Boua ou , A e ; Gha bi, Sa a; Kalai, Maha; Helali, Kamel
A icle
Rela ionships be ween g een echnological inno a ion,
enewable ene gy, ci cula economy, and g een g ow h
Jou nal o Inno a ion & Knowledge (JIK)
P o ided in Coope a ion wi h:
Else ie
Sugges ed Ci a ion: Boua ou , A e ; Gha bi, Sa a; Kalai, Maha; Helali, Kamel (2025) : Rela ionships
be ween g een echnological inno a ion, enewable ene gy, ci cula economy, and g een g ow h,
Jou nal o Inno a ion & Knowledge (JIK), ISSN 2444-569X, Else ie , Ams e dam, Vol. 10, Iss. 4, pp.
1-13,
h ps://doi.o g/10.1016/j.jik.2025.100748
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/327643
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
Rela ionships be ween g een echnological inno a ion, enewable ene gy,
ci cula economy, and g een g ow h
A e BOUATTOUR
a
, Sa a GHARBI
b
, Maha KALAI
c
, Kamel HELALI
d,*
a
Assis an P o esso in Economics. Highe Ins i u e o Business Adminis a ion o S ax, Uni e si y o S ax. Economics and Managemen Resea ch Labo a o y. Ai po
Road km 4, B.P. 1013-3018, S ax, Tunisia
b
Ph.D. in Economics. Facul y o Economics and Managemen o S ax, Uni e si y o S ax. Resea ch Labo a o y in Compe i i eness, Comme cial Decisions and
In e na ionaliza ion. Ai po Road km 4.5, B.P. 1088-3018, S ax, Tunisia
c
Assis an P o esso in Quan i a i e Me hods. Facul y o Economics and Managemen o S ax, Uni e si y o S ax. Resea ch Labo a o y in Compe i i eness, Comme cial
Decisions and In e na ionaliza ion. Ai po Road km 4.5, B.P. 1088-3018, S ax, Tunisia
d
Full P o esso in Quan i a i e Me hods. Depa men o Applied Quan i a i e Me hods, Facul y o Economics and Managemen o S ax, Uni e si y o S ax. Resea ch
Labo a o y in Compe i i eness, Comme cial Decisions and In e na ionaliza ion. Ai po Road km 4.5, B.P. 1088-3018, S ax, Tunisia
ARTICLE INFO
JEL Classi ica ion:
G01
G21
M14
Q50
Keywo ds:
G een g ow h
G een echnological inno a ion
Renewable ene gy
Ci cula economy
27 Eu opean Union coun ies
ABSTRACT
This s udy employs he panel smoo h h eshold au o eg ession model o da a spanning 27 Eu opean Union (EU)
membe s a es om 2000 o 2022 o in es iga e he nonlinea ela ionships be ween g een g ow h and g een
echnological inno a ion. This app oach enables ou examina ion o he dynamics o economies’ ise om
weakness o s eng h. This in es iga ion in oduces a comp ehensi e amewo k ha combines g een echno-
logical inno a ion, enewable ene gy sou ces, and he ci cula economy in o a single g een g ow h model. The
esul s con i m wo op imal h esholds ha impac how g een echnological inno a ion a ec s g een g ow h,
dis inguishing be ween he h ee egimes. Addi ionally, he esul s demons a e he posi i e in luence o g een
echnological inno a ion on g een g ow h. In con as , he s udy e eals a weake e ec wi h inc easing le els o
g een echnological inno a ion ansi ions om in e media e (Regime 2) o high (Regime 3) g een g ow h.
Fu he mo e, insu icien g een echnological inno a ion has no de eloped e ec i e enewable ene gy ech-
nologies ha can os e g een g ow h. Con e sely, enewable ene gy begins o posi i ely a ec g een g ow h
when EU coun ies inc ease g een echnological inno a ion in es men . Finally, a obus egime o g een ech-
nological inno a ion acili a es ci cula i y and p omo es EU economies’ g een g ow h, indica ing ha he in-
luence o he ci cula economy on g een g ow h depends on inc eased g een echnological inno a ion.
P ac ically, policymake s in EU economies should ocus on sus ainable inno a ion o p omo e g een g ow h and
ad ance he Uni ed Na ions Sus ainable De elopmen Goals.
In oduc ion
Inc eased clima e change awa eness has inc eased he ocus on
ansi ioning owa d a g een economy (Ali e al., 2021) ha ea u es low
ca bon emissions, e icien esou ce use, and social inclusi eness. A
g een economy is in ended o p o ec human heal h by de eloping
cleane , mo e cos -e ec i e, and en i onmen ally iendly echnologies
and main aining ecological s anda ds ha a e a global conce n. G een
echnological inno a ion con ibu es o add essing hese s anda ds.
While mos na ions, including hose in he Eu opean Union (EU), a e
commi ed o igh ing clima e change, he ze o ca bon e a emains
elusi e (Ahmad & Wu, 2022).
In line wi h hei commi men s o en i onmen al p ese a ion,
coun ies ha e implemen ed measu es o educe ca bon emissions and
achie e g een g ow h. This has caused a no able inc ease in he numbe
o g een echnological inno a ions in he pas ew decades (Su & Moa-
niba, 2017). The ongoing discou se su ounding po en ial g een pa h-
ways o economic ad ancemen has p omp ed nume ous economis s o
ocus on g een echnological inno a ion as a s a egy o en i onmen al
s ewa dship. The concep o g een echnological inno a ion emphasizes
* Co esponding au ho a : Full P o esso in Quan i a i e Me hods. Depa men o Applied Quan i a i e Me hods, Facul y o Economics and Managemen o S ax.
Uni e si y o S ax. Ai ield Road Km 4, BP 1088-3018, Rack no: 188. Resea ch Labo a o y in Compe i i eness, Comme cial Decisions and In e na ionaliza ion.
Ai po Road km 4.5, B.P. 1088-3018, S ax, Tunisia.
E-mail add ess: [email p o ec ed] (K. HELALI).
Con en s lis s a ailable a ScienceDi ec
Jou nal o Inno a ion & Knowledge
jou nal homepage: www.else ie .com/loca e/jik
h ps://doi.o g/10.1016/j.jik.2025.100748
Recei ed 12 Oc obe 2024; Accep ed 29 May 2025
Jou nal o Inno a ion & Knowledge 10 (2025) 100748
A ailable online 5 June 2025
2444-569X/© 2025 The Au ho s. Published by Else ie España, S.L.U. on behal o Jou nal o Inno a ion & Knowledge. This is an open access a icle unde he CC
BY-NC-ND license (
h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/ ).
he o ien a ion o inno a ion o p ese e he en i onmen , in con as o
Schumpe e ’s de ini ion, which conside s inno a ion o be a gene al
p ocess ha encompasses any o m o modi ica ion ha dis up s o
imp o es economic sys ems wi hou any pa icula dis inc ion con-
ce ning i s aims o impac s (Sica, 2016).
G een echnological inno a ion and g ow h a e he p ima y d i e s
o he global g een ansi ion. This shi is an aspec o he economic
dynamics elucida ed by Aus ian economis Schumpe e , who delin-
ea ed he undamen al economic d i e s ha cha ac e ize he change
p ocess as “c ea i e des uc ion” (Schumpe e , 1942). Al hough p e i-
ous economic li e a u e has ecognized he impo ance o g een ech-
nological inno a ion o en i onmen al p o ec ion, he e m g een
echnology emains ague and ambiguous. The li e a u e has no p o-
ided a clea and dis inc de ini ion o g een inno a ion, e e ing o i
by di e en names such as sus ainable inno a ion (Fe nandes e al.,
2021), en i onmen al inno a ion (Luan e al., 2024), g een echnological
inno a ion (Li e al., 2022), and cleane inno a ions (Deng e al., 2024).
Simila ly, he concep o g een g ow h lacks a clea , uni e sal de i-
ni ion. The concep o a g een economy wi h sus ainable economic
g ow h eme ged as a cen al heme in he inal Uni ed Na ions documen
a he Rio Con e ence on Sus ainable De elopmen in 2012 (Hickel &
Kallis, 2020). In heo y, g een g ow h e e s o a o m o economic
g ow h ha endea o s o s ike a balance be ween na u al esou ce use
and economic expansion and p omo es social equi y and en i onmen al
p o ec ion. Consequen ly, when mapping ou a pa h owa d g een
g ow h, i is essen ial o ecognize he pi o al in luence o enewable
ene gy, non enewable ene gy, and g een echnological inno a ion. A
e iew o he li e a u e on g een g ow h e eals a consensus among
scien is s ega ding i s abili y o p ese e he en i onmen while
achie ing economic bene i s. Resea che s ha e ocused on de ining,
measu ing, and examining he ela ionship o economic g ow h wi h he
en i onmen and de e mining how g een echnological inno a ion e-
duces ca bon emissions o de ine g een g ow h (Belmon e-U e˜
na e al.,
2021).
G een echnological inno a ion is a undamen al elemen o ecoin-
no a ion and a p omising app oach o acili a ing sus ainable de el-
opmen . This p ocess is cha ac e ized by he coexis ence o g een and
inno a i e a ibu es ha con ibu e o p omo ing a mo e sus ainable
u u e (Chen e al., 2023). Indeed, g een echnological inno a ion con-
ibu es o en i onmen al quali y and can be employed o e ec i e
clima e ac ion (Cagla e al., 2024). Mo eo e , acco ding o Sha i e al.
(2020), g een inno a ion educes equipmen loss, p oduc ion cos s, and
uel was e, imp o ing p oduc ion e iciency and enhancing g een
g ow hCliquez ou appuyez ici pou en e du ex e.. The deploymen o
g een echnological inno a ion bene i s he p og ession o ecycling
ini ia i es and he epu posing o p oduc ion esiduals (Zhao e al.,
2021) while also s imula ing he ad anced de elopmen o enewable
ene gy (Esmaeilpou Moghadam & Ka ami, 2024). G een echnological
inno a ion expedi es he ansi ion o ci cula i y, minimizing economic
ha m and dispa i y while also sa egua ding he en i onmen (Suchek
e al., 2021; Chauhan e al., 2022; Khan & Khu shid, 2022;
S´
anchez-Ga cía e al., 2024).
Al hough he ela ionship be ween g een echnological inno a ion
and g een g ow h has been ex ensi ely examined, some conce ns emain
unse led. P e ious esea ch on his opic has e ealed ha he e is s ill
much o be unde s ood abou he nuances o his ela ionship wi h a
endency o adop linea amewo ks o analysis (Chen & Lee, 2020;
Mei un e al., 2021). Ne e heless, a pauci y o li e a u e examining his
ela ionship om a nonlinea pe spec i e emains (Li e al., 2021;
Zhang, 2021; Liu e al., 2022; Zeng e al., 2022; Zhang e al., 2023).
Complemen ing p e ious esea ch, we employ he panel smoo h
h eshold au o eg ession (PSTAR) model o panel da a spanning 27 EU
membe s a es om 2000 o 2022 o in es iga e he nonlinea ela-
ionship be ween g een g ow h and g een echnological inno a ion. This
app oach was mo i a ed he unde s anding ha he EU has se he goal
o spea heading he global ansi ion owa d a g een economy. Gi en
he challenges al eady ou lined, he EU has es ablished he goal o
hal ing g eenhouse gas (GHG) emissions by 2030 (Ci uen es-Fau a,
2022; P esno e al., 2021). In addi ion, he 2019 G een Deal o Eu ope,
assessing he impac o clima e legisla ion, in oduced new legisla ion o
ad ance he ci cula economy and g een inno a ion. Fu he mo e, he
Eu opean Commission’s ini ia i e o Ho izon Eu ope, he EU F ame-
wo k P og am o Resea ch and Inno a ion (R&I) 2021–2027, also
highligh s his ambi ion (Kas inos & Webe , 2020). A he same ime,
echnological inno a ions ha e enabled EU coun ies o join he anks o
o he na ions pu suing g een g ow h (Nosheen e al., 2021). The selec-
ion o EU coun ies o his s udy is judicious, gi en hei commi men
o ad ancing g een g ow h h ough g een echnological inno a ion,
enabling ou examina ion o he nonlinea ela ionship be ween g een
echnological inno a ion and g een g ow h. Fu he mo e, his s udy
employs a comp ehensi e amewo k ha uni ies g een echnological
inno a ion, enewable ene gies, and he ci cula economy in o a single
g een g ow h model.
This s udy con ibu es o he exis ing body o li e a u e in se e al
ways. Fi s , p e ious s udies ha e p ima ily concen a ed on how g een
echnology impac s inno a ion and economic g ow h (Mei un e al.,
2021). These s udies ha e a gued ha g een echnological inno a ion is
pi o al o enhancing en i onmen al quali y by imp o ing esou ce e -
iciency and lowe ing ca bon emissions (Sohag e al., 2015). Howe e ,
limi ed esea ch has explo ed he impac o g een echnological inno-
a ion on g een g ow h (Saqib e al., 2024). Second, he majo i y o
s udies in his ield ha e no conside ed he nonlinea ajec o y o his
ela ionship (Wang, 2023). Thi d, while a e iew o ex an li e a u e
e eals ha g een echnological inno a ion, enewable ene gy, and he
ci cula p oduc ion sys ems o en in luence en i onmen ally esponsible
g ow h, o he bes o ou knowledge, no s udy has conside ed hem in a
uni ied amewo k. This s udy examines hese ac o s o add ess his
gap, assuming a nonlinea ela ionship be ween g een echnological
inno a ion and en i onmen ally esponsible g ow h, making a no e-
wo hy con ibu ion by employing an econome ic app oach ha in-
co po a es h eshold e ec s. Fo his pu pose, he PSTAR model
analyzes whe he he nonlinea impac on g een g ow h has an iden i-
iable h eshold e ec ha he e ogeneously in luences he impac o
enewable ene gy and ci cula i y on g een g ow h a a ying le els o
g een echnological inno a ion. The indings can enable EU policy-
make s o unde s and he signi icance o he e ec o g een echnolog-
ical inno a ion on g een g ow h and he ela ionship be ween his
g ow h, enewable ene gy, and he ci cula economy. A mo e p o ound
unde s anding can enable hem o de elop and implemen e ec i e
s a egies.
The emainde o his s udy is o ganized in he ollowing manne .
Sec ion 2 p esen s a li e a u e e iew and subsequen esea ch hypo h-
eses. Sec ion 3 de ails ou empi ical model and da a. Sec ion 4 conduc s
he PSTAR es . Sec ion 5 discusses he main esul s o he PSTAR es
es ima ion. Finally, Sec ion 6 d aws conclusions and ou lines ela ed
policy ecommenda ions.
Li e a u e e iew
Impac o g een echnological inno a ion on g een g ow h
The pi o al in luence o echnological inno a ion on enhancing
en i onmen ally esponsible g ow h has been widely acknowledged.
Howe e , he inc eased u gency o en i onmen al conce ns and
ad ancing clima e change has p omp ed esea che s o p io i ize
in es iga ing he in e connec ions be ween sus ainable inno a ion and
sus ainable g ow h. New endogenous g ow h models in eg a ing en i-
onmen al ac o s and inno a ion (Acemoglu e al., 2016) ha e inspi ed
se e al applied s udies examining hese in e connec ions (Ma hews,
2020; Nosheen e al., 2021; Li e al., 2022; Wei e al., 2023) and
emphasized he c ucial in luence o echnological inno a ion on
imp o ing en i onmen al quali y by op imizing ene gy use and
A. BOUATTOUR e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100748
2
educing ca bon emissions o acili a e g een g ow h.
In a ecen s udy, he au o eg essi e dis ibu ed lag (ARDL) ech-
nique employed by Mei un e al. (2021) examined he ela ionship be-
ween he a iables unde conside a ion o Singapo e om 1990 o
2018, e ealing ha sus ainable echnological inno a ion posi i ely
a ec s g een g ow h and educes long- and sho - e m ca bon emissions.
A ecen s udy by Khan e al. (2023) on 26 EU coun ies be ween 2000
and 2020 demons a ed ha g een echnological inno a ion ad ances
g een g ow h. In addi ion, he esea che s pos ula ed ha educed ca -
bon emissions compensa e o he a o able in luence o con en ional
g ow h on hese emissions. The same indings we e epo ed in s udies o
o he coun ies (Chen & Lee, 2020; Li e al., 2022; Sun e al., 2023).
Echoing hese obse a ions, a ecen s udy by Saqib e al. (2024) on he
impac o g een echnological inno a ion on g een g ow h in leading
coun ies wi h he mos signi ican ecological implica ions om 1990 o
2019 con i med he posi i e e ec . Fu he mo e, he esul s indica ed
imp o emen in en i onmen al quali y ela ed o g een echnological
inno a ion and g een g ow h.
In a simila ein, Fe nandes e al. (2021) examined he impac o
sus ainable echnology ans e and sus ainable inno a ion on g een
g ow h in 32 membe coun ies o he O ganisa ion o Economic
Co-ope a ion and De elopmen (OECD) be ween 1990 and 2013. The
use o wo-s age gene alized me hod o momen s (GMM) es ima o s,
comp ising one model o ep esen he lag on he dependen a iable
and ano he o he wo g een inno a ion a iables o sus ainable
echnology ans e and sus ainable pa en s demons a ed ha hese
a iables a o g een g ow h. The esul s con i med ha a ansi ion
owa d inno a ion is compa ible wi h coun ies’ g een g ow h
aspi a ions.
In con as , Nosheen e al. (2021) showed ha he e ec o g een
echnological inno a ion in he EU is condi ioned by he speci ic inno-
a ion cha ac e is ics examined. The au ho s’ in es iga ion o he e ec
o a ious sus ainable echnological inno a ions on en i onmen ally
esponsible g ow h in 27 EU economies be ween 2000 and 2017
e ealed ha ene gy- ela ed g een echnological inno a ion has a pos-
i i e e ec , and g een echnologies ela ed o anspo a ion and p o-
duc ion ha e a nega i e impac . These ad e se e ec s we e u he
co obo a ed by he indings o se e al o he s udies (Wang e al., 2014;
Zhao e al., 2017; Mensah e al., 2018).
F amewo k and esea ch hypo heses
An in es iga ion o p e ious esea ch e eals ha g een echnolog-
ical inno a ion and eco iendly g ow h a e he p ima y d i e s o he
global ecological ansi ion (Esmaeilpou Moghadam & Ka ami, 2024).
Mo eo e , while he majo i y o he ex an li e a u e on he subjec has
employed linea amewo ks o examine he nexus be ween g een
inno a ion and g een g ow h (Mei un e al., 2021), ew s udies ha e
explo ed his ela ionship using a nonlinea amewo k (Liu e al., 2022;
Zhang e al., 2023; Wang, 2023; Luan e al., 2024).
Fo example, Zeng e al. (2022) employed h eshold panel models o
explo e his ela ionship in 30 Chinese p o inces be ween 2001 and
2019, e ealing he nonlinea e ec o g een echnological inno a ion
ha depends on egional de elopmen whe ein, as economic de elop-
men ises, he posi i e inno a ion e ec on g een g ow h dec eases. he
au ho s employed an imp o ed nonlinea andom eg ession S ochas ic
Impac s by Reg ession on Popula ion, A luence, and Technology
(STIRPAT) model, demons a ing ha echnology pa en s a e co ela ed
wi h he educ ion o ca bon emissions, con ibu ing o g een g ow h.
Using a h eshold panel model, Liu e al. (2022) showed ha he ela-
ionship exhibi s s uc u al b eaks when he selec ed h eshold a iables
a e in dispa a e egimes. The au ho s a gued ha inc eased knowledge
s ock, inancial de elopmen , and human capi al will ha e a mo e sig-
ni ican e ec on g een echnological inno a ion. Howe e , hey
s essed he necessi y o implemen ing igo ous and app op ia e en i-
onmen al egula ions o p omo e he impac o inno a ion on g een
g ow h. A subsequen analysis by Luan e al. (2024) employed he
nonlinea ARDL model o examine he nonlinea i y o his ela ionship
in de eloping coun ies om 1995 o 2019. The esul s e ealed ha
only posi i e shocks in g een inno a ion acili a e en i onmen al sus-
ainabili y and consequen long- e m g een g ow h.
The nonlinea ela ionship occu s in wo dis inc phases. The ini ial
phase o economic de elopmen equen ly incu s high in es men ,
hea y pollu ion, and low e iciency (Li e al., 2022; Mensah e al., 2018).
Inc eased inancial suppo o g een echnological inno a ion in u ban
a eas can ad ance he decoupling o economic g ow h om en i on-
men al p og ess, which can inc ease ene gy consump ion, wo sen ai
pollu ion, and damage en i onmen al quali y (Rezende e al., 2019;
Saliba de Oli ei a e al., 2018). The second phase ea u es a signi ican
in eg a ion o g een echnological inno a ions, wi h a ocus on en i-
onmen ally iendly p oduc s, op imized manu ac u ing p ocesses, and
enhanced ene gy e iciency (Sohag e al., 2015). This ansi ion signi -
ican ly educes a mosphe ic pollu an emissions and imp o es en i-
onmen al quali y (Zhang e al., 2023), os e ing g een g ow h.
Acco dingly, we p opose hypo hesis H01 as ollows:
H01.The ela ionship be ween g een echnological inno a ion and g een
g ow h is nonlinea .
Despi e a conside able body o li e a u e examining he impac o
g een echnological inno a ion on GHG emissions and subsequen e -
ec s on g een g ow h, p e ious esea ch has no p o ided su icien
e idence o each a de ini i e conclusion (Mei un e al., 2021). Some
s udies ha e yielded posi i e indings (e.g., Fe nandes e al., 2021;
Mei un e al., 2021; Khan e al., 2023), whe eas o he s ha e demon-
s a ed ad e se e ec s (e.g., Mensah e al., 2018; Chen & Lee, 2020;
Nosheen e al., 2021), and a mixed pic u e has eme ged om o he s
(Suki e al., 2022; Su e al., 2023). The need o u he esea ch in o
how g een echnological inno a ion impac s o he ac o s ha in luence
g een g ow h is appa en , pa icula ly enewable ene gy and he ci cula
economy.
A subs an ial co pus o s udies has demons a ed he pi o al ole o
enewable ene gies in acili a ing g een g ow h (Aneja e al., 2024).
G een ene gies ha e signi ican ly educed en i onmen al impac
(Jahange e al., 2022), equi e less dependence on non enewable en-
e gy impo s (Khan e al., 2023), and dec eased he use o non enewable
ene gy sou ces (Chen & Lei, 2018). Fu he mo e, p omo ing g een
g ow h enables he ansi ion o e icien , high-pe o mance enewable
ene gies based on sus ainable echnological inno a ions (Khan e al.,
2022). Saqib e al. (2024) ound ha ad ancemen s in hese inno a ions
and adop ing enewable ene gy sou ces ha e a combined e ec ha
posi i ely impac s long- e m g een g ow h. Mo eo e , employing he
panel causali y app oach, he au ho s demons a ed a bidi ec ional
causal associa ion be ween g een echnological inno a ion, enewable
ene gy, g een g ow h, and he ecological oo p in . This asse ion is
suppo ed by esea ch conduc ed by Saqib and Usman (2023).
The ela ionship be ween sus ainable echnological inno a ion,
enewable ene gies, and en i onmen ally esponsible g ow h has been
he subjec o se e al ecen s udies (Aydin & Degi menci, 2024;
Esmaeilpou Moghadam & Ka ami, 2024; Li e al., 2022; Sohag e al.,
2021). Fo example, Esmaeilpou Moghadam and Ka ami (2024) ound
a posi i e in e ac ion be ween sus ainable inno a ion and enewable
ene gies, highligh ing he impo ance o g een echnological inno a ion
in p omo ing enewable ene gies. In o he wo ds, ad anced g een
echnological p og ess can inc ease he adop ion o enewable ene gy
sou ces and s imula e g een economic g ow h. Simila ly, Sohag e al.
(2021) in es iga ed he impac o enewable ene gy on en i onmen ally
esponsible g ow h in 21 OECD na ions, e ealing a posi i e co ela ion
be ween enewable ene gy de elopmen , g een echnological inno a-
ion, and long- e m g een g ow h. F om his pe spec i e, as Li e al.
(2022) no ed, g een echnological inno a ion is an essen ial media o
be ween enewable ene gies and g een g ow h. Acco dingly, we p opose
hypo hesis H02 as ollows:
A. BOUATTOUR e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100748
3
H02.G een echnological inno a ion in luences he e ec o enewable
ene gy on g een g ow h.
In addi ion o he signi ican in luence o clean ene gies in g een
g ow h, op imizing esou ce u iliza ion is also pi o al o imp o ing he
equilib ium be ween he economy, he en i onmen , and socie y,
esul ing in he concep o a ci cula economy (Ma jamaa & M¨
akel¨
a,
2022), which is conside ed one o he pa hways o ad ancing he
ansi ion o g een g ow h (Figge e al., 2023). The ci cula economy can
be de ined as a esou ce u iliza ion sys em ha elimina es i gin e-
sou ces, allowing o closed esou ce loops. P omo ing ecological equi-
lib ium is acili a ed by de eloping inno a i e solu ions, as e idenced
by Bianchi and Co della (2023). Fu he mo e, in eg a ing he ci cula
economy wi h mul iple sus ainabili y objec i es ad ances he achie e-
men o he Uni ed Na ions 2030 Sus ainable De elopmen Goals
(SDGs), as highligh ed by Ga cia-Sa a ia O iz-de-Mon ellano e al.
(2023). Applying a simila ocus, De Pascale e al. (2023) analyzed he
implemen a ion o ci cula p ac ices in he EU om 2015 o Ap il 2023,
e ealing ha he ansi ion o an al e na i e ci cula pa adigm in-
c eases sus ainable g ow h in he EU.
The ci cula economy ansi ion is accele a ed by g een echnolog-
ical inno a ion, which con ibu es o educing en i onmen al damage
and consequen ly p omo es g een g ow h (Suchek e al., 2021; Chauhan
e al., 2022; Khan & Khu shid, 2022; S´
anchez-Ga cía e al., 2024).
S´
anchez-Ga cía e al. (2024) highligh ed he ans o ma i e in luence o
g een echnological inno a ion on ad ancing ci cula i y. Ou ex ensi e
su ey o ecen li e a u e e eals ha hese echnologies accele a e he
ansi ion o inc eased ci cula i y and sus ainabili y. Imp o ed esou ce
e iciency, op imized supply chains, and imp o ed p oduc li ecycle
managemen esul om g een inno a ion acili a ed by hese new
echnologies.
In his ega d, Cliquez ou appuyez ici pou en e du ex e.examining
27 EU coun ies om 2010 o 2017, Busu and T ica (2019) e ealed a
posi i e ela ionship be ween g een echnological inno a ion, he ci -
cula economy, and economic g een g ow h. Using a mul iple eg ession
model and se e al ci cula economy indica o s, he s udy e ealed he
c ucial unc ion o g een inno a ion in he e ec o he ci cula economy
on he en i onmen ally esponsible g ow h in EU coun ies. Fu he -
mo e, T igue o e al. (2022) analyzed he elemen s ha in luence he
implemen a ion o g een inno a ions in a o o a ci cula economy in
he EU and showed ha inancial and echnological capaci ies a e
decisi e o ad ancing EU economies’ ci cula i y. The indings
con i med he signi icance o g een echnological inno a ion in pu su-
ing ci cula i y and consequen p omo ion o g een g ow h in EU na-
ions.. Acco dingly, we p opose hypo hesis H03 as ollows:
H03.G een echnological inno a ion in luences he impac o he ci cula
economy on g een g ow h.
In summa y, ou e iew o he ex an esea ch indica es a endency
o conside g een echnological inno a ion, enewable ene gies, and
ci cula i y as ac o s impac ing g een g ow h independen o one
ano he , assuming a linea ela ionship. The e o e, we p opose
add essing his gap by combining hese ac o s wi hin a uni ied ame-
wo k in a nonlinea ela ionship, as pe he app oach o some p e ious
s udies (Li e al., 2021; Zhang e al., 2023). Consequen ly, we employ he
PSTAR model o in es iga e c oss-sec ional he e ogenei y, enabling
smoo h ansi ion analysis.
Me hodology
We in es iga e he e ec o sus ainable echnological inno a ion,
enewable ene gy, and ci cula i y on eco iendly g ow h and es ou
esea ch hypo heses. The panel da a model is hus o mula ed building
upon he speci ica ions o C´
a denas Rod íguez e al. (2018) and Fe -
nandes e al. (2021) as ollows:
EAMFPi =β0+β1PERTi +β2RECi +β3ICGTDi +β4GVACEi
+β5INFLi +β6ICTETi +λi+
μ
+
ε
i
(1)
whe e EAMFP ep esen s coun y i’s en i onmen ally adjus ed mul i-
ac o p oduc i i y in pe iod ; independen a iables a e deno ed by
PERT, ICGTD, REC, GVACE, INFL, and ICTET; λi ep esen s coun y ixed
e ec (FE),
μ
ep esen s ime FE, and
ε
i is a andom e o e m. Table 1
p o ides a summa y o all a iables.
Re e encing Boua ou e al. (2024) and Kalai e al. (2024), he
PSTAR model inco po a es a con inuous ansi ion unc ion, as sug-
ges ed by Gonzalez e al. (2005) and Gonzalez e al. (2017). Fu he -
mo e, i accommoda es c oss-sec ional a iabili y, making i an
a ac i e applica ion o mul isec ion panel da a esea ch, esul ing in
he ollowing o mula ion o a PSTAR model wi h wo egimes:
yi =
μ
i+∑
p
j=1
ρ
jyi −j+βʹ
1Xi +βʹ
2Xi G(qi ;γ;c) +
ε
i (2)
whe e Xi =(X1
i ,…,Xk
i )deno es a ma ix comp ising k exogenous
a iables ha lack any lagged explana o y a iables and β= (β1,…,βk)
is he ec o o hei coe icien s.
μ
i ep esen s he ec o o indi idual
FEs, and
ρ
j indica es he au o eg essi e coe icien s o p ocess yi .
G(qi ;γ;c)deno es he ansi ion unc ion associa ed wi h ansi ion
a iable qi , he h eshold pa ame e c, and a smoo hing coe icien γ.
The e o e m is
ε
i ∼ ℵ(0,
σ
2). In PSTAR modeling, a con inuous an-
si ion unc ion modi ies ansi ion unc ion G(qi ;γ;c), which spans he
in e al [0, 1]. Gonzalez e al. (2005) ecommended a logis ic ansi ion
unc ion ha has an o de designa ed as m as ollows:
G(qi ;γ;c) = [1+exp(−γ∏
n
j=1(qi −cj))]−1
,γ>0 and c1
≤c2≤cm
(3)
c= (c1,…,cm)ʹ ep esen s an m-dimensional ec o o loca ion pa-
ame e s. We use he slope pa ame e γ o de e mine he egula i y o he
ansi ions. The c i e ia γ>0 and c1≤c2≤…≤cm a e s ic ly
en o ced in his iden i ica ion p ocess. Since hese numbe s can explain
commonly no ed pa ame e luc ua ions, m =1 o m =2 is usually
adequa e o conside in eal-wo ld applica ions. Gonzalez e al. (2017)
p oposed a i s -o de Taylo expansion as a subs i u e o G(qi ;γ;c). We
cons uc he ollowing supplemen al eg ession using his me hod a e
epa ame e iza ion:
yi =
μ
i+∑
p
j=1
ρ
jyi −j+βʹ∗
1Xi +βʹ∗
2Xi G(qi ;γ;c) +
ε
∗
i (4)
whe e βʹ∗
1 is di ec ly ela ed o he slope coe icien γ and is compu ed
using
ε
∗
i =
ε
i +R1βʹ∗
1Xi , whe e R1 ep esen s he emainde o he
Taylo expansion. I is essen ial o e alua e linea i y be o e es ima ing
he PSTAR model as his assessmen in luences he s a is ical signi i-
cance o he impac o egime change. Tes ing H0:γ=0 is simila o
es ing he null hypo hesis H∗
0:βʹ∗
2=0. Wald (LM), Fishe (LR), and
likelihood a io (LMF) es s can be used o his pu pose. The app op ia e
s a is ics o hese es s a e as ollows:
LM =TN(RSS0−RSS1)/RSS0;LR = − 2[log(RSS0) − log(RSS1)]
LMF = [TN(RSS0−RSS1)/K]/[RSS0/(TN −N−K)] (5)
whe e RSS0 deno es he panel esidual sum o he squa es o a linea
panel model inco po a ing indi idual e ec s, and RSS1 ep esen s he
esidual sum o squa es o a nonlinea panel model cha ac e ized by wo
A. BOUATTOUR e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100748
4
egimes. Unde he null hypo hesis, he Wald (LMw) and likelihood a io
(LR) s a is ics ollow a chi-squa ed dis ibu ion wi h K deg ees o
eedom, whe e K is he numbe o explana o y a iables, and Fishe
Lag ange mul iplie (LMF) s a is ics adhe e o a chi-squa ed dis ibu ion
wi h wo deg ees o eedom.
Mo eo e , he PSTAR speci ica ion es ablishes he undamen al
model ou lined unde hese condi ions, p esupposing wo dis inc e-
gimes (ℎ =1 and ℎ =2) as ollows:
whe e
ρ
1j1 and
ρ
1j2 a e he coe icien s o he lagged en i onmen ally
adjus ed mul i ac o p oduc i i y (EAMFPi −j) a each egime, h =1 o
h=2.
Resul s
This sec ion p esen s he desc ip i e analysis; uni oo , coin eg a-
ion, independence es s; PSTAR es ima ion; and he causali y es on
panel da a.
Desc ip i e s a is ics
As illus a ed in Table 2, he EAMFP a iable de i ed om 621 ob-
se a ions has a mean alue o 1.831 and a co esponding s anda d
de ia ion o 3.198. The se ies also exhibi s a lep oku ic dis ibu ion,
wi h a skewness o −0.261 <0 and a ku osis o 4.839 >3. All alues
all wi hin he ange o −12.933–11.722, and he concen a ion is 1.566.
As no ed p e iously and subs an ia ed by he Ja que and Be a (1987)
es conduc ed on he en i e sample, he se ies does no adhe e o a
no mal dis ibu ion. Fu he mo e, he se ies exhibi s a lack o au oco -
ela ion a he 1% le el, as e idenced by he Bo n and B ei ung (2016)
es .
The mean o he PERT a iable is 12.608 wi h a s anda d de ia ion
o 6.870. The concen a ion o he dis ibu ion is 11.816, wi h a ange o
alues om 0 o 91.428. This se ies exhibi s a igh -asymme ic dis i-
bu ion, as indica ed by he skewness alue o 3.663, which is g ea e
han 0. Addi ionally, he ku osis alue o 34.261 e eals ha he dis-
ibu ion is p edominan ly lep oku ic. The esul s o Bo n and
Table 1
Va iable de ini ions.
Va iables Na u e P oxy De ini ion Re e ences Sou ce
EAMFP Endogenous G een g ow h En i onmen ally adjus ed mul i ac o
p oduc i i y.
Alb izio e al. (2017); Hao e al. (2021); He man e al.
(2023); Liu e al. (2022)
OECD
PERT Exogenous
and
h eshold
G een
echnological
inno a ion
Ra io o en i onmen ally- ela ed
echnology pa en s o o al pa en s.
Bekhe & La i (2018); Du e al. (2021); Esmaeilpou
Moghadam & Ka ami (2024); Fe nandes e al. (2021);
Mensah e al. (2018); Yu e al. (2021)
OECD
REC Exogenous Renewable ene gy Sha e o enewable ene gy in o al
inal ene gy consump ion.
Jahange e al. (2022); Khan e al. (2022); Khan e al.
(2023)
Wo ld Bank Wo ld
De elopmen Indica o s
(WDI)
ICGTD Exogenous Sus ainable
echnology ans e
In e na ional collabo a ion in global
echnology de elopmen .
Aghabalaye & Ahmad (2022); Fe nandes e al. (2021) OECD
GVACE Exogenous Ci cula economy G oss alue added in he ci cula
economy, exp essed as a p opo ion
o GDP.
Chauhan e al. (2022); Khan & Khu shid (2022);
S´
anchez-Ga cía e al. (2024); Suchek e al. (2021);
V anjanac e al. (2023)
Eu os a
INFL Con ol In la ion Ra e o in la ion as a p opo ion o he
consume p ice index.
Chen e al. (2022) WDI
ICTET Con ol Digi aliza ion Ra io o in o ma ion and
communica ion echnology (ICT)
expo s o ICT impo s.
Bilal e al. (2022); Cao e al. (2021) WDI
Table 2
Desc ip i e s a is ics.
S a is ics EAMFP PERT REC ICGTD GVACE INFL ICTET
Mean 1.831 12.608 17.776 61.645 1.726 2.850 75.881
Median 1.566 11.816 15.720 64.000 1.600 2.183 67.548
S anda d de ia ion 3.198 6.870 12.051 21.961 0.854 3.691 38.510
Minimum −12.933 0.000 0.000 5.000 0.400 −4.478 13.290
Maximum 11.722 91.428 58.770 119.000 6.400 45.666 300.451
Skewness −0.261 3.663 0.837 −0.405 3.450 4.804 1.707
Ku osis 4.839 34.261 3.175 2.818 18.145 43.112 7.329
Ja que–Be a (JB) 94.65 2.7 e
+04
73.32 17.89 7168 4.4 e
+04
786.700
JB p obabili y 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Bo n–B ei ung (BB) 26.11 23.59 45.09 13.33 7.91 27.58 10.56
BB p obabili y 0.000 0.000 0.000 0.001 0.019 0.000 0.005
EAMFPi =(∑
p
j=1
ρ
1j1EAMFPi −j+β01+β11PERTi +β21ICGTDi +β31RECi +β41GVACEi +β51INFLi +β61ICTETi +λ1i+
μ
1 +
ε
1i )
+(∑
p
j=1
ρ
1j2EAMFPi −j+β02+β12PERTi +β22ICGTDi +β32RECi +β42GVACEi +β52INFLi +β62ICTETi +λ2i+
μ
2 +
ε
2i )
×G(PERTi −1;γ;c)
(6)
A. BOUATTOUR e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100748
5
B ei ung’s au oco ela ion es (2016) and Ja que and Be a’s no mali y
es (1987) indica e a p obabili y o unde 1%.
The ICGTD a iable has a mean o 61.645 and a s anda d de ia ion o
21.961 wi h alues anging om 5 o 119 and a concen a ion o 64.
Fu he mo e, his se ies exhibi s a pla yku ic dis ibu ion wi h a
skewness o −0.405 <0 and a ku osis o 2.818 <3. The esul s o
au oco ela ion and no mali y es s indica e a less han 1% p obabili y
o a second-o de au oco ela ion p oblem, ejec ing he null hypo hesis.
The GVACE a iable has minimal he e ogenei y, wi h a 0.495 coe -
icien o a ia ion, indica ing a mean alue o 1.726 and a co e-
sponding s anda d de ia ion o 0.854. The concen a ion o he alues is
1.600, anging be ween 0.4 and 6.4. The se ies’ dis ibu ion is cha ac-
e ized by a igh -asymme ic shape, as e idenced by a skewness alue
g ea e han 0, and a p edominan ly lep oku ic na u e, as indica ed by a
ku osis alue g ea e han 3. Addi ionally, he p obabili y o he
obse ed da a being a andom sample is less han 1%, which is sub-
s an ia ed by he au oco ela ion and no mali y es s.
A e he desc ip i e analysis we asce ain whe he a dependen
ela ionship is e iden be ween indi idual a iables in he model. As he
p esence o c oss-sec ional dependence in panel da a could esul in
skewed and con adic o y empi ical indings, i is impe a i e o examine
his possibili y (Phillips & Sul, 2003). The obse a ion ha wo coun-
ies a e highly dependen on one ano he does no imply ha hey ha e
he same de elopmen dynamics, and i is concei able ha each na ion
has unique cha ac e is ics.
As no ed by Wolde-Ru ael (2014), he supposi ion o pa ame e ho-
mogenei y complica es he accommoda ion o po en ial a iabili y
based on coun y-speci ic ac o s (B ei ung, 2005). Consequen ly, a
ange o es s ha e been de eloped o asce ain he p esence o a
dependen ela ionship in his con ex . In he con ex o his analysis, we
o mula e an H
0
hypo hesis, which pos ula es he absence o
c oss-sec ional dependence, and an H
1
hypo hesis, which p oposes i s
p esence. I he null hypo hesis (H
0
) is subs an ia ed, he analysis can
p oceed wi h i s -gene a ion panel uni oo es s. Con e sely, when
al e na i e one (H
1
) is alida ed, he in es iga ion should ansi ion o
second-gene a ion panel uni oo es s (Bal agi, 2021).
When examining he e ogeneous slopes, he same logic applies o he
alse assump ion o homogenei y as o he he e ogenei y o he ac ual
slopes. We use he Hashem Pesa an and Yamaga a (2008) slope he -
e ogenei y es o examine possible he e ogenei y, which is an ex ended
e sion o he Swamy (1970) es . Acco ding o Hashem Pesa an and
Yamaga a (2008), his es examines he panel’s homogenei y.
Table 3 examines he p esence o dependen ela ionships be ween
indi iduals using hese es s. Each es e i ies he p esence o an indi-
idual dependence wi h p obabili ies ha a e less han 1%. The e o e
he homogenei y hypo hesis is uled ou a he 1% h eshold acco ding
o Pesa an and Yamaga a’s (2008) homogenei y es . The indings e eal
ha he model coe icien s do no exhibi homogenei y. Consequen ly, i
is impe a i e o implemen he Pesa an (2007) second-gene a ion uni
oo es , which can elimina e c oss-sec ional dependence by using a
su oga e a iable and employs delayed c oss-sec ional means in i s
a ian o he ex ended augmen ed Dickey–Fulle uni oo es . The
indings o his es in Table 4 indica e ha all model a iables a e s a-
iona y in he i s di e ence.
Re e encing Ka a ias and Tza alis (2014), his s udy uses he uni
oo es ha inco po a es a s uc u al b eak o p o ide addi ional
suppo o ou conclusions. As shown in Table 5, he esul s demon-
s a e ha he se ies a e s a iona y a he le el, wi h s uc u al b eaks
occu ing in 2001, 2003, 2005, 2014, and 2021. Examining b eakpoin
da es e eals signi ican pe iods o s uc u al change in he ime se ies.
The subsequen analysis u nishes a amewo k o unde s anding he
dynamics o s uc u al shi s in he da a unde sc u iny. These b eak-
poin s indica e pe iods when he unde lying s a is ical p ope ies ha e
unde gone no able a ia ions.
Oil p ice luc ua ions and he geopoli ical ins abili y ha ollowed
he wa in A ghanis an in 2001 signi ican ly a ec ed EU economies
(Ba sky & Kilian, 2004). Addi ionally, signi ican economic u bulence
in 2003 esul ed om luc ua ing oil p ices d i en by geopoli ical un-
ce ain y. This global economic uphea al a ec ed EU economies,
pa icula ly conside ing he inancial allou om he I aq Wa (Husain
e al., 2024). Mo eo e , in 2005, he EU aced speci ic challenges asso-
cia ed wi h in e nal complexi ies, including ICT and egion-speci ic
dynamics. No able issues encompassed dispa i ies in economic condi-
ions ac oss EU membe s a es, deba es abou he Eu opean Cons i u-
ion, and unce ain y o e ene gy and oil cos s (Hol g ewe, 2014).
Fu he mo e, he EU economy expe ienced signi ican epe cussions
om he deb c isis in 2014, which a ec ed speci ic na ions in he
eu ozone, pa icula ly G eece, Po ugal, and Spain. Ele a ed public
deb , subs an ial budge de ici s, and conce ns ega ding he long- e m
sus ainabili y o public inances con ibu ed o he ad en o his
c isis. The a ec ed coun ies expe ienced inc eased bo owing cos s
om he cau ious esponse o inancial ma ke s (Sacchi, 2016). Se e al
coun ies wi hin he eu ozone we e compelled o implemen aus e e
iscal policies ha educed public spending and inc eased axa ion o
es o e iscal equilib ium. Implemen ing hese aus e i y measu es had
ad e se economic consequences, including declining economic ac i i y,
inc eased unemploymen , and social con lic s. Fu he mo e, he global
economic impac o he COVID-19 pandemic was signi ican , and he EU
was no exemp om he economic epe cussions o his wo ldwide
heal h eme gency. Implemen ing con ainmen measu es, he imposi ion
o a el es ic ions, and supply chain dis up ions esul ed in a sub-
s an ial economic down u n in some EU coun ies.
Be o e e alua ing econome ic models, i is impe a i e o ho oughly
examine he coin eg a ion ela ionships be ween a iables. We conse-
quen ly employ he second-gene a ion coin eg a ion es p oposed by
Pe syn and Wes e lund (2008) as i conside s ci cums ances o in-
di iduals’ c oss-sec ional eliance. Table 6 e eals ha he null hy-
po hesis o no coin eg a ion be ween he a iable models should be
ejec ed. This indica es ha he model’s a iables a e ela ed o one
ano he o e he long e m and ha a leas one coin eg a ing link is
e iden among hem.
PSTAR es ima ion and in e p e a ion
We employ he PSTAR me hodology o es ima e he undamen al
model and examine he nonlinea ela ionship be ween g een inno a-
ion and g een g ow h o he 27 EU coun ies o de e mine whe he he
o de m equals one o es o linea i y. The esul s o he speci ica ion
es in Table 7 demons a e ha he al e na i e logis ic PSTAR speci i-
ca ion (m =2) is accep ed and he null hypo hesis o linea i y is ejec ed.
The linea i y es s a e only he i s s ep be o e mo ing on o he inal
PSTAR model es ima ion. An equally impo an second s ep is de e -
mining he ideal numbe o ansi ion unc ions, which enables us o
de e mine he numbe o egimes ha bes desc ibe he dynamics o he
ela ionship be ween g een inno a ion and g een g ow h. I m =2, wo
h esholds pe egime appea . Fo he i s egime, c11 =8.026 and
c12 =9.284, which gi es a mean i s h eshold o c1=8.655. Fo he
Table 3
Dependence and homogenei y es s.
Tes s Value p- alue Decision
Dependence es s
B eusch and Pagan (1980) 2155 0.000 Dependence
F iedman (1937) 212.013 0.000 Dependence
F ees (1995, 2004) 3.997 0.000 Dependence
Pesa an (2004) 39.53 0.000 Dependence
Pesa an (2006) 42.285 0.000 Dependence
Pesa an e al. (2008) 131.5 0.000 Dependence
Pesa an (2015) 42.116 0.000 Dependence
Homogenei y es
Pesa an and Yamaga a (2008) 3.875 0.000 He e ogenei y
4.799 0.000
A. BOUATTOUR e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100748
6
second egime, c21 =11.769 and c22 =12.408, which gi es a mean
second h eshold o c2=12.089.
The esul s in Table 8 e eal wo op imal h esholds o 8.655 and
12.089, indica ing ha he ansi ion pa ame e o he PSTAR model,
wi h alues o 88.547 and 116.88, has a pi o al in luence on his phe-
nomenon. These indings indica e a nonlinea co ela ion be ween
EAMFP and PERT. The Akaike in o ma ion c i e ion (AIC) and he
Bayesian in o ma ion c i e ion (BIC) a e 1.841 and 2.055, espec i ely,
while he esidual sum o squa es (RSS) is 3375.571.
Table 9 e eals h ee egimes. Regime 1 ep esen s a h eshold o
8.655%, Regime 2 ep esen s a h eshold be ween 8.655% and
12.089%, and Regime 3 ep esen s a h eshold o 12.089%. In Regime 1,
g een inno a ion has a 5% signi ican posi i e impac on g een g ow h.
ICGTD, REC, and GVACE signi ican ly in luence EAMFP. An 1% inc ease
in he la e esul s in a espec i e dec ease o 0.116, 0.379, and 0.351 in
he EAMFP.
Regime 2 ep esen s a h eshold be ween 8.655% and 12.089%,
indica ing ha a 1% inc ease in PERT posi i ely impac s EAMFP by
0.091%. Simila ly, a 1% inc ease in ICGTD and REC esul s in a signi -
ican imp o emen , espec i ely inc easing EAMFP by 0.126% and
0.124%. In con as , a 1% inc ease in GVACE and INFL has an ad e se
e ec , educing EAMFP by 0.446% and 0.424%, espec i ely.
Con e sely, a 1% inc ease in ICTET has a posi i e ma ginal impac ,
inc easing EAMFP by 0.006%.
Regime 3 ep esen s a h eshold o 12.089%, e ealing ha a 1%
inc ease in PERT inc eases EAMFP by 0.066%. Simila ly, a 1% inc ease
ICGTD and REC signi ican ly inc ease EAMFP by 0.003% and 0.019%,
espec i ely. In addi ion, a 1% inc ease in GVACE, INFL, and ICTET
espec i ely aises EAMFP by 0.124%, 0.039%, and 0.015%.
Figs. 1 and 2 illus a e he ansi ion unc ion o PERT. This unc ion
explains how i a ies o e ime, pa icula ly as i mo es om a low and
in e media e egime o a high egime. The i s ansi ion unc ion has a
subs an ial alue (γ11 =88.547). In con as , he second ansi ion
unc ion is γ12 =116.88, which is also s ong, indica ing ha he
ansi ion is ab up . The high alue indica es ha he PERT a iable,
which explains echnological inno a ions associa ed wi h he en i on-
men , has an immedia e in luence on he EAMFP a iable, which mea-
su es mul i ac o p oduc i i y adjus ed o he en i onmen . The
con inuous e ec o PERT on EAMFP indica es ha ansi ions owa d
g een inno a ion ha e a signi ican and las ing in luence. G een in-
no a ions enhance p oduc ion p ocess e iciency, cu ail he gene a ion
o ou pu s ha a e de imen al o he en i onmen and p ese e na u al
esou ces. These posi i e e ec s and he con inuous impac o PERT on
EAMFP sugges ha ansi ions owa d g een inno a ion ha e a signi -
ican and las ing in luence, meaning ha he bene i s o g een inno a-
ion a e immedia e and p olonged o e ime.
F om a con ol pe spec i e, he esul s demons a e a s a is ically
signi ican posi i e impac o ICTET on EAMFP, which aligns wi h p e-
ious li e a u e. (Chen e al., 2022). Simila ly, he posi i e impac o
INFL on EAMFP indica es ha INFL imp o es en i onmen al quali y
(Bilal e al., 2022), enhancing g een g ow h.
Robus ness G ange noncausali y es
The G ange noncausali y es is a s a is ical model ha is used o
assess whe he he his o ical alues o one a iable can help p edic
Table 4
Second-gene a ion uni oo es .
Va iables EAMFP PERT REC ICGTD GVACE INFL ICTET
In le el
Wi h cons an −3.296***−3.204***−2.594***−1.546 −1.874 −3.350***−2.012*
Wi h end −3.271***−3.682***−2.776** −2.553 −2.908***−3.635***−2.528
Decision S S S NS NS S NS
In i s di e ence
Wi h cons an −4.767***−5.312***−4.802***−2.857***−5.321***−4.713***−4.168***
Wi h end −5.033***−5.515***−4.884***−3.604***−5.386***−4.913***−4.269***
Decision S S S S S S S
No e. NS ep esen s nons a iona i y; S ep esen s s a iona i y. Signi icance: ***p <0.01
**
p <0.05.
Table 5
Uni oo es wi h s uc u al b eaks.
Va iables In le el Fi s di e ence
EAMFP −23.770*** (2021) −35.892*** (2021)
PERT −23.862*** (2021) −41.600*** (2021)
REC −25.634*** (2001) −35.906*** (2001)
ICGTD −2.128* (2005) −29.696*** (2021)
GVACE −3.934*** (2014) −38.982*** (2021)
INFL −18.678*** (2001) −31.140*** (2021)
ICTET −6.882*** (2003) −29.381*** (2021)
No e. Signi icance: ***p <0.01.
Table 6
Second-gene a ion coin eg a ion es s.
Pe syn and Wes e lund (2008) Z- alue Robus p- alue Decision
G
−6.416 0.000 Coin eg a ion
G
a
0.746 0.000 Coin eg a ion
P
−5.040 0.000 Coin eg a ion
P
a
−1.232 0.000 Coin eg a ion
Table 7
PSTAR linea i y es .
Tes s H
0
: =0 s. H
1
: =1 H
0
: =1 s. H
1
: =2 H
0
: =2 s. H
1
: =3
-S a is ic P obabili y -S a is ic P obabili y -S a is ic P obabili y
Wald (LM) 64.198 0.000 65.156 0.000 6.073 0.639
Fishe (LMF) 4.165 0.000 4.117 0.000 1.893 0.059
Likelihood a io (LR) 67.765 0.000 68.834 0.000 6.285 0.615
Table 8
PSTAR op imal h eshold.
O de Th esholds (c) T ansi ion
pa ame e (γ)
RSS AIC BIC
=1c11 =8.026;c12 =
9.284 →c1=8.655
γ1=
88.547
3,375.571 1.841 2.055
=2c21 =11.769;c22 =
12.408 →c2=
12.089
γ2=
116.88
A. BOUATTOUR e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100748
7
Table 9
PSTAR es ima ion.
Va iables Regime 1: PERT
i -1
≤8.655 Regime 2: 8.655 <PERT
i -1
≤12.089 Regime 3: PERT
i -1
>12.089
EAMFP
i -1
0.654 2.625*** 0.425 2.327** −1.111 −6.444***
EAMFP
i -2
−0.648 −2.991*** 0.051 0.325 0.668 4.241***
PERT
i
0.109 2.408** 0.091 2.453** 0.066 2.394**
ICGTD
i
−0.116 −3.635*** 0.126 4.579*** 0.003 2.277**
REC
i
−0.379 −6.798*** 0.124 2.753*** 0.019 2.795***
GVACE
i
−0.351 −2.013** −0.446 −2.823*** 0.124 2.165**
INFL
i
0.485 1.275 −0.424 −2.172** 0.039 2.321**
ICTET
i
0.004 0.318 0.006 2.500** 0.015 2.238**
Obse a ions 119 61 441
No e. Signi icance: ***p <0.01
**
p <0.05.
Fig. 1. EAMFP i s es ima ed ansi ion unc ion o PERT.
Fig. 2. EAMFP second es ima ed ansi ion unc ion o PERT.
A. BOUATTOUR e al.
Jou nal o Inno a ion & Knowledge 10 (2025) 100748
8
