Hansmeie , Hend ik; Losacke , Sebas ian; Be sch, Johannes; K oll, Henning
A icle — Published Ve sion
Regional Specializa ions in G een Incumben s and G een
S a ‐ups in he Ge man T anspo Sec o
G ow h and Change
P o ided in Coope a ion wi h:
John Wiley & Sons
Sugges ed Ci a ion: Hansmeie , Hend ik; Losacke , Sebas ian; Be sch, Johannes; K oll, Henning
(2025) : Regional Specializa ions in G een Incumben s and G een S a ‐ups in he Ge man T anspo
Sec o , G ow h and Change, ISSN 1468-2257, Wiley, Hoboken, NJ, Vol. 56, Iss. 1,
h ps://doi.o g/10.1111/g ow.70025
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/319367
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/
G ow h and Change
-
ORIGINAL ARTICLE
OPEN ACCESS
Regional Specializa ions in G een Incumben s and G een
S a ‐ups in he Ge man T anspo Sec o
Hend ik Hansmeie
1,2
| Sebas ian Losacke
3,4
| Johannes Be sch
5
| Henning K oll
1,2
1
F aunho e Ins i u e o Sys ems and Inno a ion Resea ch ISI, Ka ls uhe, Ge many |
2
Ins i u e o Economic and Cul u al Geog aphy, Leibniz Uni e si y,
Hanno e , Ge many |
3
Depa men o Geog aphy, Jus us Liebig Uni e si y, Hessen, Ge many |
4
CIRCLE–Cen e o Inno a ion Resea ch, Lund Uni e si y,
Lund, Sweden |
5
ZEW ‐ Leibniz Cen e o Eu opean Economic Resea ch, Mannheim, Ge many
Co espondence: Hend ik Hansmeie ([email p o ec ed])
Recei ed: 1 Sep embe 2023 | Re ised: 23 Sep embe 2024 | Accep ed: 3 Janua y 2025
Keywo ds: ac o s | eco‐inno a ion | g een egional de elopmen | incumben s | egional specializa ion | s a ‐ups
ABSTRACT
The egional a ie y o ac o s is conside ed a key de e minan in he las decade's ich li e a u e on he geog aphy o eco‐
inno a ion and g een egional de elopmen . Howe e , li le is known abou he ex en o which egions di e in hei
specializa ion in new and es ablished eco‐inno a ion ac o s. In his a icle, we p opose a egional ypology based on g een
specializa ions conce ning bo h incumben s and s a ‐ups in he Ge man anspo sec o . While many egions show g een
specializa ions in ei he s a ‐ups o incumben s, only some egions manage o specialize in bo h. We ind ha he abo e‐a e age
egional specializa ion in eco‐inno a ion does no seem o be p ima ily a phenomenon o u ban a eas, bu a he depends on
egions' human capi al endowmen s and echnological capabili ies. The obse ed he e ogenei y in eco‐inno a ion specializa-
ions, bo h in inno a ion cen e s and lagging egions, calls o egional policies ha a e mo e sensi i e o hese di e ences.
1
|
In oduc ion
The g eening o he economy is decisi e o achie e ans-
o ma i e change owa d en i onmen al sus ainabili y. The
anspo sec o is unde pa icula p essu e o ans o m; be-
sides nega i e en i onmen al impac s esul ing om land seal-
ing, local pollu ion and mic oplas ic was e (Baensch‐Bal uscha
e al. 2020), i is la gely based on ossil uels, accoun ing o
app oxima ely one qua e o ene gy‐ ela ed and 15% o global
g eenhouse gas (GHG) emissions (IPCC 2022). As mo e goods
and passenge s a e anspo ed by oad, ail, ai and wa e ,
sec o emissions in coun ies o he Global No h a e e en
highe (20%–30%), wi h no signi ican decline obse ed in ecen
yea s (Clima eWa ch 2022; F ansen e al. 2019).
Coun ies and egions di e no only in e ms o causing global
en i onmen al challenges, bu also in add essing and sol ing
hem. The la e is he ocus o esea ch on he geog aphy o eco‐
inno a ion and sus ainabili y ansi ions ha has eme ged in he
pas decade (Hansen and Coenen 2015; Hansmeie and
K oll 2024; Losacke e al. 2023), poin ing o he impo ance o
place‐speci ic and egional ac o s o eco‐inno a ion p ocesses.
Geog aphical and ela ed o ms o p oximi y acili a e he eme -
gence, di usion and applica ion o knowledge and inno a ion
h ough access o ne wo ks and esou ces. As analyzed by p e-
ious wo k, he place‐speci ic na u e o g een echnology, in-
dus y and egional de elopmen is mainly due o (in‐) o mal
ins i u ions including policies (e.g., Bækkelund 2022; Bugge,
Ande sen, and S een 2022), echnological capabili ies (e.g., Co -
adini 2019; San oalha and Boschma 2021) as well as local ac o s
(e.g., Binz, T u e , and Coenen 2016; MacKinnon e al. 2019).
In he li e a u e, nume ous s udies ocus on he speci ic oles
o di e en g oups o egional eco‐inno a o s as egional
This is an open access a icle unde he e ms o he C ea i e Commons A ibu ion License, which pe mi s use, dis ibu ion and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly
ci ed.
© 2025 The Au ho (s). G ow h and Change published by Wiley Pe iodicals LLC.
G ow h and Change, 2025; 56:e70025 1 o 15
h ps://doi.o g/10.1111/g ow.70025
condi ions o eco‐inno a ion de elopmen . These include p i-
ma ily ac o s om indus y, poli ics and academia, bu also
in e media ies and socie al ac o s (e.g., Gibbs and Jensen 2022;
T ippl e al. 2020). Howe e , he ela ionship be ween es ab-
lished and new ac o s is also c ucial o inc ease ou unde -
s anding o egional sou ces o ans o ma i e change ( an den
Be ge, We e ings, and Alkemade 2020). Al hough inno a ion
s udies ha e analyzed he in luence o egional con ex s on he
eme gence o eco‐inno a ions in incumben i ms (e.g., Ho bach
and Ramme 2018; Kli kou and Coenen 2013) as well as in
g een s a ‐ups (e.g., Abdesselam, Kedja , and Renou‐Mais-
san 2024; Co adini 2019), bo h aspec s ha e ha dly been
conside ed oge he in o de o be e g asp he egionally
p e alen balance o di e en eco‐inno a ion ac o s. On he
o he hand, ansi ion s udies' in e es in s abili y and change is
e lec ed bo h in he ocus on he o ces o ine ia and in no el
de elopmen s depending on expe imen al alignmen (Boschma
e al. 2017; Gibbs and Jensen 2022). While o e a ching s udies
on di e en combina ions o egional ac o s a e missing, ecen
s udies a gue, howe e , ha incumben s ope a e qui e he e o-
geneously and do no only p ese e he s a us quo, as o en
assumed. Acco dingly, change can eme ge om bo h in-
cumben s and new en an s (S een and Wea e 2017; S ambach
and P li sch 2020; Tu nheim and So acool 2020).
The aim o his pape is o gain a be e unde s anding o e-
gions' g een specializa ion, ha is he deg ee o which inno a-
ion ac i i ies in he anspo sec o ha e become g eened in
he domains o (economic) incumben s and s a ‐ups. Eco‐
inno a i e ac i i ies in he Ge man anspo sec o a e o
c ucial impo ance no only om an ecological poin o iew,
bu also om an economic one. The au omo i e indus y alone
comp ises a ound 1000 companies (manu ac u e s and sup-
plie s) wi h a ound 830,000 employees, wi h medium‐sized o
la ge companies domina ing (Des a is 2023). T anspo ‐ ela ed
indus ies a e o conside able economic impo ance ac oss e-
gions, which makes he sec o an ideal subjec o in es iga ion,
gi en ha his s udy ocuses on egional pa icula i ies o eco‐
inno a ion eme gence.
Following Hocke s and Wüs enhagen (2010), we concei e
g een incumben s as o ganiza ions, usually i ms, ha a e well
es ablished and engage in eco‐inno a ion ac i i ies, while g een
s a ‐ups a e ecen ly ounded companies wi h an inno a i e
and en i onmen ally iendly business model o p oduc . By
ocusing on egional specializa ions, we in en ionally ocus less
on how he join exis ence o es ablished and newly ounded
companies de elops (which would be a biased compa ison), bu
a he on he ex en o which he wo ypes o ac o s in a egion
con ibu e o he p ocess o g eening. In addi ion o he gene al
ques ion o how egions di e in e ms o ela i e ac o spe-
cializa ions, we a e pa icula ly seeking o analyze he join
egional specializa ion o g een anspo incumben s and s a ‐
ups. Finally, we es which egional condi ions explain he
obse ed spa ial pa e ns.
Empi ically, we es ima e he g eening o incumben s based on
en i onmen ally o ien ed pa en s, which a e almos exclusi ely
iled by la ge i ms in he anspo sec o (Eu os a 2014), and
g een s a ‐ups based on da a om he Mannheim En e p ise
Panel (MUP) o he pe iod 2009–2018 in he 96 labo ma ke
egions in Ge many. As we analyze no only echnological
change ia pa en s bu also business model o se ice in-
no a ions, he combined analysis allows us o examine he
b oade eco‐inno a i e change necessa y o sus ainabili y
ansi ions in he anspo sec o (Töd ling, T ippl, and
Desch 2022).
The ou line o his pape is as ollows. The nex sec ion p esen s
he heo e ical and empi ical s a us quo o geog aphical
esea ch on eco‐inno a o s wi h a pa icula ocus on he
anspo sec o . Based on his, esea ch ques ions a e de i ed.
Subsequen sec ions desc ibe he da a and me hods o his s udy
be o e we p esen key indings. This is ollowed by a discussion
o he esul s and a conclusion, poin ing o u he esea ch
possibili ies and policy implica ions.
2
|
Regional Pe spec i es on Eco‐Inno a ion
Incumben s and S a ‐Ups in he T anspo Sec o
Inno a ions a e he esul o new combina ions o goods, skills,
knowledge, echniques and esou ces. This implies he need o
collabo a ing and in e ac i e ela ions, which led o he sys emic
unde s anding, ha is sec o al, echnological, egional/na ional,
o inno a ion eme gence and applica ion. F om he pe spec i e
o geog aphical esea ch on eco‐inno a ions and sus ainabili y
ansi ions, ques ions a ise no only abou spa ial di e ences in
echnology and indus ial de elopmen (Capasso e al. 2019;
Töd ling, T ippl, and F angenheim 2020), bu also abou he
ele an ac o s. Mo e p ecisely, whe he , acco ding o he
Schumpe e ian dicho omy, he necessa y change is p ima ily
ca ied ou by es ablished ac o s o whe he new en an s a e
equi ed and how egions di e in his espec (T ippl
e al. 2020; Tu nheim and Geels 2019).
2.1
|
The Role o Incumben s in Eco‐Inno a ion
Ac i i ies
Gi en he he e ogenei y o ac o s ha cons i u e complex
inno a ion and socio‐ echnical sys ems, a mul iplici y o in-
cumben s can be iden i ied. Hence, incumben s exis and ac o
a ying deg ees a di e en (spa ial) le els and social sphe es,
wi h incumben beha io pa icula ly isible among powe ul
i ms and go e nmen ac o s (Ansa i and K op 2012; Tu nheim
and Geels 2019). Incumben ac o s a e closely linked o he
egime concep o ansi ions esea ch, acco ding o which
ou ine agency o es ablished ac o s ep oduces (in‐) o mal in-
s i u ions and ules, which in u n s abilize socio‐ echnical
sys ems ha lead o pa h dependencies and inc emen al
changes (Miö ne and Binz 2021). These pa h‐dependen de-
elopmen s a e he esul , o example, o incumben i ms
p edominan ly adhe ing o exis ing and po en ially success ul
business logics, which slows down o p e en s ansi ions o-
wa d sus ainabili y (Bohnsack, Pinkse, and Kolk 2014; S een
and Wea e 2017).
Wi h ega d o he anspo sec o , hese pa e ns a e pa icu-
la ly e iden o he globally dominan au omo i e egime
ha has e ol ed o mo e han a cen u y, leading o s able
2 o 15 G ow h and Change, 2025
p oduc ion and consump ion p ac ices (Senge s and Ra-
en 2015). This egime has been mani es ed h ough echnical,
ins i u ional and social adap a ions wi h li le sys em‐changing
inno a ion ac i i ies o e he las decades (So acool, Noel, and
O sa o 2017). Al hough es ablished i ms a e ac i e in egional
and na ional esea ch e o s and p oduc ion, hey a e also
s ongly in eg a ed in o in e na ional ma ke s and collabo a-
ions, p omo ing simila anspo modes a ound he wo ld
(Nilsson, Hillman, and Magnusson 2012).
Despi e hese cha ac e is ics, howe e , inno a ion and ansi-
ion s udies poin o a mo e di e se pic u e o incumben s (e.g.,
Ansa i and K op 2012; S een and Wea e 2017) and plu alizing
pe spec i es on incumbencies ac oss egions and sec o s
(Tu nheim and So acool 2020). These iews a e based on he
indings ha es ablished i ms and o he incumben s may pu -
sue eco‐inno a ion de elopmen . Al hough incumben i ms
seem less likely o be ini ial leade s o en i onmen ally o ien ed
inno a ion, di e ences eme ge depending on he indus y
en i onmen and i m p ope ies, he mode o inno a ion, as
well as challenges om new en an s (Ansa i and K op 2012;
Tsou i, Hanson, and No mann 2021; Tu nheim and So a-
cool 2020). In addi ion, he e a e a ying ins i u ional ield
logics, such as di e ences in s anda diza ion (Dewald and
Ach e nbosch 2016), which a e likely o ansla e in o spa ial
di e ences (Miö ne and Binz 2021). Fo he anspo sec o ,
incumben s a e signi ican ly in ol ed in he de elopmen o
en i onmen ally iendly anspo echnologies and business
models in egions, o example, sha ed o au onomous mobili y
(Bohnsack, Pinkse, and Kolk 2014; Meelen, F enken, and
Hob ink 2019; Nilsson and Nyk is 2016). A he same ime, i
was ound ha incumben s can also in luence he ins i u ional
amewo k in a way ha p omo es ans o ming he anspo
sec o a he egional le el (Bugge, Ande sen, and S een 2022;
Miö ne and T ippl 2019).
Agains his backg ound, he geog aphy o eco‐inno a ion has
ecei ed inc easing scien i ic a en ion o e he pas decade
(Hansmeie and K oll 2024; Losacke e al. 2023). While he
complexi y o g een ansi ions and ans o ma i e change is
cha ac e ized by mul i‐scala in e dependencies in inno a ion
p ocesses, ne wo ks o ac o s and knowledge dynamics (Capa-
sso e al. 2019; MacKinnon e al. 2019; Tsou i, Hanson, and
No mann 2021), nume ous geog aphical s udies poin o he
impo ance o di e si ied capabili ies and knowledge bases a
he egional le el. In he la e case, he ela edness o p e‐
exis ing capabili ies is o pa icula ele ance o g een
egional di e si ica ion (Mon eso and Qua a o 2020; San-
oalha and Boschma 2021). As such, a na ow egional
specializa ion seems insu icien (Coenen 2015). Since he g een
ansi ion will a ec he anspo sec o in pa icula , egions
wi h incumben s uc u es will be called upon o add ess change
p ocesses a an ea ly s age and o de elop al e na i e di e si i-
ca ion pa hways. O he wise, hese egions will be con on ed
wi h pa icula ly nega i e de elopmen s (Rod íguez‐Pose and
Ba alucci 2023). In his con ex , Miö ne and T ippl (2019)
poin o he impo ance o es ablishing new ne wo ks wi h
anspo s akeholde s o ensu e access o and co‐c ea e
knowledge abou mo e sus ainable anspo solu ions in he
u u e. Al hough g een inno a ion ac i i ies o i ms in u ban
en i onmen s a e no necessa ily in luenced by spa ial
ex e nali ies (Galliano, Nadel, and T iboule 2023), agglome a-
ion e ec s, clus e ing, co‐loca ion, in as uc u e densi y and
linkages be ween di e en ac o s, which a e undamen al o
many u ban a eas, a e ound o posi i ely impac o e all eco‐
inno a ion ou pu (Del Río, Peñasco, and Rome o‐Jo -
dán 2016; G illi sch and Hansen 2019; Ho bach 2020) and g een
inno a ion ac i i ies in he anspo sec o in pa icula (Ca -
alho, Minga do, and an Haa en 2012; K auss, K ail, and
Axhausen 2022). This is in line wi h he equen ly con i med
inding ha inno a ion ac i i ies a e concen a ed in mo e
densely popula ed egions (e.g., Balland e al. 2020; Mewes and
B oekel 2022).
2.2
|
The Role o S a ‐Ups in Eco‐Inno a ion
Ac i i ies
Eco‐inno a i e s a ‐ups a e c ucial in ans o ming p oduc ion
and consump ion pa e ns, bo h wi hin and beyond he egion.
They in oduce en i onmen ally iendly p oduc inno a ions
and g een echnologies, bu also new business models and se -
ices (Abdesselam, Kedja , and Renou‐Maissan 2024; Colom-
belli and Qua a o 2019). New en an s seem o be pa icula ly
in luen ial in he ea ly s ages o an indus y ans o ma ion
(Hocke s and Wüs enhagen 2010), as hey bundle new p oduc s
and se ices in a unique way o gene a e new knowledge and
capabili ies ha no ably de ia e om p e ious knowledge bases.
S a ‐ups can he e o e boos eco‐inno a ion ac i i ies and in-
luence es ablished playe s (Bohnsack, Pinkse, and Kolk 2014;
Dewald and Ach e nbosch 2016; Nilsson and Nyk is 2016).
This is also ue o he anspo sec o , in which s a ‐ups play
a leading ole in he de elopmen o eco‐inno a ion despi e he
abo e‐men ioned s ong egime s uc u es. In he case o elec-
ic mobili y, o example, s a ‐ups seem o ha e de eloped new
p oduc s and business models a ound he same ime as in-
cumben s (Bohnsack, Pinkse, and Kolk 2014; So acool, Noel,
and O sa o 2017; Töd ling, T ippl, and Desch 2022).
Agains his backg ound, concep ual and empi ical esea ch on
knowledge spillo e s shows ha s a ‐ups bene i g ea ly om
unused and non‐comme cialized ideas, knowledge and ne wo k
esou ces o es ablished ac o s (e.g., Cojoianu e al. 2020;
Colombelli and Qua a o 2019; Doblinge , Su ana, and Ana-
don 2019; Kli kou and Coenen 2013). As such, bo h ex a‐ and
in a‐ egional linkages a e seen o posi i ely a ec g een s a ‐
up eme gence. Al hough he impo ance o exis ing s uc u es
a ies g ea ly be ween sec o s (Dewald and Ach e nbosch 2016),
geog aphe s in he ield o egional s udies and ansi ions
esea ch a gue ha en ep eneu ial ac i i ies o s a ‐ups and
spin‐o s a e an essen ial mechanism o egional b anching in o
ela ed and new indus ies (Boschma e al. 2017; MacKinnon
e al. 2019). G een indus ial de elopmen in pa icula bene i s
om eme ging inno a ion ac o s, as hey a e especially capable
o de eloping skills and knowledge combina ions beyond
egion‐in e nal pa h dependencies. The p ocess o g een pa h
c ea ion is he e o e mo e likely o occu in egions whe e he
suppo s uc u es and he exis ing skill base a e well‐de eloped
(Gibbs and Jensen 2022; Töd ling, T ippl, and F angen-
heim 2020; T ippl e al. 2020). Based on he abo e, we de i e he
i s esea ch ques ion as ollows:
3 o 15
RQ1: A e hose egions ha display a specializa ion in g een
s a ‐ups also mo e specialized in g een incumben s?
S a ‐ups a e pa icula ly dependen on he ex e nal a ailabili y
o knowledge and ne wo ks, no only due o he g ea e
complexi y o en i onmen al inno a ions bu also because o
limi ed in e nal esou ces (Ho bach 2020). Agains his back-
g ound, a la ge numbe o ecen s udies ha e examined
egional de e minan s o g een s a ‐up eme gence. Fi s , i has
been shown ha he knowledge accumula ed in a egion is
posi i ely associa ed wi h he eme gence o g een s a ‐ups (e.g.,
Coll‐Ma ínez, Malia, and Renou‐Maissan 2022b; Colombelli
and Qua a o 2019; Ho bach 2020). As new ac o s ex ensi ely
build on p e‐exis ing s uc u es, a di e en ia ed bu comple-
men a y knowledge pool is pa icula ly c ucial (Giudici, Gue-
ini, and Rossi‐Lamas a 2019; Vedula, Yo k, and Co be 2019).
Likewise, Cojoianu e al. (2020) and Coll‐Ma ínez, Jo é‐Llopis,
and Te uel (2022a,2022b) epo a posi i e co ela ion be ween
egional human capi al endowmen as well as he p esence o
esea ch ins i u ions and he eme gence o new eco‐inno a ion
en an s. Looking a he echnological dimension, i is in e -
es ing o no e ha spillo e s om bo h g een and non‐g een
knowledge c ea ion seem o in luence g een egional s a ‐up
ac i i ies (Cojoianu e al. 2020; Colombelli and Qua a o 2019).
In addi ion o skills and echnological capabili ies, egula ion
and ( egional) policies in luence g een s a ‐up ac i i ies. Fo
example, by di ec ly p omo ing eco‐inno a i e and en i on-
men ally iendly s a ‐ups, nega i ely impac ing non‐g een
s a ‐ups o c ea ing demand incen i es, he eby s eng hening
ma ke ce ain y o g een s a ‐ups (Giudici, Gue ini, and
Rossi‐Lamas a 2019; Hoogendoo n, an de Zwan, and
Thu ik 2020; Ho bach 2020). In line wi h hese indings, Bio e ,
Dechezlep ê e, and Fadic (2021) show ha he majo i y o
en u e capi al has been in es ed in g een anspo s a ‐ups in
ecen yea s. Mo eo e , nume ous s udies show ha egions'
in o mal ins i u ional amewo k condi ions a e c ucial. As
such, social no ms, sha ed meanings, beha io al pa e ns and
en i onmen al awa eness in luence i m o ma ions and g een
en ep eneu ship a he egional le el. On he one hand, hese
can lend addi ional legi imacy o en i onmen ally iendly
echnologies and p ac ices, and on he o he hand, hey can
simpli y oppo uni y ecogni ion. Bo h ac o s inc ease egional
eco‐inno a i e i m en y (e.g., Cojoianu e al. 2020; Töd ling,
T ippl, and F angenheim 2020; Vedula, Yo k, and Co -
be 2019). Gi en he a ie y o egional and sec o al speci ici ies
ha in luence eco‐inno a ion ac i i ies, we o mula e he sec-
ond esea ch ques ion as ollows:
RQ2: Which ex e nal de e minan s help o explain (di e en
ypes o ) egional specializa ions in g een incumben s and g een
s a ‐ups?
3
|
Me hodology
To iden i y and explain ac i i ies o eco‐inno a ion incumben s
and s a ‐ups a he egional le el, we use wo di e en da a
sou ces. We use pa en da a o p oxy eco‐inno a ion ac i i ies
o incumben s. O he company‐ ela ed pa en s iled a he
Eu opean Pa en O ice (EPO), he majo i y a e held by la ge
co po a e en i ies; in he ield o anspo e en mo e han 95%
(Eu os a 2014). Acco dingly, we can use pa en applica ions as
a e y di ec p oxy o in en i e ac i i ies d i en by incumben s.
In con as , ac i i ies o new en ep eneu ial ma ke en an s
can be measu ed di ec ly by he numbe o s a ‐ups. Gi en he
no el y o ac i i ies in he domain o g een anspo , a la ge
sha e o new i ms' ac i i ies can hus be conside ed inno a i e
(e.g., Cojoianu e al. 2020; Vedula, Yo k, and Co be 2019).
Using bo h egionalized pa en and s a ‐up da a, we hen de i e
me hodological app oaches o analyze he egional p e alence
o g een incumben s and g een s a ‐ups.
3.1
|
Pa en Da a
Pa en s a e a key measu e in inno a ion esea ch and ela ed
ields o s udy he eme gence and di usion o knowledge and
echnological in en ions. The use o pa en da a is accompanied
by limi a ions, such as he p e ailing ocus on echnologies, he
low in o ma i e alue wi h ega d o he quali y and impac o
in en ions and he possible non‐pa en abili y he eo . Howe e ,
pa en s a e pa icula ly aluable o geog aphical analyses
because hey allow o ace he p oduc ion o ( echnological)
knowledge in a comp ehensi e and spa ially nuanced way
(G iliches 1990; an den Be ge, We e ings, and Alkemade 2020).
This s udy d aws on he PATSTAT da abase o he Eu opean
Pa en O ice (EPO). In a i s s ep, we collec all pa en appli-
ca ions wi h p io i y da es om 2009 o 2018 and assign hem
egionally using he in en o s' add esses. Like an den Be ge,
We e ings, and Alkemade (2020), we use whole coun s in he
case o mul iple in en o s, hence we conside knowledge o be a
non‐di isible good. Since an o e ly small‐scale app oach would
isk ha many egions ha e no o only a small olume o pa -
en s, we eso o he 96 Ge man labo ma ke egions. These
ep esen unc ional e i o ial uni s
1
and a e gene ally
composed o se e al dis ic s (“K eise”/NUTS3 egions). A he
same ime, his egional se ing allows annual socioeconomic
da a o be agg ega ed om he municipal o dis ic le el,
whe eas o smalle e i o ial uni s his da a is o en missing,
pa icula ly sec o al in o ma ion o adequa ely es ima e inno-
a ion ac i i ies.
In a second s ep, we aim o iden i y all pa en s om he ans-
po sec o as well as hose classi ied as g een anspo ech-
nologies. Fo he o me , we d aw on he WIPO echnology
conco dance, which links codes o he In e na ional Pa en
Classi ica ion (IPC) o echnology ields (Schmoch 2008). To
iden i y g een anspo echnologies, we ely on he Y02 class o
he Coope a i e Pa en Classi ica ion (CPC), an ex ension o he
IPC (see Table A1 o a de ailed o e iew o da a sou ces and
sea ch s a egies). The subg oup Y02 T con ains pa en s ela ed
o clima e change mi iga ion echnologies in he anspo
sec o , co e ing oad, ail, ai and ma i ime/wa e ways ans-
po echnologies as well as enabling echnologies such as
cha ging o elec ic ehicles, and uel cells (Coll‐Ma ínez,
Malia, and Renou‐Maissan 2022b; Haščič and Migo o 2015;
an den Be ge, We e ings, and Alkemade 2020). Thus, o he
10‐yea obse a ion pe iod, we iden i ied 46,228 pa en s in he
4 o 15 G ow h and Change, 2025
echnology ield o anspo and a ound 15,000 pa en s a e
classi ied as g een anspo echnologies.
3.2
|
S a ‐Up Da a
The da a o iden i y (g een) s a ‐up a e aken om he Man-
nheim En e p ise Panel (MUP) which is gene a ed by he ZEW–
Cen e o Eu opean Economic Resea ch since 1992. By coop-
e a ing wi h he la ges Ge man c edi a ing agency “C ed-
i e o m e.V.,” which con ibu es da a wice a yea on he o al
Ge man co po a e landscape, he MUP p o ides a comp ehen-
si e mic o da abase on legally independen companies. As a
esul , pa en companies and subsidia ies a e epo ed sepa-
a ely. In o al, i con ains de ailed in o ma ion on a ound 9
million economically ac i e and closed companies (Be sch
e al. 2014).
Unlike esea ch ha iden i ies and examines g een s a ‐up
ac i i ies ia su ey da a (e.g., Hoogendoo n, an de Zwan,
and Thu ik 2020; Ho bach 2020), he MUP, wi h a co e age o
abou 90% o he ull s ock o i ms,
2
allows o mo e ep e-
sen a i e s udies o business dynamics o e longe obse a ion
pe iods and small‐scale disagg ega ion (Be sch e al. 2014).
Al hough he me e numbe o s a ‐ups does no necessa ily
e lec hei quali y o impac (Cojoianu e al. 2020), he mea-
su e in ac o e s a sui able way o co e ing en ep eneu ship
ac i i ies o a wide ange o economic ac o s a he egional
le el.
O pa icula impo ance o he analyses in his pape is he
in o ma ion on he company add ess o he spa ial alloca ion,
he da e o he ounda ion o he empo al limi a ion, and he
classi ica ion o economic ac i i ies (NACE Re . 2). The la e is
used o exclude se ice‐o ien ed and p edominan ly non‐
inno a i e sec o s and o delinea e he anspo sec o (see
Table A1). Again, we use he yea s 2009–2018 as he obse a ion
pe iod and he 96 Ge man labo ma ke egions as he uni s o
analysis (c . chap e 3.1).
G een s a ‐ups in he anspo sec o canno be iden i ied by
sec o assignmen s. Ins ead, we sc een he company de-
sc ip ions con ained in he MUP using a keywo d‐based sea ch
s a egy, ollowing p e ious s udies wi h a simila app oach
(Hansmeie and Losacke 2024; Shapi a e al. 2014). In o de o
limi he bias in he iden i ica ion o s a ‐ups, which is also due
o he ac ha many newly bo n (mic o) en e p ises and s a ‐
ups ha e imp ecise company desc ip ions, we ollow a b oad
sea ch app oach. We de i e he echnology and sec o ‐speci ic
sea ch e ms o g een anspo s a ‐ups om a ious da a-
bases and sou ces: he Y02 T pa en class, anspo s a ‐ups
lis ed in he “S a G een” ne wo k
3
and he wo k by Cojoianu
e al. (2020) (Table A 1). To check he alidi y o he iden i ied
companies, we sc eened hei b ie desc ip ion and conduc ed
in e ne sea ches whe e necessa y. This s ep elimina ed hose
ha a e no applicable (e.g., companies p ima ily ac i e in he
ene gy sec o ). On he o he hand, we add hose g een anspo
ela ed s a ‐ups ha we e no ini ially ound, bu appea in he
“S a G een” ne wo k o ecei ed unding by he Ge man
Fede al En i onmen al Founda ion (DBU).
4
O e all, o he
42,000 s a ‐ups ela ed o he anspo sec o , abou 900 can be
classi ied as eco‐inno a i e.
3.3
|
Regional Pa en and S a ‐Up Specializa ions
Compa ing egions acco ding o hei ela i e p og ess in g een
incumben s and g een s a ‐ups is he main objec i e o his
pape . Technically, howe e , pu e sha es a e o en di icul o
compa e as hei a e ages and a iance di e be ween domains.
I will also be impo an o he analysis o de e mine whe he
eco‐inno a ion ac i i ies o incumben s and s a ‐ups in egions
a e abo e o below he na ional a e age. An es ablished and
common p ac ice in (economic) geog aphy esea ch o achie e
his double objec i e is o measu e specializa ion based on he
loca ion quo ien (LQ). The LQ allows o de e mine whe he a
egion is echnologically o economically specialized compa ed
o all o he egions o e a gi en pe iod o ime. Usually, LQ >1
indica es abo e‐a e age specializa ion, LQ =1 a e age special-
iza ion and LQ <1 below a e age specializa ion. To minimize
dis o ions by ou lie s, we no malize he specializa ion in-
dica o s by means o a log ans o ma ion be ween þ100 and
−100, wi h posi i e alues indica ing abo e‐a e age egional
specializa ions (Losacke and Lie ne 2020). Simila o he
seminal wo k o Balland and Rigby (2017) who d aw on ela i e
echnology ad an age measu es, we will e e o he no malized
LQs as he ela i e pa en ad an age (RPA) and ela i e s a ‐up
ad an age (RSA) (see Equa ions 1a and 1b). Fo each egion in
he g een anspo domain g he indica o s a e gi en by:
RPA g =100 × anhln [p g/∑ pg
p /∑ p ](1a)
RSA g =100 × anhln [s g/∑ sg
s /∑ s ](1b)
Bo h specializa ion measu es a e calcula ed as he ela ion o
g een anspo ac i i ies o pa en po s a ‐up ac i i ies sin he
anspo sec o . This allows o con ol o app op ia e base-
lines, as egional specializa ions in en i onmen ally iendly
inno a ion ac i i ies migh depend on s ong ac i i ies in he
o e a ching anspo sec o .
4
|
Regional Specializa ions in G een T anspo
Incumben s and S a ‐Ups
Agains he backg ound o his pape 's co e empi ical p ospec ,
we a e in e es ed in iden i ying egional specializa ions in no el
and es ablished eco‐inno a o s. I espec i e o he absolu e
inno a ion capaci y, we seek o de e mine whe he he cu en
sha e o g een ac o ac i i y in a egion's anspo sec o is
abo e o below na ional a e age. By analyzing his si ua ion
sepa a ely o bo h incumben s and s a ‐ups, i is possible o
dis inguish be ween di e en ypes o egional eco‐inno a ion
ac o specializa ions.
5 o 15
4.1
|
Typology o Regions
The compa ison o he ela i e pa en and s a ‐up ad an age
enables us o c ea e a ypology o he 96 Ge man spa ial plan-
ning egions. The RPA and RSA a e de e mined o each egion
o he en i e pe iod om 2009 o 2018, allowing o depic ou
di e en ypes o egional eco‐inno a ion ac o specializa ion in
he anspo sec o (Figu e 1).
1. Ho spo : Region wi h bo h abo e‐a e age g een s a ‐up
and incumben specializa ion.
2. S a ‐up‐d i en: Region wi h abo e‐a e age g een s a ‐up
specializa ion bu below‐a e age g een incumben
specializa ion.
3. Incumben ‐d i en: Region wi h abo e‐a e age g een
incumben specializa ion bu below‐a e age g een s a ‐up
specializa ion.
4. Lagga d: Region wi h bo h below‐a e age g een s a ‐up
and incumben specializa ion.
When looking a he dis ibu ion o egions by ype, i is
no iceable ha 35 o he 96 egions ha e nei he a s a ‐up no
an incumben specializa ion. These so‐called lagga d egions
ep esen he la ges sha e (app oxima ely 37%), ollowed by 29
egions in which he g eening in he anspo sec o is d i en o
an abo e‐a e age ex en by s a ‐ups bu less by es ablished
ac o s. Con e sely, 18 egions a e incumben ‐d i en and a he
same ime ha e a below‐a e age g een s a ‐up specializa ion.
Only 14 egions, ha is a mino i y o oughly 15%, can be
ca ego ized as g een anspo ho spo s, ea u ing specializa ion
in bo h g een s a ‐ups and g een incumben s.
O e all, his ypology e eals a ela i ely he e ogeneous and
ambiguous pic u e o egional eco‐inno a ion ac o specializa-
ions in he anspo sec o in Ge many. Answe ing esea ch
ques ion 1, his ini ial inding is ema kable in ha i nei he
sugges s ha g een anspo s a ‐ups and incumben s mainly
eme ge in spa ial p oximi y due o he impo ance o knowledge
spillo e s and echnological specializa ion, no does i seem o
suppo ha eco‐inno a ions p ima ily eme ge ou side es ab-
lished s uc u es, as concep ualized in ansi ion s udies. In he
FIGURE 1 |Regional ypology acco ding o eco‐inno a ion ac o specializa ions in he anspo sec o in Ge many (2009–2018).
6 o 15 G ow h and Change, 2025
la e case, we would expec a mo e nega i e co ela ion be ween
incumben and s a ‐up specializa ion, while a co‐occu ence o
s a ‐ups and incumben s would show a posi i e co ela ion be-
ween RPA and RSA, ha is ha egions all p ima ily in o ei he
he lagga d o he ho spo ca ego y. Ra he , ou desc ip i e esul s
co obo a e he indings o ecen wo k. No only does he
impo ance o speci ic eco‐inno a o s a y be ween egions (e.g.,
Bugge, Ande sen, and S een 2022; T ippl e al. 2020), bu i is also
e iden ha s a ‐up and incumben ac i i ies ake place whe e
he espec i e o he is less dominan , indica ing he in luence o
o he /non‐g een egional de e minan s (e.g., Colombelli and
Qua a o 2019; an den Be ge, We e ings, and Alkemade 2020).
We do no obse e ha g een anspo s a ‐ups a e mo e likely o
eme ge in egions ha a e also specialized in eco‐inno a ion
anspo incumben s. Only a ew egions ea u e a specializa-
ion in bo h g een incumben s and g een s a ‐ups.
Fu he mo e, Figu e 2shows ha he di e en egional spe-
cializa ions o eco‐inno a ion ac o s can be ound ac oss Ge -
many. A he same ime, ou esul s indica e ha ce ain spa ial
clus e s o he speci ic ypes o egions eme ge. G een anspo
ho spo s and s a ‐up‐d i en egions p edomina e in he
sou he n and eas e n pa s o he coun y. Consis en ly, lagging
egions clus e in he cen e and (no h) wes o Ge many, while
incumben ‐d i en egions seem o be e enly dis ibu ed. In e -
es ingly, impo an loca ions o he au omo i e indus y, such as
Wol sbu g, Munich, Ingols ad o S u ga , a e mainly s a ‐up
d i en o e en egional ho spo s o eco‐inno a i e anspo
ac i i ies. This inding sugges s ha sec o al g eening is also
aking place in hose egions ha a e cha ac e ized by s ong
indus ial egime s uc u es. The ac ha ho spo s a ely appea
in isola ion om s a ‐up o incumben egions could also
indica e in e ‐ egional spillo e e ec s (Guo e al. 2020).
Al hough some spa sely popula ed egions a e conside ed ac o
ho spo s, he geog aphical dis ibu ion seems o co ela e wi h
he popula ion densi y in gene al. Acco dingly, none o he
depic ed la ge ci ies has a below‐a e age specializa ion wi h
bo h g oups o ac o s.
In o de o be e compa e he egions, Table 1p o ides
desc ip i e s a is ics o se e al socio‐economic indica o s. In
line wi h p e ious s udies, we e e o key egional ac o s ha
ha e been shown o in luence (g een) inno a ion ac i i ies,
pa icula ly in he anspo sec o . These include, o example,
he capi al s ock (GDP pe capi a), he popula ion densi y o
measu e agglome a ion ex e nali ies and he a ailabili y o high‐
skilled indi iduals measu ed as he sha e o employees wi h an
academic backg ound, ha is e ia y educa ion (e.g. Cojoianu
e al. 2020; Coll‐Ma ínez, Malia, and Renou‐Maissan 2022b;
Co adini 2019). As a sec o a iable, we also conside he
egional mo o iza ion a e, which is calcula ed om he numbe
o ca s pe 1000 inhabi an s.
In ac , he a o emen ioned inding seems o be con i med, ac-
co ding o which egional ho spo s o eco‐inno a o s ela ed o
anspo end o be mo e u ban, ha is ha e on a e age a
highe popula ion densi y. Mo eo e , hese egions ha e he
compa a i ely highes GDP pe capi a, he la ges sha e o
FIGURE 2 |Region ypes o eco‐inno a ion ac o specializa ions in he anspo sec o in Ge many (2009–2018).
7 o 15
highly quali ied people and he ewes ca s pe inhabi an . These
co ela ions a e almos consis en ac oss he di e en ypes o
egions, wi h he excep ion o s a ‐up d i en egions ha ha e
on a e age ewe highly quali ied people han incumben e-
gions and bo h he lowes GDP pe capi a and popula ion
densi y. While hese indings di e o some ex en om p e i-
ous s udies in which s a ‐ups appea o be pa icula ly depen-
den on quali ied pe sonnel and p o i om agglome a ion
e ec s (e.g. Giudici, Gue ini, and Rossi‐Lamas a 2019; Hoo-
gendoo n, an de Zwan, and Thu ik 2020; Ho bach 2020), i
should be no ed ha in his s udy we conside all i m bi hs as
s a ‐ups wi hou explici ly con olling o hei inno a i eness.
4.2
|
Explana ion o Regional Di e ences
To alida e he desc ip i e esul s and add ess ou second
esea ch ques ion, we make use o econome ic models. To his
end, we p edic he egional classi ica ion in o he p oposed
ypology (see Figu e 1). In a i s s ep, he ocus is on es ima ing
he p obabili y ha a egion is an eco‐inno a ion ho spo in he
anspo sec o . Fo his objec i e, a logis ic eg ession model
wi h obus s anda d e o s by egions is es ima ed, whe e he
dependen a iable is dicho omous and akes he alue 1 i a
egion is specialized in o bo h eco‐inno a ion incumben s and
s a ‐ups (RPA & RSA >0), o he wise i akes he alue 0. In
o de o a oid biases d i en by li le inno a ion ac i i ies and
ou lie s, we ha e summed he s a ‐up and pa en coun s o he
co esponding yea wi h he wo p e ious yea s (mo ing win-
dow), esul ing in a panel da a se o he dependen a iable
co e ing he pe iod 2011–2018. In addi ion, we use a one‐yea
ime lag
5
o he independen a iables, which hus co e he
yea s 2010–2017. The logis ic eg ession model is gi en by:
logi (Type , )=α+β1Popula ion densi y , −1
+β2GDP pe capi a , −1
+β3High ‐ skilled indi iduals , −1
+β4Ca indus y si e , −1
+β5La ge en e p ises , −1
+β6R&D ‐ in ensi e indus ies , −1
+β7Public anspo ,2020
+β8G een o es , −1+θ +ε
(2)
whe e he subsc ip s and deno e he egion and ime pe iod,
θ
is he dummy a iable o each ime window ( ime ixed
e ec ) and εis he e o e m. We include a numbe o common
egional‐le el ac o s (see also Table 1): popula ion densi y, GDP
pe capi a and sha e o high‐skilled indi iduals. Mo eo e , we
include a egion's sha e o la ge en e p ises o p oxy he in lu-
ence o mul i‐na ional companies and a dummy a iable indi-
ca ing main loca ions o he au omo i e indus y (headqua e
o ca manu ac u e s wi h mo e han 10,000 employees). The
egional sha e o employees in esea ch‐in ensi e indus ies
e lec s echnological capabili ies, while a u he dummy a i-
able indica es whe he he dis ance o he nea es public
anspo s op is abo e o below Ge man a e age ( alues o
2020). The egional sha es o o es o he poli ical pa y “ he
G eens” in he 2009, 2013 and 2017 ede al elec ions se e as a
p oxy o ecological ideals and li es yles (e.g. Coll‐Ma ínez,
Malia, and Renou‐Maissan 2022b; Ho bach 2020).
6
The
de ailed desc ip ions o he a iables a e included in he annex
(Table A2).
In a second s ep, we employ a mul inomial logi model,
conside ing all ou ypes o egions in he dependen a iable.
Tha is, he mul inomial model uses he g oup o lagga d e-
gions as a baseline and es ima es he p obabili y ha a egion
alls in one o he o he g oups as opposed o he e e ence
g oup. This eg ession model akes a simila o m as he logis ic
model desc ibed abo e, excep ha he p obabili y dis ibu ion
o he ou come a iable is mul inomial no binomial.
Table 2p esen s he eg ession esul s, whe e model (1) is he
binomial logis ic model and models (2), (3) and (4) belong o he
mul inomial logis ic eg ession. Likelihood a io es s indica e
ha all models a e signi ican ly be e a explaining di e ences
among he g oups han in e cep ‐only models. The models do
no su e om mul icollinea i y as indica ed by co ela ions
(see Table A2) and a iance in la ion ac o s (<5). We a e awa e
ha he e migh be mino p oblems wi h endogenei y o some
a iables, which a e add essed by he use o ime‐lags and he
implemen ed panel s uc u e. Howe e , despi e hei obus ness
o he use o di e en ime lags, he eg ession esul s canno be
seen as e lec ing causali y in he s ic sense. In e ms o he
esea ch ques ion, howe e , hey can be seen as con ibu ions
and explana ions a he han me e co ela ions. In doing so, we
ollow he common p ac ice in he geog aphical li e a u e o
using analy ical s a is ics o pa ially explain e ec s ha may be
complex and di icul o disen angle (see e.g. Ho bach, Ol a,
and Belin 2013). The main esul s p esen ed below a e obus o
se e al changes in he econome ic s a egy. Tha is, esul s hold
when using subse s o he o iginal da a se (e.g. excluding yea s/
TABLE 1 |Mean alues o socio‐economic and sec o al cha ac e is ics by ype o egion.
Type o egion N
Popula ion
pe km
2
GDP pe
capi a (EUR)
High‐skilled
indi iduals (%)
Ca s pe 1000
inhabi an s
Ho spo 14 325 37,941 17.2 496
S a ‐up‐d i en 18 180 32,146 11.5 543
Incumben ‐
d i en
29 285 35,194 13.5 514
Lagga d 35 201 33,467 11.0 551
To al 96 228 34,510 13.0 529
Sou ce: Calcula ions based on da a p o ided by Fede al Ins i u e o Resea ch on Building, U ban A ai s and Spa ial De elopmen (2022).
8 o 15 G ow h and Change, 2025
TABLE A2 |Desc ip i e s a is ics and co ela ions o a iables.
Va iable Desc ip ion Mean SD Min. Max. 1 2 3 4 5 6 7 8
1 Popula ion
densi y
Popula ion pe km
2
329 502 42 4055 1
2 GDP pe
capi a
G oss domes ic p oduc (GDP)
pe capi a (in EUR)
32,676 8220 17,935 66,429 0.35*** 1
3 High‐
skilled
indi iduals
Sha e o employees wi h an
academic deg ee
11.3 3.8 5.0 28.3 0.49*** 0.56*** 1
4 Ca
indus y
si e
Loca ion o a ca manu ac u e
wi h mo e han 10,000
employees (1 i yes)
0.1 0 1 0.10*** 0.37*** 0.31*** 1
5 La ge
en e p ises
Sha e o companies wi h mo e
han 250 employees
0.3 0.1 0.1 0.6 0.32*** 0.55*** 0.35*** 0.15*** 1
6 R&D‐
in ensi e
indus ies
Sha e o employees wo king in
knowledge‐ o esea ch‐
in ensi e indus ies
10.2 5.5 0.8 28.3 −0.11*** 0.43*** −0.06** 0.25*** 0.23*** 1
7 Public
anspo
Dis ance o he nea es public
anspo s op wi h a leas 20
depa u es pe day in 2020 (1 i
below median alue)
0.5 0 1 0.40*** 0.32*** 0.53*** 0.24*** 0.38*** 0.10*** 1
8 G een
o es
Sha e o o es o “ he g eens”
in he 2009, 2013, 2017 ede al
elec ions
8.4 3.1 2.8 17.4 0.38*** 0.49*** 0.31*** 0.19*** 0.16*** 0.25*** 0.37*** 1
**p<0.05; ***p<0.01.
15 o 15
