ow h models o
empi ical se ings. Howe e , he opposi e e ec o
economic de elopmen on ins i u ions and educa ion is less
esea ched. Sys ema ic e idence is sca ce when looking a
he in e play o educa ion and ins i u ions. Mos o he li e -
a u e ocuses on pos -wa e idence s a ing in he 1960s
only. Howe e , ins i u ions and educa ion, bo h, only
change slowly o e ime and, hus, should be analysed o e
a longe pe iod. In his a icle, we y o close a gap by
examining he ela ionship be ween all h ee a iables o e
138 yea s. By es ima ing a ec o e o co ec ion model o
a sample o 20 O ganisa ion o Economic Co-ope a ion
and De elopmen (OECD) coun ies, we a e addi ionally
able o di e en ia e be ween sho - and long- un esul s.
We ind posi i e e ec s o ins i u ions and educa ion on
g ow h and ice e sa. In con as , ins i u ions and educa-
ion do no signi ican ly a ec each o he . Ou esul s a e
obus o se e al obus ness es s and ex ensions.
KEYWORDS
g ow h, human capi al, ins i u ions, long- un, PMG es ima o
Recei ed: 15 Decembe 2023 Re ised: 10 July 2024 Accep ed: 26 Augus 2024
DOI: 10.1111/kykl.12416
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.
© 2024 The Au ho (s). Kyklos published by John Wiley & Sons L d.
Kyklos. 2025;78:243–270. wileyonlinelib a y.com/jou nal/kykl 243
1|INTRODUCTION
Acco ding o ins i u ional economis s, such as Douglass (e.g. No h, 1990), James (e.g. Buchanan, 1975),
Da on Acemoglu (e.g. 2001) and o he s, ins i u ions play a c ucial ole in economic de elopmen , shaping he
incen i es, beha iou s and in e ac ions o indi iduals and o ganisa ions wi hin a socie y. Ins i u ions impac
economic de elopmen in h ee main ways: Fi s , p ope y igh s a e undamen al o economic de elopmen .
When indi iduals and businesses ha e con idence ha hei p ope y will be p o ec ed and ha con ac s will
be en o ced, hey a e mo e likely o in es , inno a e and engage in p oduc i e economic ac i i ies. Second,
e ec i e ma ke ins i u ions, such as compe i ion laws, egula o y amewo ks and consume p o ec ions, c ea e
a le el playing ield o businesses. They p e en monopolies, encou age ai compe i ion and ensu e ma ke s
ope a e e icien ly, leading o inno a ion and economic g ow h. Thi d, s able poli ical ins i u ions gua an ee
checks and balances and, he e o e, p e en exploi a ion by he s a e, c ea e an en i onmen o ce ain y, keep
co up ion down and encou age long- e m in es men . While neoclassical economis s o en implici ly included
ins i u ions in hei models, i ook some ime un il hey used such a iables egula ly and explici ly
(No h, 1990). Dawson (1998), o example, ex ends he neoclassical model by Mankiw e al. (1992) by inco po-
a ing an ins i u ional a iable.
Besides ins i u ions, educa ion (o human capi al) is ano he c i ical de e minan o economic de elopmen . One
ob ious explana ion o his ela ionship is he highe p oduc i i y o wo ke s esul ing om mo e educa ion. The
highe le el o p oduc i i y leads o mo e ou pu and hus economic de elopmen o g ow h (e.g. Lucas, 1988).
The empi ical li e a u e (e.g. Acemoglu e al., 2018; Mankiw e al., 1992. sec ion 2) suppo s he hypo hesis ha
ins i u ions as well as educa ion a e bo h c i ical ac o s in luencing he o al ac o p oduc i i y. Thus, in his a icle,
we will examine he ela ionships be ween hese h ee a iables in mo e de ail.
Howe e , de elopmen is no a one-way s ee . Highe economic de elopmen , and hus highe weal h, migh
lead o highe educa ion spending and a highe demand o educa ion (e.g. Glewwe & Jacoby, 2004). Addi ionally,
wi h highe weal h, people migh be be e able o pa icipa e in he poli ical p ocess and push o be e ins i u ions
(e.g. Tian e al., 2020). E en ually, wha can we say abou he ela ionship be ween ins i u ions and educa ion?
People migh only ully e lec on he impo ance o ins i u ions and make he igh o ing decisions i hey a e
su icien ly educa ed (e.g. Fo una o & Panizza, 2015). Mo eo e , suppose he e is no oom o de elopmen due
o a lack o p ope y igh s o o he ins i u ional es ic ions. In ha case, he e is no need o in es in one's own
educa ion. This could be seen in many So ie coun ies o o he au oc acies whe e ca ee s a e no he esul o
pe sonal e o bu a e ied o pa y membe ships o loyal y o he egime. These hypo heses will be discussed in
mo e de ail in Sec ion 2.
Agains his backg ound, we would like o close wo esea ch gaps. Fi s , as laid ou abo e, we do no only
analyse o which ex en ins i u ions and human capi al os e GDP g ow h. We will also analyse he impac o human
capi al and GDP g ow h on ins i u ion-building and whe he GDP g ow h and ins i u ions will aise human capi al.
Secondly, we will analyse hese in e connec ions o e a long pe iod ( om 1870 o 2007) o 20 OECD coun ies
using panel da a. We will use an e o co ec ion model o cap u e long- un simila i ies be ween hese coun ies and
sho - un di e ences ac oss he coun ies. In he li e a u e, he ela ionship be ween hese h ee a iables has ye o
be sys ema ically examined, especially o e such a long pe iod.
The pape 's main indings a e ha ins i u ions and human capi al signi ican ly os e g ow h. These esul s a e
obus o a ious ex ensions. Fu he , GDP g ow h posi i ely a ec s human capi al and ins i u ion building.
Howe e , he e ec s o educa ion and ins i u ions on one ano he a e no obus .
The cou se o his pape is as ollows: Sec ion 2p o ides an o e iew o he li e a u e wi h espec o ou
hypo heses. Sec ion 3in oduces he da a and he empi ical amewo k, whe eas Sec ion 4p esen s he main
indings. Ex ensions a e discussed in Sec ion 5. The pape inishes wi h a sho conclusion.
244 RÖTHEL and LESCHKE
2|LITERATURE REVIEW
O e ime, inc eases in he quan i y and quali y o physical capi al, labou and echnical and o ganisa ional p og ess
should shi po en ial GDP p.c. (Islam, 1995; Mankiw e al., 1992). This is he wisdom o he ‘old g ow h heo y’
(Solow, 1956). In es men in a compe i i e se ing os e s GDP g ow h h ough imp o ed p oduc i i y, echnological
p og ess, openness and secu e p ope y igh s. Mo eo e , he geog aphical loca ion and cul u e may in luence
ins i u ion-building and in es men and, he e o e, should impac GDP p.c. (e.g. Acemoglu e al., 2001; Eas e ly &
Le ine, 2003; Rod ik e al., 2004).
Since he 1990s, economis s such as Rod ik (2000) and many o he s a gued ha ins i u ions a e impo an
de e minan s o g ow h and ha p ope y igh s ep esen a co e ca ego y o economic ins i u ions. The heo e ical
backg ound o his hypo hesis is much olde s a ing wi h publica ions o ea ly enligh enmen philosophe s like
Hobbes (1651) and Locke (1690). The amous philosophe Hobbes (1651) a gues ha wi hou law and o de
imposed by he s a e, li e will be a wa o ‘each agains all’(Bellum omnium con a omnes), ‘…and he li e o man
soli a y, poo , nas y, b u ish, and sho ’. Locke (1690) la e emphasises ha people should ha e undamen al igh s,
such as he igh s o li e, libe y and p ope y. Fu he , Mon esquieu (1748) s a es ha a bi a y go e nmen al ac ion
agains he will and he wel a e o he people can only be a oided by implemen ing ce ain cons ain s on he
go e nmen , o example, he sepa a ion o di ision o he s a e's powe s (in o legisla i e, execu i e and judicia y
powe ). The economic philosophe s Hayek (1960) and Buchanan (1975) ha e pushed his idea u he .
Mo e ecen ly, Acemoglu e al. (2004) a ibu ed g ea impo ance o he s a e's ole in o malising and
p o ec ing p ope y igh s. The ocus on ins i u ions, such as p ope y igh s and poli ical cons ain s, does no mean
we should neglec o he a iables' impac on GDP p.c. I en ails only he hypo hesis ha wi hou p ope y igh s and
poli ical cons ain s, sus ainable GDP p.c. g ow h canno se iously be explained in he long- un. Un o una ely,
ins i u ional a iables, such as ule o law, p ope y igh s, o cons i u ional law, canno be di ec ly measu ed (a leas
his o ically). This migh be a p oblem, especially i one is in e es ed in de e mining he quali y o ins i u ions in de ail.
Agains his backg ound, one solu ion is o use p oxies.
Glaese e al. (2004) use se e al di e en ins i u ional measu es o es ima e he e ec on economic g ow h.
Some a iables, o example, execu i e cons ain s o he exp op ia ion isk, a e signi ican and posi i e; o he s, o
example, judicial independence, a e no . Ano he mo e ecen s udy by Góes (2016) indica es a posi i e ela ionship
be ween economic eedom and GDP g ow h. Finally, in hei me a- eg ession analyses, E endic e al. (2011) and
Colag ossi e al. (2020) show ha o e all he e idence indica es a posi i e e ec o ins i u ions o democ acy on
economic pe o mance. This leads us o ou i s hypo hesis.
Hypo hesis 1. Ins i u ions ha e a posi i e e ec on economic g ow h.
In he Keynesian g ow h model as well as in he neo-classical models o he Solow ype, physical capi al and
exogenous echnological p og ess played a dominan ole. Since he mid o he 1980s, endogenous g ow h heo ies
(e.g. Lucas, 1988; Rome , 1986,1990) iden i ied human capi al and educa ion as impo an ac o s con ibu ing o
long- un economic g ow h. A as amoun o empi ical esea ch suppo s he hypo hesis ha be e educa ion
(human capi al) will lead o highe g ow h a es (e.g. Acemoglu & Ang is , 2000; Ba o, 2000; K uege &
Lindahl, 2001; Mankiw e al., 1992). Eh lich and Mu phy (2007) desc ibe human capi al as an ‘in angible asse , bes
hough o as a s ock o embodied and disembodied knowledge, comp ising educa ion, in o ma ion, heal h, en ep e-
neu ship and p oduc i e and inno a i e skills’. Inc eases in human capi al a ise h ough in es men in schooling, job
aining and heal h, as well as h ough esea ch and de elopmen p ojec s and in o mal knowledge ans e s. Ma coni
(2018) inds a posi i e e ec o he a e age yea s o educa ion on GDP g ow h o e a long ime be ween 1980 and
2005 o 118 coun ies. Indeed, he idea ha educa ion and human capi al a e impo an ac o s o he de elop-
men o an indi idual and a coun y can al eady be ound in he wo ks o ‘old’economis s such as Smi h (1776),
Pigou (1920) and Machlup (1972). E en hough he e idence o a posi i e e ec o educa ion seems clea a i s ,
o he s udies do no ind any signi ican e ec s o highe school en olmen and educa ional a ainmen
RÖTHEL and LESCHKE 245
(e.g. Islam, 1995; Knowles & Owen, 1995). This e ec migh e en u n nega i e abo e a ce ain h eshold o
educa ion. McGowan and And ews (2017) a gue ha highe skill and quali ica ion misma ch is associa ed wi h lowe
labou p oduc i i y and, hus, po en ially wi h lowe g ow h a es. Finally, Benos and Zo ou (2014) summa ise hese
esul s in hei me a- eg ession analysis, poin ing a he la ge p oblem o publica ion bias inhe en in his body o
li e a u e. Ne e heless, we ollow he endogenous g ow h heo ies and p opose he ollowing hypo hesis.
Hypo hesis 2. Human capi al has a posi i e e ec on economic g ow h.
E en i mos esea che s analyse he impac o ins i u ions on g ow h, a e e sed in luence is concei able, oo.
Diamond (1999) a gues ha ag icul u al p og ess and ea ly economic g ow h enabled be e ins i u ions. Mo eo e ,
Lipse (1959) and Moo e (1967) suppo he idea ha highe income is a basis o c ea ing be e ins i u ions. Also,
some olde analyses o democ a ic ansi ion p omo e he hypo hesis ha highe economic de elopmen makes i
easie o in oduce democ a ic ins i u ions (Kuzne s, 1965,1966). Following he a gumen by Acemoglu and
Robinson (2000), a highe le el o economic de elopmen migh lead o he popula ion demanding mo e edis ibu-
ion om he poli ical eli es. In o de o a oid un es o e olu ion, he eli es ex end democ a ic ins i u ions ensu ing
u u e pa icipa ion o he public in he edis ibu ion p ocess. Ne e heless, a c i ical junc u es hypo hesis o mu-
la ed by No h (1990) and Acemoglu e al. (2004) emphasises ha he main di ec ion o causali y is om democ acy
o ins i u ions o income.
In hei heo e ical wo k, Zak and Feng (2003) demons a e ha economic g ow h is an impo an ac o in he
speed o ansi ion o democ acy. The esul s o hese heo e ical wo ks a e suppo ed empi ically by Heid e al.
(2012), who ind a posi i e ela ionship be ween income and democ acy.
In he long- un, Gundlach and Paldam (2009) and Tian e al. (2020) p o ide e idence o a s ong and posi i e
e ec o income on democ acy and ins i u ions, espec i ely. They a gue ha income migh be one o he mos c i i-
cal de e minan s o democ a ic ansi ion. Acemoglu e al. (2008) ind a posi i e co ela ion bu no causal e ec o
income on democ acy o e a pe iod o 500 yea s. All in all, we hypo hesise he ollowing ela ionship.
Hypo hesis 3. Economic g ow h has a posi i e e ec on ins i u ions.
Acco ding o Lipse (1959), weal h may lead o be e educa ion sys ems and social capi al, inducing be e
democ a ic ins i u ions (e.g. p ope y igh s). Djanko e al. (2003) add ha mo e educa ed people a e mo e likely o
sol e con lic s in a egula ed manne and less likely o use iolence. Glaese e al. (2004) ind ha he le el o
schooling signi ican ly a ec s he change in ins i u ions in he ollowing yea s. Howe e , Acemoglu e al. (2005)
show ha ime e ec s migh d i e his e ec . When including hese ime e ec s in he eg ession o Glaese e al.
(2004), hey canno con i m he signi ican e ec o schooling on democ acy.
In con as , Ape gis and Payne (2017) ind a posi i e ela ionship be ween educa ion and democ acy. The e ec
o educa ion on democ acy appea s o be s onge in coun ies wi h lowe income le els. The s udies by Bobba and
Co iello (2007) and Cas elló-Climen (2008) suppo hese esul s o e a longe pe iod be ween 1965 and 2000
and 1970 and 2000, espec i ely, o a la ge sample o coun ies. On he o he hand, Fo una o and Panizza (2015)
ind a posi i e co ela ion be ween educa ion and he quali y o go e nmen in coun ies wi h highe le els o
democ acy.
Glaese e al. (2007) de elop a heo e ical model o analyse he channel be ween educa ion and
democ acy. They a gue ha educa ion leads o highe ci ic engagemen , and well-educa ed people bene i ela i ely
mo e om democ acy. Thus, in an au oc a ic egime, once people demand mo e educa ion due o a highe e u n on
educa ion, he au oc a needs o p o ide mo e educa ion in o de o p e en a e olu ion. Pa en e e al. (2022) a gue
ha his leads o a mode nisa ion p ocess which e en ually ends in a democ acy. This heo y is suppo ed by an
empi ical s udy by Ape gis (2018), p o iding new e idence om 161 coun ies ha educa ion suppo s democ acy.
Simila ly, Milligan e al. (2004) ind ha educa ion inc eases a en ion o and in e es in poli ics and hus leads o
mo e in o med and ‘be e ’ o ing beha iou . Thus, we p opose he ollowing hypo hesis.
Hypo hesis 4. Human capi al has a posi i e e ec on ins i u ions
246 RÖTHEL and LESCHKE

Bjø nsko (2009) examines he e ec o ini ial GDP p.c. and social capi al on he g ow h a e o he a e age
schooling leng h. Social capi al is posi i ely ela ed o schooling leng h. Howe e , ini ial GDP p.c. nega i ely a ec s
he a e age schooling leng h. The au ho in e p e s his esul as he high ini ial in es men o weal hie coun ies o
imp o e he educa ion sys em. Oke ch (2006) a gues ha iche coun ies in es mo e in educa ion in o de
o achie e highe g ow h a es. He inds ha pe capi a g ow h posi i ely a ec s in es men in educa ion and, hus,
inc eases human capi al o ma ion. Addi ionally, on a mic o le el, se e al s udies in es iga e he e ec o amily
income on child pe o mance in school. Duncan e al. (2014), o example, a gue ha low-income amilies lack he
esou ces o in es in educa ion. A boos in pa en s' income leads o be e school pe o mance. This is in line wi h
Glewwe and Jacoby (2004). They ind ha an inc ease in household weal h leads o a signi ican inc ease in school
en olmen a es. They conclude ha human capi al does no only a ec g ow h, bu a ein o cing p ocess be ween
hese wo a iables exis s. O e all, we assume ha highe GDP p.c. leads o a posi i e e ec on human capi al.
Hypo hesis 5. Economic g ow h has a posi i e e ec on human capi al.
Eh lich and Lui (1999) p o ide a heo e ical amewo k o he e ec o ins i u ions on human capi al. They
a gue ha weak ins i u ions will ge indi iduals o in es in en -seeking ac i i ies ins ead o human capi al. Thus, in
he absence o s ong ins i u ions, he le el o human capi al should be ela i ely lowe . Na kho and Polishchuk
(2019) demons a e how he alloca ion o human capi al be ween p oduc i e and unp oduc i e sec o s is in luenced
by he quali y o ins i u ions. Glaese e al. (2004) also es ima e he e ec o he ini ial le el o ins i u ions and o
GDP p.c. on he g ow h o educa ion. While GDP p.c. has a obus and signi ican posi i e e ec , he di e en mea-
su es o ins i u ions exhibi no such e ec . In con as , Rod ik e al. (2004) show ha ule o law posi i ely in luences
he human capi al pe wo ke . B own and Hun e (2004) and S asa age (2005) bo h examine he e ec o democ acy
on educa ion spending, especially o p ima y schools. They ind a posi i e ela ion o La in Ame ican and A ican
coun ies, espec i ely. Samuels and Va gas (2023) con i m his esul o a la ge sample, in he absence o ag a ian
eli es. O he s udies examine he link be ween ins i u ions and a ious educa ional ou comes. Tiongson (2000), Mo
(2001) and Abdulla (2021) ind nega i e e ec s o co up ion on educa ional ou come a iables.
In con as , Pelleg ini and Ge lagh (2004) ind no signi ican ela ionship be ween co up ion and schooling. The
esul has been con i med by Pelleg ini (2011) using a newe da ase . Finally, his leads o ou las hypo hesis.
Hypo hesis 6. Ins i u ions ha e a posi i e e ec on human capi al.
In his a icle, he long- un e idence is o pa icula in e es . Thus, we will u n o long- un s udies, o example,
using g ow h o GDP p.c. as he dependen a iable. Almos all long- un s udies examine he pe iod be ween 1960
and he beginning o he millennium. Ge ing e al. (2005) and Acemoglu e al. (2018), o example, look a he e ec
o democ acy on g ow h in a la ge sample o coun ies be ween 1950 and 2000 and 1960 and 2010, inding posi i e
and signi ican e ec s. G ündle and K iege (2016) con i m hese esul s. Simila ly, Hanushek and Woessmann
(2007,2012) examine he e ec o schooling on economic g ow h and ind posi i e and signi ican e ec s. The s ud-
ies abo e mainly use ixed e ec s o sys em GMM es ima o s o iden i y he e ec s. Bassanini and Sca pe a (2002)
es ima e he human capi al augmen ed g ow h model using he pooled mean g oup (PMG) es ima o . The e in he
long- un, human capi al signi ican ly inc eases he g ow h a e. Lee and Kim (2009), Ri e a-Ba iz (2002) and Dias and
Tebaldi (2012) show ha bo h, educa ion and policies, ma e o economic g ow h. An olde s udy by O'Rou ke
and Williamson (1997) p o ides some his o ical e idence wi h espec o globalisa ion, schooling and g ow h in he
Eu opean pe iphe y be ween 1870 and 1913. They iden i y schooling as a d i e o con e gence o some coun ies
wi hin hei sample. Finally, Rö hel and Leschke (2023) examine he e ec o ins i u ions and educa ion in a neoclas-
sical g ow h model be ween 1870 and 2007. They ind obus posi i e and signi ican e ec s o bo h a iables.
In con as o mos o hese s udies, we analyse he mu ual in luence o he h ee a iables, GDP p.c., human
capi al and ins i u ions, essen ial o social de elopmen . This allows us o es i hese a iables ollow a i uous
ci cle since he dynamics be ween he a iables a e su icien ly accoun ed o he i s ime. Since we es ima e he
ela ionship o e a long ime, we a e able o di e en ia e be ween sho - and long- un e ec s o he a iables on
RÖTHEL and LESCHKE 247
one ano he . As in Bassanini and Sca pe a (2002), we use he PMG es ima o , oo, since i is he mos app op ia e
es ima o o a se ing wi h many pe iods. We will discuss he es ima o in mo e de ail in he ollowing sec ion.
3|EMPIRICAL FRAMEWORK
3.1 |Da a
In his s udy, he ela ionship be ween economic g ow h, ins i u ions and educa ion, in he long- un, is o pa icula
in e es . The da ase comp ises 20 coun ies be ween 1870 and 2007. Da a a ailabili y limi s ou ime ho izon.
Since ou da a s em om di e en da a sou ces p o iding such da a o e a long ime, we ace a ade-o be ween
he leng h and wid h o he da a. By analysing 20 coun ies o e he abo e-men ioned pe iod, we y o sol e
his issue. As in many p e ious s udies (e.g. Acemoglu e al., 2008), we measu e economic g ow h using he GDP
p.c. da a om he Maddison P ojec Da abase (Bol & Van Zanden, 2020). Ins i u ions may be o mal, such as laws,
o in o mal, such as no ms o alues (No h, 1990). In his a icle, we ocus on o mal ins i u ions, such as he legal
amewo k in a coun y. E en hough some limi a ions may a ise, we decided o p oxy he legal amewo k by
using one ins i u ional a iable. This is he usual app oach in he li e a u e. Conce ning his ins i u ional a iable,
we al eady discussed he a gumen by Buchanan (1975) in a ou o p ope y igh s o binding ules o poli icians
abo e. This is in line wi h he esul s o Acemoglu and Johnson (2005) ha p ope y igh s ha p o ec he
ci izens' igh s agains he go e nmen a e o i s -o de impo ance o economic ou comes. Thus, as a p oxy o
ins i u ions, we use he libe al democ acy index om he Va ie ies o Democ acy Ins i u e. This index akes a
nega i e iew o poli ical powe and emphasises indi idual igh s agains he y anny o he s a e (Coppedge e al.,
2022). Finally, we u n o he educa ion a iable. Fo ou pu pose, educa ional a ainmen ega ding seconda y
educa ion seems o be sui able as a p oxy o o e all educa ion. P ima y educa ion shows less di e ences ac oss
coun ies, and e ia y educa ion is an inapp op ia e measu e o educa ion in he 19 h cen u y. The educa ion
a iable is aken om he Ba o–Lee da ase and measu es he a e age yea s o seconda y schooling wi hin he
popula ion be ween 15-64 yea s (Ba o & Lee, 2015). Thus, i measu es he s ock o educa ion wi hin a coun y.
Many p e ious s udies (e.g. Acemoglu e al., 2005) used his a iable, oo. The educa ion a iable is only a ailable
e e y 5 yea s. Since he o he wo a iables con ain yea ly da a, we linea ly app oxima e he educa ion a iable o
ob ain yea ly da a, oo.
We assume ha ins i u ions and educa ion a e no only de e minan s o o al ac o p oduc i i y bu (e en mo e
impo an ly) a p econdi ion o p i a e in es men . Thus, in he ollowing speci ica ions, we will only ocus on he
h ee abo emen ioned a iables and no include in es men . Howe e , his has been done by Rö hel and Leschke
(2023) who ollowed Dawson (1998) and in es iga ed he ac o p oduc i i y–e ec o ins i u ions and educa ion on
g ow h in a ull g ow h model, including he in es men a iable, o e he same pe iod.
Table 1p esen s he desc ip i e s a is ics. Since seconda y educa ion desc ibes he a e age yea s o seconda y
schooling, i akes a alue be ween 0 and 7.45 in he sample. Libe al democ acy a ies be ween 0.01 and 0.90, whe e
highe le els o democ acy a e associa ed wi h highe index alues. In he eg ession, we speci y all a iables in
loga i hmic o m. This is in line wi h p e ious s udies (e.g. Rö hel, 2023; Woold idge, 2020) and allows us o in e p e
TABLE 1 Desc ip i e s a is ics.
Obs. Mean S d. de . Min Max
GDP pe capi a 2747 12,162.71 10,779.54 1352.00 74,879.81
Libe al democ acy 2744 0.53 0.25 0.01 0.90
Seconda y educa ion 2760 1.53 1.63 0.00 7.45
248 RÖTHEL and LESCHKE
he esul s as elas ici ies. Since aking he na u al loga i hm would c ea e nega i e alues o he democ acy and
educa ion a iables, we i s apply he ollowing linea ans o ma ion on hese a iables: y¼1þx.
Figu e 1shows he a e age de elopmen o hese h ee a iables ac oss he coun ies in he sample o e ime.
S a ing in 1870 a a a he low le el, all h ee a iables exhibi con inuous and pa allel g ow h un il eaching hei
peak in 2007. The only excep ion a e he alues o libe al democ acy in he yea s a ound he Second Wo ld Wa
exhibi ing a decline be o e eco e ing again in he 1950s.
One disad an age o he leng h o ou da ase is he coun ies hemsel es. Especially, a he beginning o ou
ime ame, bo de s o some coun ies changed egula ly. We add ess his po en ial p oblem by excluding ce ain
coun ies om ou sample as pa o ou obus ness es s.
3.2 |Model
Tu ning o he eg ession equa ion and es ima ion s a egy, we assume ha lagged alues o he a iables a ec
hei cu en alues and he alues o he o he a iables. Thus, we look a a dynamic linea ela ionship. The e a e
se e al ways ha a e ypically used o es ima e dynamic panel speci ica ions. Usually, when he panel is sho wi h
many c oss-sec ional uni s, es ima o s like he wo-s age leas squa es o sys em GMM es ima o a e used.
FIGURE 1 Mean and s anda d de ia ion o he a iables (1870–2007).
RÖTHEL and LESCHKE 249
Howe e , hese es ima o s a e no necessa ily designed o o he se ings, o example, wi h bo h la ge N and T
(Roodman, 2009). Pesa an and Smi h (1995) desc ibe di e en possibili ies o dealing wi h such da a: One way
would be o apply an es ima o imposing a common slope, like he dynamic ixed e ec s (DFEs) es ima o . O he
possibili ies would be ei he a e aging he da a o e he c oss-sec ional uni s and es ima ing all oge he as a ime
se ies o a e aging he da a o e ime and es ima ing one single c oss-sec ion. Howe e , Pesa an and Smi h (1995)
show ha hese es ima o s mos ly p o ide inconsis en esul s in a dynamic en i onmen . Addi ionally, he DFE
es ima o su e s om he so-called Nickell bias (Nickell, 1981). Wi h la ge T, one could un single eg essions o
each uni and a e age he coe icien s ac oss g oups, yielding he o e all e ec . This s a egy is e e ed o as mean
g oup es ima o (MG) which is consis en a leas o la ge N and T. Fu he , when applying a pooled es ima o ( ixed
o andom e ec s), all pa ame e s ac oss uni s a e cons ained o be he same, excep o he in e cep s. In con as ,
wi h he MG es ima o , all pa ame e s a e es ima ed sepa a ely ac oss uni s. Thus, po en ial simila i ies a e neglec ed
(Pesa an e al., 1999). In o de o accoun o dynamic he e ogeneous models, Pesa an e al. (1999) in oduce he
PMG es ima o . This in e media e es ima o accoun s o bo h sho - un he e ogenei y ac oss uni s and long- un
simila i ies. Thus, i combines pooling and a e aging.
The da a desc ibed abo e ha e a mode a ely la ge N and la ge T. Thus, ei he he MG o he PMG es ima o can
be applied. Since all coun ies a e OECD coun ies, i is plausible o assume ha in he long- un, all a e a ec ed by
simila echnological and o he e ec s. Howe e , in he sho - un, hese coun ies p esumably di e in e ms o
hese ac o s and de ia e om he common long- un equilib ium. The e o e, we will assume ha in he sho - un,
he e ogenei y ac oss coun ies exis s, while in he long- un, simila i ies can be obse ed. Following he p e ious
explana ions in his sec ion, we will apply he PMG es ima o and compa e he esul s o he MG and he DFE es i-
ma o . Following Pesa an e al. (1999) and Blackbu ne and F ank (2007), a con enien s a ing poin is he ollowing
simple au o eg essi e dis ibu ed lag (ARDL) (p,q,q) amewo k:
yi ¼X
p
j¼1
λijyi, jþX
q
j¼0
δ1ijdi, jþX
q
j¼0
δ2ijsi, jþμiþϵi ,ð1Þ
whe e yi deno es GDP p.c, di libe al democ acy and si seconda y educa ion in coun y i a ime . μi ep esen s he
coun y ixed e ec s and ϵi he e o e m. Repa ame e izing, we ge he ollowing e o co ec ion equa ion:
Δyi ¼ϕiðyi, 1θ1idi θ2isi ÞþX
p1
j¼1
λ∗
ij Δyi, j
þX
q1
j¼0
δ∗
1ijΔdi, jþX
q1
j¼0
δ∗
2ijΔsi, jþμiþϵi ,
ð2Þ
whe e ϕi¼ 1P
p
j¼1
λi
!
,θ1i¼P
q
j¼0
δij
1P
k
λik
,θ2i¼P
q
j¼0
δij
1P
k
λik
,
λ∗
ij ¼ X
p
m¼jþ1
λim,j¼1,2,…,p1,
and
δ∗
1ij ¼ X
q
m¼jþ1
δim,δ∗
2ij ¼ X
q
m¼jþ1
δim,j¼1,2,…,q1:ð3Þ
ϕdeno es he e o co ec ion pa ame e and can be in e p e ed as he speed o adjus men o equilib ium. We
assume ha ou a iables a e nons a iona y (in eg a ed o o de one I(1)) and coin eg a ed. Thus, he i s pa o
250 RÖTHEL and LESCHKE
acknowledge he alue o a highe le el o ins i u ions and hus demand his kind o change in ins i u ions. In addi-
ion, he highe he le el o GDP p.c., and hus he le el o weal h o he people, he mo e ime can be used o
demand a highe le el o ins i u ions. Bo h a gumen s ollow Lipse (1959). Re u ning o Hypo heses 3and 4, we can
ejec bo h null hypo heses.
Simila o he abo e, we es he esul s when adding he same ime dummies. Table 8p esen s he esul s. The
e ec o educa ion on democ acy is posi i e and signi ican in all speci ica ions. Howe e , he signi icance le el
dec eases compa ed o he baseline esul s. In con as , he signi icance le el o GDP p.c. inc eases. The Wo ld Wa
II dummy emains signi ican and nega i e, and he Ma shall Plan dummy emains signi ican and posi i e.
Finally, we examine he ela ionship be ween GDP p.c., libe al democ acy and seconda y educa ion as he
dependen a iable. Again, we es he null hypo heses ha bo h a iables ha e no e ec on seconda y educa ion
agains he al e na i es 5 and 6 o mula ed abo e. Table 9p esen s he esul s. The con e gence coe icien is signi -
ican and nega i e in all speci ica ions, indica ing con e gence. Addi ionally, as in he p e ious cases, he Hausman
es a ou s he PMG es ima o . The long- un coe icien o GDP p.c. signi ican ly and posi i ely a ec s seconda y
educa ion, while he e ec o libe al democ acy is insigni ican . All sho - un coe icien s, excep o he lagged
TABLE 9 Baseline esul s seconda y educa ion.
(1) (2)
Hausman es
(3)
PMG MG DFE
Con e gence
coe icien
LogSeconda yEduca ion
1
0.0071*** 0.0116*** 0.0045***
(0.0017) (0.0024) (0.0007)
Long- un
coe icien s
LogLibe alDemoc acy
0.0075 2.7731 0.4360
(0.1313) (4.0061) (0.3540)
LogGDPp.c.
0.7616*** 0.7770 0.6810***
(0.0294) (0.6218) (0.0640)
0.80
Sho - un
coe icien s
ΔLogSeconda yEduca ion
1
0.8736*** 0.8474*** 0.8820***
(0.0103) (0.0106) (0.0094)
ΔLogLibe alDemoc acy
0.0087 0.0048 0.0023
(0.0063) (0.0072) (0.0049)
ΔLogGDPp.c.
0.0031 0.0054** 0.0014
(0.0025) (0.0026) (0.00260)
ΔLogGDPp.c.
1
0.0033 0.0046 0.0000
(0.0030) (0.0028) (0.0024)
Cons an 0.0420*** 0.0586*** 0.0235***
(0.0102) (0.0122) (0.0041)
Numbe o obse a ions 2686 2686 2686
Numbe o coun ies 20 20 20
Numbe o pe iods 136 136 136
No e: The Hausman es is indica ing ha he PMG es ima o is p e e ed o e he MG es ima o .
*Deno es signi icance a he 10% le el.
**Deno es signi icance a he 5% le el.
***Deno es signi icance a he 1% le el.
RÖTHEL and LESCHKE 257

dependen a iable, a e insigni ican . The sho - un esul s may no be su p ising since i akes ela i ely long un il a
shock owa ds he educa ion sys em ansla es in o a change in school a ainmen . In he long- un, GDP p.c. is
a coun y's po en ial weal h o in es in he educa ion sec o . Thus, wi h a highe le el o GDP, a coun y can in es
mo e in educa ion, which may lead o highe a ainmen a es in he u u e. The causal link be ween ins i u ions and
educa ion is no s aigh o wa d and emains insigni ican . Re u ning o Hypo heses 5and 6, we can ejec he null
TABLE 10 Baseline esul s seconda y educa ion wi h ime dummies.
(1) (2)
PMG PMG
Con e gence coe icien LogSeconda yEduca ion
0.0076*** 0.0079***
(0.0019) (0.0019)
Long- un coe icien s LogGDPp.c.
0.7450*** 0.7499***
(0.0256) (0.0240)
LogLibe alDemoc acy
0.1595 0.0748
(0.1245) (0.1076)
Wo ld Wa I (1914–1918) 0.0220 0.0110
(0.0580) (0.0506)
Wo ld Wa II (1939–1945) 0.2483*** 0.2094***
(0.0622) (0.0541)
G ea Dep ession (1929–1930) 0.0896
(0.0770)
Ma shall Plan (1946–1952) 0.1095**
(0.0487)
Oil C isis (1973–1974) 0.0124
(0.0781)
Fall o he Be lin Wall (1989–1990) 0.0910
(0.0563)
Sho - un coe icien s ΔLogSeconda yEduca ion
1
0.8706*** 0.8694***
(0.0109) (0.0112)
ΔLogLibe alDemoc acy
0.0094 0.0089
(0.0062) (0.0063)
ΔLogGDPp.c.
0.0022 0.0031
(0.0025) (0.0025)
ΔLogGDPp.c.
1
0.0023 0.0030
(0.0028) (0.0029)
Cons an 0.0450*** 0.0466***
(0.0118) (0.0118)
Numbe o obse a ions 2686 2686
Numbe o coun ies 20 20
Numbe o pe iods 136 136
No e: All speci ica ions we e es ima ed using he PMG es ima o .
*Deno es signi icance a he 10% le el.
**Deno es signi icance a he 5% le el.
***Deno es signi icance a he 1% le el.
258 RÖTHEL and LESCHKE
hypo hesis o he o me while we canno ejec he null hypo hesis o he la e . Table 10 p esen s he esul s,
including ime dummies. The esul s emain almos exac ly he same and con i m hose o he baseline es ima ion.
GDP p.c. emains signi ican and posi i e, while libe al democ acy is insigni ican . Ne e heless, we ea ou esul s
as p elimina y and will u he es hem in he ollowing ex ensions sec ion.
5|EXTENSIONS
Se e al ac o s migh in luence ou esul s ha we ha e ye o con ol o . Fi s , we will ocus on he es ima ion wi h
GDP p.c. as he dependen a iable. The coe icien s, o example, migh be d i en by single coun ies only. Thus, we
ha e again es ima ed he PMG model o Equa ion (2), es ing o he sensi i i y o a educ ion o coun ies. Figu e 2
displays he poin es ima e and he 95%-con idence in e al o libe al democ acy when each o he coun ies is
emo ed om he sample. The poin es ima e o he le shows he main ‘baseline es ima ion’ om Table 5. The sec-
ond poin es ima e displays he esul s when Aus ia (AUS) is emo ed om he sample. This p ocedu e is epea ed
un il e e y coun y has been emo ed om he equa ion once. Simila ly, Figu e 3shows he sensi i i y o he educa-
ion a iable when single coun ies a e emo ed. Again, on he le , ‘MAIN’deno es he baseline es ima ion ollowed
by he o he coun ies. All poin es ima es emain posi i e and signi ican , con i ming he esul s abo e. Howe e ,
he magni ude o he coe icien s sligh ly di e s when emo ing single coun ies.
Figu es 4and 5show he baseline esul s o bo h a iables om Equa ion (4) when coun ies a e excluded om
he sample. The long- un coe icien o GDP p.c. emains signi ican and posi i e, con i ming he p e ious esul s. In
con as , he coe icien o seconda y educa ion is insigni ican o mos o he speci ica ions. Table 7p o ides a sig-
ni ican coe icien on he 10%-le el. Howe e , e en on he 10%-le el, ou esul s a e no obus o he exclusion o
single coun ies.
Figu es 6and 7show he esul s o he exclusion o coun ies wi h espec o Equa ion (5), whe e seconda y
educa ion is he dependen a iable. Simila o he p e ious esul s, he long- un coe icien o GDP p.c. is signi ican ,
FIGURE 2 Baseline GDP es ima ion: Sensi i i y o libe al democ acy coe icien .
RÖTHEL and LESCHKE 259
while he libe al democ acy coe icien is insigni ican . O e all, mos esul s a e obus o emo ing single coun ies
om he sample. Only seconda y educa ion does no seem o a ec libe al democ acy in he long- un.
Fu he , one could a gue ha he lag s uc u e a ec s he PMG esul s. E en hough he lag s uc u e has been
chosen based on he AIC, we will also look a he esul s o he ARDL(1,1,1), ARDL(2,1,1), ARDL(1,2,1), ARDL(1,1,2),
FIGURE 3 Baseline GDP es ima ion: Sensi i i y o seconda y educa ion coe icien .
FIGURE 4 Baseline libe al democ acy es ima ion: Sensi i i y o GDP p.c. coe icien . [Colou igu e can be
iewed a wileyonlinelib a y.com]
260 RÖTHEL and LESCHKE
ARDL(1,2,2), ARDL(2,2,1) and ARDL(2,1,2). Fo GDP p.c. as he dependen a iable, hey a e p esen ed in Table 11.
Gene ally, he coe icien s all emain s able h oughou he di e en speci ica ions and con i m he p e iously ound
esul s. Bo h, libe al democ acy and seconda y educa ion, emain posi i e, signi ican and o oughly he same
FIGURE 5 Baseline libe al democ acy es ima ion: Sensi i i y o seconda y educa ion coe icien . [Colou igu e
can be iewed a wileyonlinelib a y.com]
FIGURE 6 Baseline seconda y educa ion es ima ion: Sensi i i y o libe al democ acy coe icien . [Colou igu e
can be iewed a wileyonlinelib a y.com]
RÖTHEL and LESCHKE 261
magni ude as in p e ious speci ica ions. Table 12 shows he same o libe al democ acy as he dependen
a iable. Again, he esul s o he con e gence coe icien and GDP p.c. emain he same and s able. Howe e , he
long- un seconda y educa ion coe icien is ei he insigni ican o posi i e and signi ican . This con i ms he
esul s om he p e ious exclusion o coun ies. Las ly, we also es he sensi i i y o he lag s uc u e when
seconda y educa ion is he dependen a iable. The e, he s uc u e o Table 13 di e s om hose p esen ed be o e.
Since he model wi h only one lag o he dependen a iable is no sol able, we ha e es ed he sensi i i y o a
di e en lag s uc u e only o he ARDL(2,1,1), ARDL(2,2,1) and ARDL(2,2,2). Remembe ha he AIC indica ed ha
he ARDL(2,1,2) is he bes choice in he baseline es ima ion. Again, he esul s emain simila o hose o he
p e ious es ima ions. GDP p.c. posi i ely and signi ican ly a ec s seconda y educa ion, while libe al democ acy is
insigni ican .
E en ually, one could a gue ha , ins ead o using yea ly da a, i would be mo e app op ia e o use 5-yea
a e ages. This migh be ad an ageous since ins i u ions and educa ion usually only change slowly o e ime. When
using a e ages, we also a oid pu ing oo much emphasis on small changes. Finally, as explained abo e, he
educa ion da a a e only a ailable e e y 5 yea s. Table 14 p esen s he PMG baseline esul s, including ime dummies
o all h ee dependen a iables. In column one, we ind simila posi i e and signi ican coe icien s o he long- un
es ima es o libe al democ acy and seconda y educa ion. In columns 2 and 3, we also see a signi ican and posi i e
e ec o GDP p.c. on ins i u ions and educa ion, espec i ely. Addi ionally, libe al democ acy does no seem o a ec
educa ion in he long- un. Again, hese esul s con i m he p e ious indings. Simila o he abo e, he long- un
educa ion coe icien alls ou o line, when libe al democ acy is he dependen a iable. In column 2, we ind a
signi ican and nega i e e ec on ins i u ions. This esul con i ms he non obus ness o he e ec o his a iable on
ins i u ions.
We will now e u n o he hypo heses p oposed in Sec ion 2. The null Hypo heses 1and 2, ha ins i u ions and
human capi al ha e no e ec on economic g ow h, can bo h be ejec ed. The null Hypo heses 3and 5, ha economic
g ow h has no e ec on ins i u ions and human capi al, espec i ely, can also be ejec ed. Mo eo e , ou esul s
indica e ha we canno ejec he null Hypo hesis 4 ha human capi al does no a ec ins i u ions. The same holds
o null Hypo hesis 6 ha ins i u ions ha e no e ec on human capi al.
FIGURE 7 Baseline seconda y educa ion es ima ion: Sensi i i y o GDP p.c. coe icien .
262 RÖTHEL and LESCHKE

TABLE 11 Sensi i i y o lag s uc u e GDP p.c.
(1) (2) (3) (4) (5) (6) (7)
ARDL(1,1,1) ARDL(2,1,1) ARDL(1,2,1) ARDL(1,1,2) ARDL(1,2,2) ARDL(2,2,1) ARDL(2,1,2)
Con e gence coe icien LogGDPp.c.
1
-0.0232*** -0.0243*** -0.0218*** -0.0241*** -0.0224*** -0.0194*** -0.0249***
(0.0041) (0.0044) (0.0038) (0.0043) (0.0039) (0.0037) (0.0045)
Long- un coe icien s LogLibe alDemoc acy
3.3757*** 2.9176*** 3.2492*** 3.2674*** 3.1963*** 2.5970*** 2.8652***
(0.4927) (0.4422) (0.5268) (0.4687) (0.5083) (0.5246) (0.4294)
LogSeconda yEduca ion
0.8067*** 0.8663*** 0.8451*** 0.8237*** 0.8613*** 1.0166*** 0.8837***
(0.1003) (0.0928) (0.1065) (0.0972) (0.1040) (0.1144) (0.0909)
Cons an 0.1836*** 0.1934*** 0.1736*** 0.1907*** 0.1785*** 0.1579*** 0.1980***
(0.0295) (0.0325) (0.0277) (0.0310) (0.0286) (0.0274) (0.0335)
Time dummies YES YES YES YES YES YES YES
Numbe o obse a ions 2705 2686 2682 2687 2682 2681 2686
Numbe o coun ies 20 20 20 20 20 20 20
Numbe o pe iods 137 136 136 136 136 136 136
No es: All speci ica ions we e es ima ed using he PMG es ima o . *, ** and *** deno e signi icance a he 10%-, 5%- and 1%-le el, espec i ely.
RÖTHEL and LESCHKE 263
TABLE 12 Sensi i i y o lag s uc u e libe al democ acy.
(1) (2) (3) (4) (5) (6) (7)
ARDL(1,1,1) ARDL(2,1,1) ARDL(1,2,1) ARDL(1,1,2) ARDL(1,2,2) ARDL(2,2,1) ARDL(2,1,2)
Con e gence coe icien LogLibe alDemoc acy
1
0.0745*** 0.0845*** 0.0758*** 0.0753*** 0.0765*** 0.0853*** 0.0848***
(0.0251) (0.0258) (0.0253) (0.0252) (0.0254) (0.0255) (0.0258)
Long- un coe icien s LogGDPp.c.
0.0767*** 0.0865*** 0.0604*** 0.0780*** 0.0652*** 0.0764*** 0.0873***
(0.0204) (0.0185) (0.0207) (0.0205) (0.0206) (0.0192) (0.0186)
LogSeconda yEduca ion
0.0383 0.0334 0.0589** 0.0374 0.0538** 0.0471** 0.0339
(0.0249) (0.0226) (0.0251) (0.0252) (0.0251) (0.0233) (0.0229)
Cons an 0.0202*** 0.0312*** 0.0105*** 0.0215*** 0.0138*** 0.0244*** 0.0321***
(0.0069) (0.0092) (0.0040) (0.0072) (0.0049) (0.0072) (0.0094)
Time dummies Yes Yes Yes Yes Yes Yes Yes
Numbe o obse a ions 2705 2682 2686 2687 2686 2681 2682
Numbe o coun ies 20 20 20 20 20 20 20
Numbe o pe iods 137 136 136 136 136 136 136
No e: All speci ica ions we e es ima ed using he PMG es ima o .
*Deno es signi icance a he 10% le el.
**Deno es signi icance a he 5% le el.
***Deno es signi icance a he 1% le el.
264 RÖTHEL and LESCHKE
6|CONCLUSION
A la ge sha e o he li e a u e and empi ical analyses suppo he idea ha bo h ins i u ions and human capi al
(educa ion) ha e a signi ican impac on GDP p.c. (g ow h). Ou empi ical analysis is in line wi h hese indings.
Based on he long ime ho izon, om 1870 un il 2007, we a e able o show he high impac o ins i u ions and
educa ion on GDP p.c. Bu we do no s op a his poin . Some s ands o li e a u e also p o ide ou
addi ional hypo heses: (1) GDP p.c. should posi i ely in luence ins i u ions. (2) Also, human capi al migh a ec
ins i u ion building in a posi i e manne . (3) GDP p.c. is claimed o ha e a posi i e in luence on human capi al
because he iche a coun y is he mo e money can be spen o imp o e he educa ion sys em. (4) Fu he , be e
ins i u ions should suppo he educa ion sys em, oo. In o de o es hese hypo heses, we es ima e a modi ied
ARDL model, he e o co ec ion model, using he PMG es ima o . Human capi al is measu ed by he a e age
yea s o seconda y schooling be ween 15 and 64 yea s and economic g ow h by GDP p.c., and as a p oxy o
ins i u ions, we use he libe al democ acy index, which indica es ins i u ional cons ain s agains a bi a y policy
ac ion o he s a e.
The model and he da a seem o be sui able o analysing long- un equilib ium ela ionships be ween he
a iables GDP p.c., ins i u ions, and human capi al. As men ioned abo e, ou esul s con i m he hypo hesis ha
ins i u ions and educa ion posi i ely a ec he GDP p.c.; his e ec is s onge o ins i u ions compa ed o
educa ion. Fu he mo e, we ind ha in he long- un, GDP p.c. has a posi i e e ec on ins i u ions. Howe e ,
he same is no ue o he ela ionship be ween educa ion and ins i u ions. In addi ion, ou esul s show ha GDP
p.c. signi ican ly and posi i ely a ec s seconda y educa ion. Again, his is no he case o he e ec o ins i u ions
on educa ion. These indings hold i we ex end he basic model by including ime dummies, i we exclude single
coun ies om he sample and i we use di e en lag s uc u es o a e age he da a.
This pape has some limi a ions. Due o he panel s uc u e o ou da ase , he esul s a e agg ega ed ac oss
coun ies. Thus, i is no possible o con incingly assign he coe icien s o speci ic coun ies. Addi ionally, we y o
app oach causali y by using a ec o e o co ec ion model. Ou esul s can be in e p e ed as being g ange causal
TABLE 13 Sensi i i y o lag s uc u e seconda y educa ion.
(1) (2) (3)
ARDL(2,1,1) ARDL(2,2,1) ARDL(2,2,2)
Con e gence coe icien LogSeconda yEduca ion
1
0.0075*** 0.0076*** 0.0078***
(0.0019) (0.0019) (0.0019)
Long- un coe icien s LogLibe alDemoc acy
0.1652 0.1219 0.1014
(0.1296) (0.1298) (0.1214)
LogGDPp.c.
0.7406*** 0.7449*** 0.7505***
(0.0264) (0.0266) (0.0254)
Cons an 0.0439*** 0.0447*** 0.0458***
(0.0112) (0.0115) (0.0118)
Time dummies Yes Yes Yes
Numbe o obse a ions 2687 2682 2681
Numbe o coun ies 20 20 20
Numbe o pe iods 136 136 136
No e: All speci ica ions we e es ima ed using he PMG es ima o .
*Deno es signi icance a he 10% le el.
**Deno es signi icance a he 5% le el.
***Deno es signi icance a he 1% le el.
RÖTHEL and LESCHKE 265
(see, e.g., Yoo & Ku, 2009). This can be seen as an app oach o a causal in e p e a ion; howe e , one should ne e be
ho oughly con inced. Ano he gene al empi ical ma e is he limi ed da a a ailabili y o e such a long ime.
F om a policy pe spec i e, se e al conclusions can be d awn. As p e ious esea ch indica es, human capi al and
ins i u ions a e impo an d i e s o economic g ow h in he long- un. Thus, policymake s should in es in a coun y's
human capi al s ock and ad oca e secu ed p ope y igh s. Howe e , his poses some d awbacks. The e ec is o a
long- un na u e, and poli icians usually p o ide sho - un o ien ed policies in o de o ge e-elec ed. The e o e, such
TABLE 14 5-yea a e age es ima ion.
(1) (2) (3)
GDP Ins i u ions Educa ion
Con e gence coe icien s 0.1081*** 0.1966*** 0.1275***
(0.0267) (0.0499) (0.0314)
Long- un coe icien s LogLibe alDemoc acy
1.5763*** 0.1006
(0.2935) (0.1097)
LogSeconda yEduca ion
1.2454*** 0.2837***
(0.0759) (0.0427)
LogGDPp.c.
0.2950*** 0.7330***
(0.0286) (0.0207)
WWI 1.1019*** 0.1006*** 0.0127
(0.2024) (0.0264) (0.0423)
WWII 0.6326*** 0.4422*** 0.1813***
(0.1804) (0.0460) (0.0627)
Sho - un coe icien s ΔLogGDPp.c.
0.0454 0.0122
(0.0704) (0.0235)
ΔLogGDPp.c.
1
0.1355*** 0.0930** 0.0417
(0.0474) (0.0373) (0.0304)
ΔLogLibe alDemoc acy
0.5494** 0.0632
(0.2712) (0.0667)
ΔLogLibe alDemoc acy
1
0.3140* 0.1889***
(0.1680) (0.0712)
ΔLogSeconda yEduca ion
0.2750** 0.0759
(0.1131) (0.0649)
ΔLogSeconda yEduca ion
1
0.2092 0.2069*** 0.6918***
(0.1477) (0.0779) (0.0410)
Cons an 0.8753*** 0.3973*** 0.7449***
(0.1978) (0.1009) (0.1855)
Numbe o obse a ions 514 514 515
Numbe o coun ies 20 20 20
Numbe o pe iods 26 26 26
No e: All speci ica ions we e es ima ed using he PMG es ima o .
*Deno es signi icance a he 10% le el.
**Deno es signi icance a he 5% le el.
***Deno es signi icance a he 1% le el.
266 RÖTHEL and LESCHKE