Yu, Lina; Guo, Zenghui; Ning, Jing
A icle
Fo eign mul ina ional global alue chain pa icipa ion and
domes ic inno a ion: Viewing GVCs inside-ou and ou side-
in
Jou nal o Inno a ion & Knowledge (JIK)
P o ided in Coope a ion wi h:
Else ie
Sugges ed Ci a ion: Yu, Lina; Guo, Zenghui; Ning, Jing (2024) : Fo eign mul ina ional global alue
chain pa icipa ion and domes ic inno a ion: Viewing GVCs inside-ou and ou side-in, Jou nal o
Inno a ion & Knowledge (JIK), ISSN 2444-569X, Else ie , Ams e dam, Vol. 9, Iss. 4, pp. 1-11,
h ps://doi.o g/10.1016/j.jik.2024.100603
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/327505
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/
Fo eign mul ina ional global alue chain pa icipa ion and domes ic
inno a ion: Viewing GVCs inside-ou and ou side-in
Lina Yu
a
, Zenghui Guo
a
, Jing Ning
b,*
a
School o Economics, Ocean Uni e si y o China, 238 Songling Road, Laoshan Dis ic , Qingdao, Shandong P o ince, 266100, China
b
School o Managemen , Ocean Uni e si y o China, 238 Songling Road, Laoshan Dis ic , Qingdao, Shandong P o ince, 266100, China
ARTICLE INFO
JEL classi ica ion:
O31
O32
O33
Keywo ds:
MNEs
Domes ic i ms
Inno a ion pe o mance
GVC–OLI nexus
GVC complexi y
GVC pa icipa ion
ABSTRACT
Two ques ions pe sis ega ding global alue chains (GVCs): Fi s , wha a e he s a egies and pa hways h ough
which mul ina ional en e p ises (MNEs) in eg a e and na iga e GVCs? Second, wha a e he cha ac e is ics o
GVCs in which MNEs pa icipa e? An analysis o hese ques ions leads o inside–ou and ou side–in pe spec i es
on GVCs. The li e a u e on MNEs mos ly ocuses on he quan i y and quali y o o eign di ec in es men a he
han on he di e ences in MNEs’p oduc ion ac i i ies and s uc u e o GVCs. This s udy examines he e ec s o
MNEs’pa icipa ion in simple and complex GVCs on domes ic inno a ion. We ind ha he posi i e e ec s a e
mainly d i en by MNEs’downs eam pa icipa ion in simple GVCs. The mechanism es s indica e ha MNEs’
ups eamness wi hin simple GVCs nega i ely a ec s domes ic i ms’abso p ion capaci y and in e nal R&D and
ha wi hin simple and complex GVCs nega i ely a ec s domes ic i ms’expo esilience.
In oduc ion
To unde s and he na u e o global alue chains (GVCs), i is c ucial
o ega d i ms, a he han coun ies o indus ies, as he p ima y
pa icipan s (An `
as, 2020). Acco ding o Cigna e al. (2022), mul ina-
ional en e p ises (MNEs) play a dominan ole in global p oduc ion and
GVC- ela ed ade, ep esen ing o e one- hi d o he o al ou pu , 65%
o expo s, and 41% o in e media e inpu impo s. MNEs acili a e he
in eg a ion o de eloping economies in o global ne wo ks o agmen ed
p oduc ion and collabo a i e inno a ion (Lee &Ge e i, 2021;Mody,
2004;Pie obelli &Rabello i, 2011), embedding local business eco-
sys ems and global di isions o labo (Sa anga e al., 2019; an Mee e en
&Kleibe , 2022).
Al hough MNEs ep esen he la ges sou ce o echnological inno-
a ion and ans e (Iamma ino &McCann, 2013), he Po e pa adox
asse s ha endu ing compe i i e ad an ages in a globalized e a o igi-
na e om local hings (Po e , 1998), impelling he analysis o he
economic consequences o MNEs’GVC- ela ed ac i i ies on hos coun-
ies’inno a ion pe o mance. Howe e , p e ious esea ch on how
MNEs a ec domes ic inno a ion ocuses mainly on he quan i y and
quali y o o eign di ec in es men (FDI), o example, he o eign eq-
ui y sha es in he ho izon al, ups eam, and downs eam sec o s
(Vujano i´
c e al., 2022) o o eign i ms’pa en applica ions and pa en
g an s (Tan e al., 2023). When he quan i y and quali y o FDI e lec a
coun y’s a ac i eness o FDI, hey canno accu a ely desc ibe he
di e ences in he echnological in ensi y and ma ke s a egies
embodied in speci ic p oduc ion ac i i ies. Thus, he esea ch ocus mus
shi o he undamen al na u e o MNEs’p oduc ion ac i i ies ha ca e
o he local ma ke s o hos coun ies and se e as he c i ical elemen s o
he global p oduc ion sequence. Speci ically, MNEs no only sell he
same p oduc s in he same way wo ldwide (Le i , 1983) bu also
con ain a conside able po ion o he p oduc ion ne wo ks spanning
mul iple coun ies (Pandya, 2016). Mo eo e , om a s uc u al
pe spec i e on MNEs’global p oduc ion, he ex an li e a u e cen e s on
he GVC go e nance heo y wi h a ocus on mic o-in e i m coo dina-
ion (Bai , 2008) while igno ing he b oade ne wo k s uc u e o GVCs
(Ambos e al., 2021). As GVC complexi y is a basic ne wo k s uc u al
ea u e ha e lec s he specializa ion pa e ns o coun ies and in-
dus ies (Wang e al., 2017), we analyze he pa icipa ion o MNEs in
simple and complex GVCs and hei impac on domes ic inno a ion. Two
ques ions d aw ou a en ion: Wha s a egies and pa hways do MNEs
use o in eg a e and na iga e GVCs? Wha a e he cha ac e is ics o he
GVCs ha engage MNEs? Thus, he analysis o hese wo issues e ec-
ua es he inside–ou and ou side–in pe spec i es in GVCs.
* Co esponding au ho .
E-mail add esses: [email p o ec ed] (L. Yu), [email p o ec ed] (Z. Guo), [email p o ec ed] (J. Ning).
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.2024.100603
Recei ed 15 July 2024; Accep ed 17 Oc obe 2024
Jou nal o Inno a ion & Knowledge 9 (2024) 100603
A ailable online 2 No embe 2024
2444-569X/© 2024 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/ ).
Based on he abo e analysis, we ill his esea ch gap by in es iga ing
he speci ic impac o MNEs’pa icipa ion in simple and complex GVCs
on domes ic i ms’inno a ion pe o mance. Addi ionally, we analyze
how domes ic i ms’he e ogenei y a ec s he impac o MNEs’pa ici-
pa ion in GVCs on domes ic i ms’inno a ion pe o mance, including
ac o s such as domes ic i ms’R&D in ensi y and e ealed compa a i e
ad an age (RCA). Finally, we analyze he heo e ical mechanisms,
including domes ic i ms’in e nal R&D, abso p ion capaci y, and expo
esilience.
This s udy con ibu es o exis ing li e a u e in se e al ways. Fi s , i
p o ides an al e na i e pe spec i e o in es iga ing inside–ou ( i m- o-
ne wo k) and ou side–in (ne wo k- o- i m) GVCs. McWilliam e al.
(2020) and Pananond e al. (2020) highligh his combina ion o com-
plemen a y pe spec i es. They in eg a e GVC go e nance heo y wi h
he owne ship, loca ion, and in e naliza ion ad an age (OLI) pa adigm
and MNEs’global s a egy. We in e p e how MNEs unde ake p oduc-
ion ac i i ies in speci ic GVC s uc u es and how hese s uc u es a ec
he in e nal o ganiza ion and ex e nal ela ionships o MNEs, which
ep esen inside–ou and ou side–in pe spec i es, espec i ely. Speci -
ically, his s udy empi ically analyzes he complexi y o in e coun y
indus ial connec ions and MNEs’GVC pa icipa ion. Second, i con-
ibu es o he li e a u e on FDI. The exis ing li e a u e ocuses on how
MNEs en e hos coun ies h ough FDI and unde es ima es he GVC
p oduc ion ac i i ies o ganized and domina ed by MNEs in hos econ-
omies (Ha ding &Ja o cik, 2012;C escenzi e al., 2015;Lin &Kwan,
2016;Lu e al., 2017;Amendolagine e al., 2019). This s udy cla i ies he
mechanisms h ough which FDI gene a es posi i e spillo e s o de el-
oping coun ies h ough GVC pa icipa ion. Finally, i con ibu es o
li e a u e on i m inno a ion and indus ial upg ades. As he in e ac i e
and open na u e o inno a ion is widely ecognized (Ambos e al., 2021;
Buciuni &Pisano, 2021), how inno a ion p og esses along he speci ic
pa hs o GVC pa icipa ion and s uc u es o GVCs mus be u he
explo ed.
The emainde o his s udy is o ganized as ollows. Sec ion 2 p o-
ides he heo e ical backg ound and in oduces he esea ch hypo he-
ses. Sec ion 3 p esen s he empi ical me hodology, and Sec ion 4 epo s
he empi ical esul s o he baseline eg ession, obus ness, and he -
e ogenei y es s, and mechanism iden i ica ion. Sec ion 5 p esen s he
discussion. Finally, sec ion 6 p esen s he conclusions and limi a ions o
his s udy.
Theo e ical backg ound and hypo hesis de elopmen
The ela ionship be ween GVC pa icipa ion and domes ic inno a-
ion is widely ecognized. F om an in e nal pe spec i e, ex an li e a u e
mainly analyzes he pa icipa ion and posi ions o domes ic i ms wi hin
GVCs (Elshaa awy &Ezza , 2023), which a ec he capabili y o alue
c ea ion and access o high-quali y in e media es. F om an ex e nal
pe spec i e, echnology spillo e s a e examined h ough bila e al in-
e na ional ade (Coe &Helpman, 1995) and indi ec spillo e s o o he
i ms, indus ies, and coun ies ha ade wi h di ec pa ne s (I o e al.,
2023). F om he la e pe spec i e, as MNEs exploi and abso b new
ideas and echnologies ac oss na ional bo de s, hei pa icipa ion in
GVCs gene a es ne wo k ex e nali ies o local inno a ions. Acco ding o
An `
as (2020), a GVC e e s o a se ies o s ages engaged in he p o-
duc ion o a p oduc o se ice des ined o inal consump ion, wi h each
p oduc ion s age adding alue and a leas wo di e en coun ies a e
in ol ed in p oduc ion. This anscends he GVC go e nance s uc u e
and local p oduc ion ac i i ies in which MNEs collabo a e wi h domes ic
i ms. In his con ex , a comp ehensi e mindse is c i ical o analyzing
he in e ac i e na u e o agmen ed global p oduc ion. Connec ing he
i m- o-ne wo k (inside–ou ) pe spec i e wi h he ne wo k- o- i m
(ou side–in) pe spec i e is he co e o he GVC–OLI nexus (McWilliam
e al., 2020). Fu he mo e, we can conside he pa icipa ion o MNEs
and o e all s uc u e o GVCs, which ep esen inside–ou and ou -
side–in pe spec i es, espec i ely. The adi ional OLI pa adigm is
widely applied o analyze he ex en , pa e n, and geog aphic dispe sion
o MNEs’ o eign alue-adding ac i i ies (McWilliam e al., 2020).
Mo eo e , GVC- ela ed specialized p oduc ion e lec s he alue-adding
p ocess, indica ing an impo an aspec ha ela es o he basic ques-
ions o he OLI pa adigm. GVC complexi y emana es om he
alue-adding ac i i ies o MNEs in GVCs, e lec ing he en i e echno-
logical p ocess, which is segmen ed and dis ibu ed ac oss se e al
coun ies. The e o e, o eign MNEs’p oduc ion ac i i ies in simple and
complex GVCs e eal he di ec ion, ex en , and mechanisms o hei
impac on he inno a ion pe o mance o domes ic i ms.
As he GVC–OLI nexus mus add ess he basic ques ions o he OLI
pa adigm om he pe spec i e o he GVC s uc u e, his s udy examines
i by ocusing on he in e naliza ion decisions o MNEs in simple and
complex GVCs. When he s uc u e and p ocesses o MNEs become
inc easingly complex (And ews e al., 2023), he “I”sub-pa adigm o
OLI is a po en analy ical ool o unde s and he o ganiza ion o in e -
nalized ac i i ies and con ol o ex e nalized ela ionships in complex
p oduc ion ne wo ks (Rugman &Ve beke, 2003;Zeng e al., 2023).
Li e a u e on he in e naliza ion heo y highligh s he end o MNEs
shi ing om e ical in eg a ion o ne wo k o ches a ion. This shi
enables MNEs o cap u e he loca ion- and owne ship-speci ic ad an-
ages o domes ic i ms wi hou di ec equi y con ols (Sco -Kennel &
Ende wick, 2004;Asmussen e al., 2022). As e iciency is an axiom o he
in e naliza ion heo y, MNEs op imize go e nance modes and exploi
posi ion ad an ages o manage o he economic ac o s and maximize
sel -in e es (Beni o e al., 2019). To add ess he challenges associa ed
wi h GVC complexi y, MNEs apply o ganiza ional mechanisms o con-
ol and coo dina e c oss-bo de p oduc ion ac i i ies, including
cen aliza ion, s anda diza ion, socializa ion, and ou pu -o ien ed
mechanisms (Zeng e al., 2023). The e o e, GVC complexi y causes
a ia ions in he in e naliza ion decisions o MNEs o g ea e e iciency
because ansac ion cos s, in o ma ion codi iabili y, and he dis ibu ion
o isks and bene i s a e dis inc , indica ing ha he o e all s uc u al
a ibu es o GVCs a e in ica ely in e wined wi h he in e naliza ion
heo y.
As a complex ne wo k phenomenon, GVC is analyzed om a social
ne wo k pe spec i e. Asso a i i y and homophily a e essen ial o
analyzing he o ma ion and e olu ion o GVC p oduc ion ne wo ks.
Acco ding o Pan e al. (2022), he o me in ol es selec i e linking and
asso a i e mixing. MNEs wi h high connec i i y p io i ize local i ms
wi h abundan connec ions. The la e pe ains o a phenomenon in
which indi idual uni s o m connec ions wi h o he s possessing simila
cha ac e is ics o a endency o di e ences be ween hem o diminish
o e ime a e building connec ions. This may e ec ua e ein o ced
connec ions wi hin simila g oups o MNEs and domes ic i ms and
diminish he ans e o in o ma ion and esou ces be ween di e en
g oups h ough seg ega ion pa e ns (Cu a ini e al., 2009;Jackson
e al., 2017). Speci ically, he po en ial posi i e spillo e s om MNEs’
pa icipa ion in complex GVCs o domes ic i ms a e limi ed and subjec
o se e al cons ain s. Fo example, complex GVCs a e o ganized by
MNEs wi h simila echnological ad an ages, impeding coope a ion
be ween MNEs and domes ic i ms in hos coun ies.
P e ious esea ch has concen a ed on he knowledge and echnol-
ogy spillo e s om MNEs indica ed by he OLI amewo k o in es iga e
he speci ic mechanisms by which MNEs in luence domes ic i ms
(Ascani &Gaglia di, 2020;Paul &Feliciano-Ces e o, 2021;Xiao &Tian,
2023). Fundamen ally, MNEs collec agmen ed ad an ages by ex e -
nalizing ce ain p oduc ion ac i i ies. Fi s , he e icacy o spillo e s o
domes ic i ms is con ingen on hei abso p ion capaci y (Duan e al.,
2021;Jha e al., 2024;Mo alles &Mo eno, 2020), which is in luenced
by ex e nal ac o s and is pa h dependen . Second, in e nal R&D inpu s
a e c i ical o p omo ing inno a ion pe o mance and a e he mos
impo an means o accumula ing unique ad an ages. Owing o he
demons a ion and compe i ion e ec s (Lu e al., 2017;Wu e al., 2023),
in e nal R&D inpu s o domes ic i ms can be augmen ed o subs i u ed
by he p esence o MNEs. Finally, as GVC go e nance ela ionships a e
L. Yu e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100603
2
uns able (Pon e e al., 2019), i is impo an o de e mine whe he MNEs
a ec he esilience o domes ic i ms. MNEs may exhibi a ying de-
g ees o oppo unism ac oss di e se p oduc ion con ex s and a ec do-
mes ic i ms’expo esilience o isk and dis up ion by in eg a ing hem
in o globalized ope a ions. In addi ion, exis ing li e a u e p o es ha
he o e all ne wo k pe spec i e is c ucial o unde s anding indi idual
s a egies and di usion o knowledge and isk (Jackson e al., 2017).
The e o e, his s udy in es iga es how MNEs’pa icipa ion in GVCs a -
ec s domes ic i ms’inno a ion pe o mance h ough in e nal R&D,
abso p ion capaci y, and expo esilience, agains he backd op o GVC
complexi y.
Viewing GVCs inside–ou and ou side–in: E ec s o MNEs’pa icipa ion in
simple and complex GVCs
Following he abo e analysis, we cons uc an a gumen ega ding
he associa ion be ween MNEs’pa icipa ion in GVCs and domes ic
i ms’inno a ion pe o mance. MNEs ein o ce he dep h and
complexi y o he hos economies’in eg a ion in o GVCs h ough hei
inhe en links o p oduc ion ne wo ks. GVC pa icipa ion pa hs indica e
di e en le els o s a egic coupling. In di e se indus ial and com-
me cial con ex s, MNEs coo dina e global p oduc ion, e ec ing asse
complemen a i y be ween hem and domes ic i ms. The ups eam and
downs eam p oduc ion s ages exhibi dis inc alue-c ea ion capabil-
i ies. F om an inside–ou pe spec i e, MNEs’di ec pa icipa ion in
ups eam p oduc ion ac i i ies indica es con ol o downs eam pa ic-
ipan s. As downs eam pa icipan s, hey ely on in e media e inpu s
om ups eam GVC pa icipan s. The mo e ups eam MNEs pa icipa e
in GVCs, he mo e imi a ion and compe i ion hey c ea e among do-
mes ic en e p ises. In con as , domes ic i ms ha e mo e oppo uni ies
o accumula e complemen a y asse s and coope a e wi h MNEs ha
ocus on downs eam GVC p oduc ion.
As GVC complexi y is a c i ical ac o a ec ing he dissemina ion o
isks, new ideas, p oduc s, and he impac o MNEs engaging in GVCs
a ies be ween simple and complex GVCs. Unlike isk, which is un-
a oidable and can be in en ionally ans e ed, knowledge ans e is
es ic ed and expensi e. F om an ou side–in pe spec i e, when complex
GVCs ep esen highly specialized di isions o labo , MNEs exploi low-
cos ad an ages in hos coun ies such as labo , land, and na u al e-
sou ces bu do no sou ce di ec ly om local i ms. This kind o ag-
men ed p oduc ion, which is di e en om local p oduc ion and sales,
acili a es economies o scale o speci ic p oduc ion s ages in hos
economies and p o ides op imal p oduc ion a angemen s o MNEs a a
global le el. Gi en he high echnological in ensi y inhe en in complex
GVCs, MNEs’pa icipa ion in in e na ional specializa ion poses chal-
lenges o domes ic i ms in eme ging economies. This is because o
s ic selec ion c i e ia, which a o only domes ic i ms possessing
ad anced echnologies. Thus, eme ging economies ha e compa a i e
ad an ages in simple GVCs, p o iding oppo uni ies o collabo a ion
be ween MNEs and domes ic en e p ises. Thus, we hypo hesize he
ollowing:
H1a. The g ea e he ups eam pa icipa ion o MNEs in GVCs, he
mo e nega i e he e ec on domes ic i ms’inno a ion pe o mance.
The e o e, he posi i e impac o MNEs’pa icipa ion in GVCs is
gene a ed h ough downs eam pa icipa ion.
H1b. The posi i e impac o he downs eam pa icipa ion o MNEs in
GVCs mainly occu s in simple GVCs.
In e nal R&D and abso p ion capaci y o domes ic i ms
MNEs p og essi ely shi hei global s a egic pa adigm om e -
ical in eg a ion o in e o ganiza ional ela ionships and ne wo ks o
main ain co e asse s, capabili ies, and knowledge. Addi ionally, in e -
i m knowledge ans e and collabo a ion syne gies ha e eme ged
(Ande sson &Fo sg en, 2000;Reilly &Sha key Sco , 2014;Meye e al.,
2020). The ex en o which ex e nal spillo e s bene i domes ic i ms
depends on hei abso p i e capaci y. Gene ally, he ad e se impac o
FDI on eme ging economies a ises om inadequa e condi ions o sup-
po abso p ion. Howe e , ex e nal ac o s associa ed wi h MNEs also
in luence hei abso p i e capaci y.
The e olu ion o abso p i e capaci y is dependen on his o ical a-
jec o ies and pa hs, en ailing a ade-o be ween inwa d-looking is-
`
a- is ou wa d-looking abso p i e capabili y; an excessi e dominance by
ei he o hem can ende an en e p ise’s inno a ion sys em dys unc-
ional (Cohen &Le in hal, 1990;Lewin e al., 2011;C escenzi &
Gaglia di, 2018). In a i m, in e nal R&D gene a es new knowledge and
boos s he abso p i e p ocess (Hejazi e al., 2023), indica ing he com-
plemen a y and subs i u i e dynamics be ween in e nal R&D and
ex e nal spillo e s. As in e nal and ex e nal ac o s p esen app op ia e
ha mony, domes ic i ms can p e en inno a i e capabili ies om being
e oded by ex e nal ac o s such as eliance on MNEs’ echnology
ans e . Some schola s analyze he asymme ic p ope ies o spillo e
e ec s, e ealing he impo ance o spillo e di ec ionali y (Kno e al.,
2009). We posi ha asymme y is ela ed o MNEs’posi ions wi hin
GVCs, which de e mine he powe –dependence ela ionship and a ec
he c ea ion and app op ia ion o alue in he en i e p oduc ion
ne wo k. As GVC o ches a o s, MNEs c ea e a chi ec u al ad an ages o
app op ia ely high le els o alue wi hou eso ing o e ical in eg a-
ion (Jacobides e al., 2006) and u ilize his powe o access esou ces
ha a e bene icial o main aining compe i i e ad an age. Gi en he
ambigui y ha whe he MNEs gene a e posi i e spillo e s (Al a o, 2017;
Rojec &Knell, 2018), how MNEs pa icipa e in GVCs explica es his
ambigui y. We an icipa e ha he ups eam GVC ac i i ies o MNEs a e
mainly o ien ed owa d ma ke -seeking mo i a ions, e ec ing less
co-specializa ion han he downs eam GVC ac i i ies. Speci ically, he
main di ec ionali y o spillo e s in GVC p oduc ion sha ing is backwa d,
simila o p e ious esea ch (Ja o cik, 2004), which excludes he GVC
con ex .
F om an in e nal pe spec i e, R&D is a ca alys p omo ing a i m’s
quan i y and quali y o inno a ion. Focusing on he R&D–p oduc ion
ela ionship, schola s unco e con lic ing indings on he e ec s o
coupling and decoupling R&D and p oduc ion (Buciuni &Fino o, 2016;
Buciuni &Pisano, 2021). The endency o sepa a e global R&D and
p oduc ion ac i i ies a ies acco ding o hei capabili ies o manage
complex and geog aphically dispe sed o ganiza ional s uc u es
(Cas ellani &La o a o i, 2020). Conside ing he agmen ed na u e o
GVCs, he sepa a ion o R&D and p oduc ion akes p ecedence because
he ocus is on accessing inno a i e concep s a he han possessing
hem (Ambos e al., 2021). Speci ically, MNEs’pa icipa ion in ups eam
GVCs may subs i u e domes ic i ms’inno a ion, whe eas downs eam
pa icipa ion incu s MNEs’inpu demands o local supplie s. Al hough
ups eam GVC p oduc ion ac i i ies make R&D a p i a e good, su icien
condi ions o acili a ing echnology and knowledge ans e s a e
lacking. Mo eo e , domes ic i ms’dependence on high-quali y inpu s
om MNEs, ce e is pa ibus, diminishes he ma ginal e u ns on inde-
penden R&D. Simila o abso p ion capaci y, how MNEs a ec domes ic
i ms’R&D in complex GVCs is ambiguous, and domes ic i ms can
engage in basic R&D in simple alue chains. Domes ic i ms engage in
complemen a y R&D ac i i ies o p o ide ex e nal echnological op-
po uni ies. The e o e, ma ginal e u ns on domes ic i ms’R&D in-
c ease i MNEs pa icipa e in downs eam p oduc ion ac i i ies in simple
GVCs. Thus, we hypo hesize he ollowing:
H2a. The ups eam pa icipa ion o MNEs in GVCs nega i ely a ec s
he abso p ion capaci y and in e nal R&D o domes ic i ms, nega i ely
a ec ing domes ic i ms’inno a ion pe o mance.
H2b. The nega i e e ec s o MNEs’pa icipa ion in GVCs on in e nal
R&D and abso p ion capaci y o domes ic i ms occu p ima ily in
simple GVCs.
L. Yu e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100603
3
Expo esilience o domes ic i ms
E iciency and esilience a e in e connec ed as hey ep esen com-
plemen a y ace s o sa egua ding i ms’compe i i e ad an ages and
capabili ies o esis isks and dis up ions. Unde emendous unce -
ain y and ins abili y (Ma in e al., 2018), obus esilience helps i ms
esis and quickly adjus o isks and dis up ions. The ade–in es men
nexus o GVCs is es uc u ing, e ec ing shi s in he geog aphical dis-
ibu ion o p oduc ion s ages and pa e ns o s a egic coo dina ion
among di e en ac o s (Ge e i, 2014). The s a egic ocus o MNEs on
ne wo k o ches a ion may indica e an inc easing cos o ansac ions
o MNEs and o he i ms because o hei dec easing con ol and
inc easing ex an e unce ain y. Especially, he con ac incomple eness
con ibu es o he s icky a ibu e o GVCs and makes pa icipan s in
GVCs ulne able o isk and dis up ion (An `
as, 2020) because o he
di icul y o es ablish c edi . Acemoglu e al. (2007) examine he co -
ela ion be ween con ac ual incomple eness, echnological comple-
men a i y, and echnology adop ion. They sugges ha heigh ened
con ac ual incomple eness causes he adop ion o less ad anced ech-
nologies and ha his in luence in ensi ies in cases whe e in e media e
inpu s exhibi highe le els o complemen a i y.
Ups eam and downs eam pa icipa ion in GVCs gene a es he e o-
geneous e ec s on domes ic i ms’expo esilience, as hey ep esen
ma ke - and esou ce-seeking ac i i ies, espec i ely, engende ing
a ying deg ees o in e dependence and echnological complemen a i y.
Wi h g ea e pa icipa ion in ups eam GVC segmen s, MNEs c ea e
mo e added alue and p o ec hei co e echnologies h ough seg e-
ga ion mechanisms. In addi ion, compa ed wi h downs eam GVC
pa icipa ion, ups eam GVC pa icipa ion signals less asse comple-
men a i y and g ea e powe o s a egic maneu e ing. Conside ing
ha he na u e o he isk men ioned in he p e ious sec ion di e s om
ha o knowledge and echnology, he e ec o MNEs’pa icipa ion in
simple and complex GVCs on domes ic i ms’expo esilience is sig-
ni ican . In simple GVCs, once he compa a i e ad an age o i ms in
eme ging coun ies su passes ha o domes ic i ms, MNEs can eloca e
hei p oduc ion ope a ions. In complex GVCs, he inc easing cos o
coo dina ion is associa ed wi h inc eased con ac incomple eness,
which nega i ely a ec s domes ic i ms’expo esilience. Hence, we
hypo hesize he ollowing:
H3a. The ups eam pa icipa ion o MNEs in GVCs nega i ely a ec s
domes ic i ms’expo esilience, nega i ely a ec ing domes ic i ms’
inno a ion pe o mance.
H3b. The nega i e e ec o MNEs’pa icipa ion in GVCs on domes ic
i ms’expo esilience occu s in bo h simple and complex GVCs.
Fig. 1 p esen s ou concep ual model.
Me hodology
Sample and da a
This s udy ocuses on China, he la ges de eloping na ion, ha has
eme ged as a new egional cen e wi hin GVCs (Gao e al., 2021) and
plays a c ucial ole in complex GVCs. Signi ican inwa d o eign di ec
in es men (IFDI) acili a es China’s indus ializa ion p ocess, in e-
g a ing i in o global p oduc ion ne wo ks and in e na ional ma ke s (Lu
e al., 2017). As he p incipal bene icia y o IFDI, he manu ac u ing
sec o se es as a i al link be ween China and he global economy.
Ou sample comp ises 15 manu ac u ing indus ies o 2005–2015.
We selec his pe iod because o he a ailabili y o key da a and ad-
an ages o indus y-le el agg ega e da a. The da a o measu ing
inno a ion pe o mance and es ing mechanisms a e ob ained om he
S a is ics Yea book on Science and Technology Ac i i ies o Indus ial
En e p ises, which selec s he ca ego y o domes ic i ms ha we p i-
o i ize du ing 2005–2015. The a ionale behind employing indus y-
le el agg ega e da a on en e p ises a ises om hei expansi e s a is i-
cal co e age, which encompasses bo h la ge-scale en e p ises and small-
and medium-sized en e p ises (SMEs). The da a indica e ha SMEs
cons i u e 70% o China’s echnological inno a ion (Zhang e al., 2023).
Fig. 1. Concep ual amewo k.
L. Yu e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100603
4
One limi a ion o his da ase is ha he indus y-le el agg ega e da a
may lose some in o ma ion abou indi idual i ms.
T adi ional in e coun y inpu –ou pu (ICIO) ables p o ide a ela-
i ely comple e pic u e o GVC- ela ed p oduc ion ac i i ies compa ed
wi h i m-le el da a wi hou e lec ing he cha ac e is ics o he e oge-
neous i ms (Fo anie e al., 2020). Howe e , i m-le el s udies only
e lec he backwa d linkages o GVCs, ha is, he pu chases o in e -
media e and inal goods om ab oad. The e o e, i is impossible o
u he di e en ia e be ween i ms’pu chases o o eign p oduc s o
domes ic use and expo s, esul ing in he a io o o eign alue added o
expo s being g ea e han 1. ICIO ables dis inguishing domes ic and
o eign i ms a he indus y le el b idge he gap be ween mic o- and
meso-da a.
ICIO ables dis inguishing domes ic and o eign i ms a e sou ced
om he OECD AMNE da abase. We consolida ed C29 and C30 o he
analy ical ac i i y o mul ina ional en e p ises (AMNE) da abase o
add ess he inconsis encies in indus y classi ica ion s anda ds be ween
he di e en da abases. Table A in Appendix p esen s an indus ial
co espondence able and indus y classi ica ion c i e ia. Da a o he
con ol a iables we e ob ained om China’s Na ional Bu eau o S a-
is ics and China Indus y S a is ical Yea book. Gi en ha China’s Na-
ional Bu eau o S a is ics and China Indus y S a is ical Yea book do no
p o ide a sepa a e classi ica ion o domes ic en e p ises, we exclude
o eign-in es ed and Hong Kong-, Macao-, and Taiwan-in es ed indus-
ial en e p ises.
Va iables and measu es
The dependen a iable is he inno a ion pe o mance o domes ic
i ms (Inn). We ocus on new p oduc de elopmen pe o mance, which
di ec ly e lec s and sus ains a i m’s inno a ion pe o mance
(Sub amanian &V ande, 2019). This is calcula ed using he na u al
loga i hm o new p oduc sales pe uni o new p oduc de elopmen
expendi u e. In ou obus ness es s, we use he a io o in en ion pa en
applica ions o o al pa en applica ions as a measu e o inno a ion
pe o mance.
Ou independen a iable is he p oduc ion posi ion o MNEs wi hin
GVCs (FGVC_PS), which measu es and in e p e s he pa h o pa icipa-
ion in GVCs. The highe he index, he mo e ups eam he MNEs a e in
GVCs. Con e sely, he lowe he index, he mo e downs eam he MNEs
a e in GVCs. Consis en wi h Koopman e al. (2010), we e e o
alue-added expo s based on o wa d linkages as ups eam pa icipa-
ion, and in e media e goods impo s based on backwa d linkages as
downs eam pa icipa ion, which e lec he p oduc ion s age a which
he pa icipan is loca ed. The GVC measu emen s a e p ima ily based
on ade o p oduc ion decomposi ion. The la e includes GDP
decomposi ion and inal goods decomposi ion using o wa d and back-
wa d linkages, espec i ely, which u he measu e he p oduc ion
leng h and ela i e ups eam posi ion o GVCs. Rega ding deepe
comp ehension o he g owing in icacy and e inemen o p oduc ion
ne wo ks, no el me ics ha can cap u e he “leng h”o he connec ions
a e necessa y (Meng e al., 2020). Acco ding o p e ious esea ch (Wang
e al., 2022,2021,2017), he posi ions o MNEs wi hin GVCs based on
p oduc ion leng h a e calcula ed by disagg ega ing he dis ibu ion o
alue addi ion ac oss he en i e GVC as ollows:
PL FGVC =X FGVC
VFGVC =
VFLLAEBY +
VFLAEBBY
VFLAEBY (1)
PLy FGVC =Xy FGVC
YFGVC =VLLAEB
YF+VLAEBB
YF
VLAEB
YF
,(2)
and
FGVC PS =PL FGVC
[PLy FGVC]ʹ.(3)
FGVC_PS is he a io o PL _FGVC o PLy_FGVC, and PL _FGVC is he
a e age o wa d p oduc ion leng h om he MNEs’p oduc ion o in-
e media e goods o inal p oduc ion in o he coun ies. V_FGVC ep e-
sen s he alue added gene a ed by MNEs h ough in e media y expo s;
X _FGVC is he agg ega e ou pu esul ing om MNEs’ alue added
h ough he expo s o in e media e goods; PLy_FGVC is he a e age
backwa d p oduc ion leng h om o eign coun ies’p oduc ion o in-
e media e goods o he inal p oduc ion o MNEs in he hos coun y;
Y_FGVC is he o eign alue added embodied in MNEs’impo s o in-
e media e goods; Xy_FGVC deno es he ou pu o inal goods gene a ed
by MNEs wi hin he coun y, e ec ua ing om he impo s o o eign
in e media e goods.
Simple and complex GVCs exhibi di e en complexi y le els. In
simple GVCs, domes ic o o eign alue added c osses na ional bo de s
only once. In complex GVCs, alue added c osses na ional bo de s mo e
han wice. The e o e, he cons uc o GVC complexi y depends on he
numbe o imes he in e media e inpu s c oss na ional bo de s. The
mo e he imes, he mo e specialized he global di ision o labo . The
dicho omy be ween simple and complex GVCs is based on he ollowing
equa ion:
VFLAEBY =
VFLAELYL+
VFLAE(BY −LYL)(4)
VFGVC S =
VFLAELYL,(5)
VFGVC C =
VFLAE(BY −LYL),(6)
VLAEB
YF=VLAEL
YL
F+VLAE(B
YF−L
YL
F)(7)
YFGVC S =VLAEL
YL
F,(8)
and
YFGVC C =VLAE(B
YF−L
YL
F).(9)
V_FGVC_S is he domes ic alue added embodied in MNEs’in e me-
dia e expo s used by he di ec ly impo ing coun y o p oduce p oduc s
consumed in domes ic ma ke s. V_FGVC_C is he domes ic alue added
embodied in MNEs’in e media e expo s, which a e used by he di ec ly
impo ing coun y o p oduce in e media e o inal p oduc s. Y_FGVC_S
is he o eign alue added ha MNEs impo di ec ly om pa ne
coun ies o p oduce p oduc s o domes ic consump ion by hos coun-
ies. Y_FGVC_C is he e u ned domes ic alue added o o eign alue
added embodied in in e media e impo s used by MNEs in hos coun ies
o p oduce inal p oduc s o domes ic consump ion o expo s.
The e o e, he posi ions o he MNEs wi hin he simple and complex
GVCs (FGVC_PS_S and FGVC_PS_C) is calcula ed as ollows:
PL FGVC S =X FGVC S
VFGVC S =
VFLLAELYL+
VFLAELLYL
VFLAELYL,(10)
PLy FGVC S =Xy FGVC S
YFGVC S =VLLAEL
YL
F+VLAEL
LYL
F
VLAEL
YL
F
,(11)
FGVC PS S =PL FGVC S
[PLy FGVC S]ʹ,(12)
PL FGVC C =X FGVC C
VFGVC C =
VFLLAE(BY−LYL)+
VFLAE(BBY −LLYL)
VFLAE(BY −LYL),
(13)
PLy FGVC C =Xy FGVC C
YFGVC C =VLLAE(B
YF−L
YL
F)+VLAE(BB
YF−LL
YL
F)
VLAE(B
YF−L
YL
F),
(14)
L. Yu e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100603
5
and
FGVC PS C =PL FGVC C
[PLy FGVC C]ʹ(15)
The a iables used o es he heo e ical mechanisms a e domes ic
i ms’abso p i e capaci y, in e nal R&D, and expo esilience. We use
he abso p ion cos , which is he a io o echnology abso p ion expen-
di u e o echnology in oduc ion expendi u e, as a p oxy o abso p i e
capaci y. Highe alues indica e a lowe abso p i e capaci y. The
numbe o S&T ins i u ions es ablished by domes ic i ms is used as a
p oxy a iable o in e nal R&D. We cons uc an indica o o expo
esilience based on he exogenous shocks caused by he 2008 inancial
c isis ( he sample pe iod begins in 2009) measu ed as ollows:
Resiliencei, =Expo i, −Expo i,2008
Expo i,2008 (12a)
We con ol o indus y– i m- and indus y-le el ac o s ha may
in luence domes ic i ms’inno a ion pe o mance. A he indus y– i m
le el, we con ol o he e ec s o capi al in ensi y, expo a io, ax li-
abili y, and s a e capi al sha e. A he indus y le el, we con ol o he
e ec o indus y scale. The de ails a e as ollows: Capi al in ensi y (KI)
is he na u al loga i hm o he a io o ne ixed asse s o he a e age
numbe o employees. Expo a io (ER) is he alue o expo deli e ies
as a pe cen age o indus ial sales. Tax liabili y (TL) is he sum o income
axes as a pe cen age o o al p o i s. S a e capi al sha e (SC) is he sha e
o s a e and collec i e capi al in paid-in capi al. Indus y scale (IS) is he
na u al loga i hm o he a e age numbe o employees.
Es ima ion me hod
We employed a panel da a model o in es iga e he impac o MNEs’
pa icipa ion in GVCs on domes ic i ms’inno a ion pe o mance
because i can deal wi h da a ac oss indus ies and yea s by assuming
ha he e ogenei y exis s be ween indus ies and yea s. The Hausman
es shows ha a ixed-e ec s es ima ion is p e e ed o es ing ou
hypo heses (chi-squa ed =17.62, P =0.0138), which is conduci e o
elimina e unobse able he e ogenei y ha emains cons an o e ime.
To add ess he p oblem o he e oscedas ici y, we use obus s anda d
e o s o all es ima ions. Acco dingly, he baseline model is as ollows:
Inni, =
α
0+
α
1FGVC PSi, +Zi, +δi+φ +
ε
i, ,(13a)
Inni, =
α
0+
α
1FGVC S PSi, +Zi, +δi+φ +
ε
i, ,(14a)
and
Inni, =
α
0+
α
1FGVC C PSi, +Zi, +δi+φ +
ε
i, ,(15a)
whe e he subsc ip ideno es he indus y and deno es he yea . δ
i
and
φ
ep esen he indus y- and yea - ixed e ec s, and
ε
i
is he e o e m.
Z
i,
deno es he se o con ol a iables.
Empi ical esul s
Main analysis
Table 1 p esen s he benchma k esul s and examines he impac o
MNEs’pa icipa ion in GVCs on domes ic i ms’inno a ion pe o mance
o es H1a and H1b. Columns (1)–(3) show he es ima ed esul s
wi hou he con ol a iables. Wi h he addi ion o he con ol a iables,
columns (4)–(6) indica e ha he eg ession esul s a e gene ally
consis en .
The mo e ups eam MNEs pa icipa e in GVCs, he mo e nega i ely
hey a ec domes ic i ms’inno a ion pe o mance. When he GVC
index is la ge , MNEs a e close o ups eam GVC p oduc ion.
Con e sely, MNEs a e close o he downs eam GVC p oduc ion. Col-
umn (4) shows ha o domes ic i ms’inno a ion pe o mance, he
coe icien o FGVC_PS is signi ican ly nega i e a he 1% le el. To
u he dis inguish simple and complex GVCs, columns (5) and (6)
p esen he impac o MNEs’pa icipa ion in simple and complex GVCs
on domes ic i ms’pe o mance. The coe icien s o FGVC_PS_S and
FGVC_PS_C a e signi ican ly nega i e and nonsigni ican , espec i ely,
indica ing ha MNEs a ec domes ic i ms’inno a ion pe o mance
mainly h ough simple GVCs. The e o e, H1a and H1b a e suppo ed.
Robus ness es
We conduc addi ional obus ness es s o he indings p esen ed in
he benchma k es ima ions; Tables 2 and 3 p esen he esul s. Fi s , we
adop al e na e measu es o he dependen and independen a iables.
Speci ically, he numbe o in en ion pa en applica ions as a sha e o
Table 1
Main analysis (Dependen a iable: Inno a ion pe o mance).
Va iable (1) (2) (3) (4) (5) (6)
FGVC_PS -1.086*** -1.290***
(0.364) (0.401)
FGVC_PS_S -1.053*** -1.279***
(0.345) (0.381)
FGVC_PS_C 0.719 0.384
(0.566) (0.593)
Cons an 3.608*** 3.587*** 1.924*** 5.687*** 5.735*** 3.831***
(0.324) (0.309) (0.567) (0.701) (0.702) (0.838)
Con ol No No No Yes Yes Yes
Indus y Yes Yes Yes Yes Yes Yes
Yea Yes Yes Yes Yes Yes Yes
Obse a ions 165 165 165 165 165 165
No e: Robus s anda d e o s a e indica ed in pa en heses. *** p<0.01, ** p<0.05, and * p<0.1 ( his applies o all ables in his s udy).
Table 2
Robus ness es (Indica o eplacemen and he PPML model).
Va iable (1) (2) (3) (4)
FGVC_PS_TR -0.954***
(0.277)
FGVC_PS -0.443***
(0.125)
FGVC_PS_S -0.442***
(0.120)
FGVC_PS_C 0.118
(0.177)
Cons an -0.456** 2.107*** 2.127*** 1.472***
(0.178) (0.237) (0.238) (0.267)
Con ol Yes Yes Yes Yes
Indus y Yes Yes Yes Yes
Yea Yes Yes Yes Yes
Wald Chi-squa ed - 50.39*** 51.81*** 46.08***
Obse a ions 165 165 165 165
L. Yu e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100603
6
he o al pa en applica ions is adop ed o measu e domes ic i ms’
inno a ion pe o mance, and he ade decomposi ion app oach (Wang
e al., 2013;Wang e al., 2021) is applied o calcula e MNEs’pa ici-
pa ion in GVCs. Gi en he limi a ions o he ade decomposi ion
app oach, we canno dis inguish be ween simple and complex GVCs.
The es ima ed esul s a e p esen ed in column (1) (Table 2).
Second, we use he Poisson pseudo-maximum likelihood (PPML)
model o eg ession. Gi en ha he le - unca ed da a in he indica o
o inno a ion pe o mance in he benchma k eg ession is g ea e han
ze o, and E[ln(Y)]∕=ln[E(Y)] may be ue because o he loga i hmic
ope a ion, he PPML model can sol e he abo e p oblems. The eg es-
sion esul s a e p esen ed in columns (2)–(4) (Table 2).
Thi d, we exclude e e se causali y, which causes endogenei y, by
lagging he independen a iables and all con ol a iables in columns
(1)–(3) (Table 3). Addi ionally, we use he one-pe iod lag o he co e
explana o y a iable as an ins umen al a iable in columns (4)–(6)
(Table 3). Conside ing endogenei y, he eg ession esul s show ha he
coe icien o he co e explana o y a iable is s ill obus ly nega i e and
inc eases signi ican ly in absolu e e ms compa ed wi h he baseline
eg ession. The nega i e e ec o MNEs’posi ions wi hin GVCs on do-
mes ic i ms’inno a ion pe o mance is unde es ima ed i endogenei y
is no con olled. To ensu e he easonableness o he selec ed ins u-
men al a iable, we conduc a weak iden i ica ion es . The epo ed
signi icance (p- alue) o he ins umen al a iables in he i s s age is
signi ican ly less han 1%, indica ing ha he p oblem o weak ins u-
men al a iables is obscu e in ou eg essions.
He e ogenei y analysis
To in es iga e whe he i m he e ogenei y exis s in he impac o
MNEs’pa icipa ion in GVCs on domes ic i ms’inno a ion pe o -
mance, we pe o m he e ogenei y es s acco ding o domes ic i ms’
R&D in ensi y and RCA. Speci ically, his analysis is based on he 50%
quan iles o each a iable. In each es , we dis inguish be ween simple
and complex GVCs.
In Table 4, columns (1)–(3) show ha he coe icien o FGVC_PS_C is
posi i e a he 5% signi icance le el, whe eas he coe icien s o
FGVC_PS and FGVC_PS_S a e nonsigni ican . Columns (4)–(6) indica e
ha he coe icien s o FGVC_PS, FGVC_PS_S, and FGVC_PS_C a e nega-
i e and s a is ically signi ican . We ind ha he nega i e e ec o
MNEs’pa icipa ion in ups eam GVCs mainly a ec s domes ic i ms
wi h high R&D in ensi y, whe eas MNEs’pa icipa ion in complex up-
s eam GVCs gene a es a posi i e impac . As domes ic i ms ace di i-
cul ies in complex GVCs owing o echnological ba ie s, MNEs gene a e
demons a ion e ec s. Howe e , MNEs’pa icipa ion in ups eam GVCs
causes domes ic i ms o ely on MNEs and subs i u e hem o inde-
penden R&D.
In Table 5, columns (1)–(3) show ha he coe icien s o FGVC_PS
and FGVC_PS_S a e nega i e a he 5% signi icance le el, whe eas he
coe icien o FGVC_PS_C is posi i e a he 10% signi icance le el. Col-
umns (4)–(6) show ha he coe icien s o FGVC_PS, FGVC_PS_S, and
FGVC_PS_C a e nega i e and s a is ically signi ican . Simila ly, MNEs’
pa icipa ion in ups eam complex GVCs bene i s domes ic i ms cha -
ac e ized by low RCA o some ex en , whe eas domes ic i ms cha ac-
e ized by high RCA a e nega i ely a ec ed by MNEs’ups eam
pa icipa ion in GVCs. Conside ing Tables 4 and 5, domes ic i ms wi h
low R&D in ensi y and RCA do no bene i om complex GVCs o ga-
nized and domina ed by MNEs h ough backwa d linkages, as complex
GVCs a e cha ac e ized by high ba ie s o en y.
This sec ion e eals ha i m he e ogenei y is impo an in de e -
mining whe he he impac is signi ican by inco po a ing he di-
cho omy o simple and complex GVCs, al hough he eg ession esul s
o complex GVCs a e unce ain in he main analysis.
Mechanism iden i ica ion
This s udy speci ies he heo e ical mechanisms h ough which
MNEs’pa icipa ion in GVCs a ec s domes ic i ms’inno a ion
Table 3
Robus ness es (Endogenei y ea men ).
Va iable (1) (2) (3) (4) (5) (6)
FGVC_PS -1.598*** -2.493***
(0.338) (0.602)
FGVC_PS_S -1.563*** -2.365***
(0.344) (0.651)
FGVC_PS_C -0.589 -0.830
(0.590) (2.405)
Cons an 5.928*** 5.965*** 4.993*** 7.732*** 7.664*** 5.961**
(0.713) (0.726) (0.952) (1.000) (1.057) (2.873)
Con ol Yes Yes Yes Yes Yes Yes
Indus y Yes Yes Yes Yes Yes Yes
Yea Yes Yes Yes Yes Yes Yes
Obse a ions 150 150 150 150 150 150
Table 4
He e ogenei y analysis (R&D in ensi y).
Low R&D High R&D
Va iable (1) (2) (3) (4) (5) (6)
FGVC_PS -1.148 -1.971***
(0.742) (0.611)
FGVC_PS_S -1.102 -1.879***
(0.734) (0.558)
FGVC_PS_C 1.706** -1.434**
(0.740) (0.601)
Cons an 7.236*** 7.233*** 3.398** 5.962*** 5.965*** 5.617***
(1.688) (1.702) (1.413) (1.231) (1.215) (1.350)
Con ol Yes Yes Yes Yes Yes Yes
Indus y Yes Yes Yes Yes Yes Yes
Yea Yes Yes Yes Yes Yes Yes
Obse a ions 82 82 82 83 83 83
L. Yu e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100603
7
pe o mance and shows ha domes ic i ms’abso p i e capaci y, in-
e nal R&D, and expo esilience a e he key a iables behind hese
mechanisms. The mechanism iden i ica ion echnique used in his s udy
ocuses on he causali y be ween he co e explana o y and media ing
a iables when he impac o he media ing a iable on he dependen
a iable is di ec and e iden .
Table 6 examines he impac o MNEs’pa icipa ion on domes ic
i ms’abso p ion capaci y and in e nal R&D. As we use abso p i e cos
o measu e abso p i e capaci y, an inc ease in abso p i e cos indica es
a dec ease in abso p i e capaci y. Inno a ion pe o mance is nega i ely
a ec ed when abso p i e cos s inc ease. Wi hou dis inguishing be-
ween simple and complex GVCs, column (1) shows ha he coe icien
o FGVC_PS is posi i e a he 1% signi icance le el o domes ic i ms’
abso p ion capaci y, whe eas column (4) shows ha he coe icien o
FGVC_PS is nega i e a he signi icance le el o domes ic i ms’in e nal
R&D. To u he dis inguish be ween simple and complex GVCs, col-
umns (2) and (3) show ha only he coe icien o FGVC_PS_S is signi -
ican ly nega i e o domes ic i ms’abso p ion capaci y, whe eas
columns (5) and (6) show ha only he coe icien o FGVC_PS_S is
signi ican ly nega i e o domes ic i ms’in e nal R&D. The e o e, H2a
and H2b a e suppo ed.
Table 7 shows he impac o MNEs’pa icipa ion in GVCs on do-
mes ic i ms’expo esilience. Column (1) shows ha he coe icien o
FGVC_PS is signi ican ly nega i e o domes ic i ms’expo esilience.
To u he dis inguish be ween simple and complex GVCs, columns (2)
and (3) show ha he coe icien s o FGVC_PS_S and FGVC_PS_S a e bo h
nega i e a he 1% signi icance le el o domes ic i ms’expo esil-
ience. The e o e, H3a and H3b a e suppo ed.
In summa y, he dicho omy be ween simple and complex GVCs is
impo an , e ealing ha he oles o abso p i e capaci y and in e nal
R&D in he ele an mechanisms a e ambiguous i GVCs a e highly
complex, whe eas he ole o expo esilience is clea . Speci ically, an
inc ease in GVC complexi y insula es he low o knowledge and
echnology and is inconduci e o mi iga e he dissemina ion p ocess o
isk and dis up ion. In b ie , he posi i e spillo e s gene a ed by MNEs
a e obscu e in complex GVCs, and he nega i e e ec s on expo esil-
ience a e e iden .
Discussion
Resul s
This s udy demons a es ha MNEs’downs eam a he han up-
s eam pa icipa ion in GVCs is he main way o gene a e posi i e e ec s
on domes ic i ms. This is simila o he indings o Ja o cik (2004) and
Lu e al. (2017). They obse e ha backwa d FDI posi i ely a ec s do-
mes ic i ms. This s udy shi s om he quan i y and quali y o FDI o he
p oduc ion ac i i ies ela ed o FDI, especially he GVC p oduc ion ac-
i i ies o MNEs.
Table 5
He e ogenei y analysis (RCA).
Low RCA High RCA
Va iable (1) (2) (3) (4) (5) (6)
FGVC_PS -1.071** -2.089**
(0.534) (0.872)
FGVC_PS_S -1.056** -2.037**
(0.503) (0.854)
FGVC_PS_C 1.206* -1.334*
(0.688) (0.738)
Cons an 6.212*** 6.263*** 3.246** 7.374*** 7.353*** 6.902***
(1.485) (1.499) (1.405) (1.191) (1.207) (1.221)
Con ol Yes Yes Yes Yes Yes Yes
Indus y Yes Yes Yes Yes Yes Yes
Yea Yes Yes Yes Yes Yes Yes
Obse a ions 83 83 83 82 82 82
Table 6
Mechanism es (Abso p ion capaci y and in e nal R&D).
Abso p ion capaci y In e nal R&D
Va iable (1) (2) (3) (4) (5) (6)
FGVC_PS 2.335** -0.988**
(0.910) (0.463)
FGVC_PS_S 2.253** -0.992**
(0.914) (0.462)
FGVC_PS_C -0.705 0.447
(1.045) (0.497)
Cons an 0.608 0.590 3.978* 0.355 0.405 -1.259*
(1.927) (1.962) (2.068) (0.605) (0.623) (0.723)
Con ol Yes Yes Yes Yes Yes Yes
Indus y Yes Yes Yes Yes Yes Yes
Yea Yes Yes Yes Yes Yes Yes
Obse a ions 165 165 165 165 165 165
Table 7
Mechanism es (Expo esilience).
Expo esilience
Va iable (1) (2) (3)
FGVC_PS -2.075**
(0.992)
FGVC_PS_S -1.963**
(0.957)
FGVC_PS_C -2.212**
(1.000)
Cons an 0.339 0.265 0.918
(1.569) (1.559) (1.704)
Con ol Yes Yes Yes
Indus y Yes Yes Yes
Yea Yes Yes Yes
Obse a ions 105 105 105
L. Yu e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100603
8
