Yu, Jiaju; Xu, Ye; Zhou, Jian; Chen, Wei
A icle
Digi al ans o ma ion, o al ac o p oduc i i y, and i m
inno a ion in es men
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, Jiaju; Xu, Ye; Zhou, Jian; Chen, Wei (2024) : Digi al ans o ma ion, o al
ac o p oduc i i y, and i m inno a ion in es men , Jou nal o Inno a ion & Knowledge (JIK), ISSN
2444-569X, Else ie , Ams e dam, Vol. 9, Iss. 2, pp. 1-10,
h ps://doi.o g/10.1016/j.jik.2024.100487
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/327392
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/4.0/
Digi al ans o ma ion, o al ac o p oduc i i y, and fi m inno a ion
in es men
Jiaju Yu
a,
*,YeXu
a
, Jian Zhou
a
, Wei Chen
b
a
School o S a is ics and Da a Science, Jiangxi Uni e si y o Finance and Economics, Nanchang, China
b
Business School, Jiangxi Uni e si y o Science and Technology, Nanchang, China
ARTICLE INFO
A icle His o y:
Recei ed 20 No embe 2023
Accep ed 24 Ma ch 2024
A ailable online 5 Ap il 2024
ABSTRACT
In he con ex o he digi al economy, empowe ing fi m inno a ion in es men h ough digi al ans o ma-
ion is an impo an s a egy o p omo ing inno a ion de elopmen in China. This s udy in es iga es he
e ec s and mechanisms o digi al ans o ma ion on fi m inno a ion in es men om he pe spec i e o o al
ac o p oduc i i y, using a sample o manu ac u ing companies lis ed on he A-sha e ma ke in China om
2012 o 2021. The ollowing h ee main findings we e ob ained. Fi s , esul s o bo h fixed- and andom-
e ec s eg ession me hods e ealed ha digi al ans o ma ion significan ly p omo es fi ms’inno a ion
in es men . This conclusion emains obus a e con olling o endogenei y issues using he ins umen al
a iable me hod and eplacing he explana o y a iable measu emen me hods. Second, s epwise eg ession
analysis e ealed a nega i e media ing mechanism o o al ac o p oduc i i y (TFP) in he impac o digi al
ans o ma ion on he le el o fi m inno a ion in es men . The main eason o his is ha an imp o emen
in TFP in ensifies he compe i ion o capi al and labo inpu be ween he p oduc ion and inno a ion depa -
men s. Thi d, he g oup eg ession o he e ogenei y analysis ound ha he o e all e ec o digi al ans o -
ma ion on fi m inno a ion in es men is significan ly posi i e in g oups wi h low financing cons ain s and
high human capi al, bu no significan in g oups wi h high financing cons ain s and low human capi al.
Mo eo e , he nega i e media ing mechanism o TFP is significan ly alid in all g oups, u he alida ing he
compe i ion be ween he p oduc ion and inno a ion depa men s o capi al and labo inpu . Based on he
findings o his s udy and om he pe spec i e o go e nance by he Chinese go e nmen , h ee policy ec-
ommenda ions a e p oposed o empowe fi m inno a ion de elopmen h ough digi al ans o ma ion.
© 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/)
Keywo ds:
Fi m inno a ion in es men
Digi al ans o ma ion
To al ac o p oduc i i y
Media ing mechanism
JEL classifica ion:
D8
M21
O3
In oduc ion
The cu en Chinese economy has en e ed a s age o high-quali y
de elopmen . The epo o he 20 h Na ional Cong ess o he Com-
munis Pa y o China also emphasizes ha one o he co e asks in
his de elopmen p ocess is o p omo e inno a i e de elopmen ,
which ine i ably elies on fi ms’inno a ion in es men a he mic o
le el. Simul aneously, wi h he apid de elopmen o he digi al
economy and he inc easingly in eg a ed combina ion o digi al ech-
nologies— ep esen ed by a ificial in elligence, blockchain, cloud
compu ing, and big da a echnologies—wi h he eal economy has
impac ed he ope a ional de elopmen p ocess o fi ms as mic oeco-
nomic en i ies (Fa o e o, 2022). Acco ding o he “Whi e Pape s on
Digi al Economy o China’s Lis ed Companies”(2022), he cu en dig-
i al pene a ion a e o Chinese lis ed companies exceeds 70 %,
signi ying ha digi al ans o ma ion has become an ine i able end
o Chinese fi ms. Exis ing esea ch sugges s ha digi al ans o ma-
ion essen ially in ol es applying new echnologies (digi al echnol-
ogy) o he p oduc ion, ope a ion, and managemen aspec s o fi ms,
he eby changing he o iginal business p ocesses (Chen e al., 2022;
Gilch & Sieweke, 2021;Ilona e al., 2018). The applica ion o new
echnologies is a mani es a ion o echnological p og ess ha can
enhance co po a e ope a ional pe o mance in a ious ways (Gaglio,
2022;Li e al., 2024;Xu e al., 2022). The e o e, can he success ul
applica ion o new echnologies s imula e fi ms o pu sue hem u -
he , he eby enhancing hei le el o inno a ion in es men ? Wha
is he mechanism o ac ion? In he e a o digi al economy, he
answe s o hese ques ions a e ela ed o how he Chinese go e n-
men should con inue o p omo e he le el o inno a i e economic
de elopmen . The e o e, his s udy ocuses on explo ing he impac
o digi al ans o ma ion on inno a ion in es men in Chinese en e -
p ises and cla i ying he unde lying mechanism om he pe spec i e
o o al ac o p oduc i i y (TFP).
* Co esponding au ho .
E-mail add ess: [email p o ec ed] (J. Yu).
h ps://doi.o g/10.1016/j.jik.2024.100487
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/)
Jou nal o Inno a ion & Knowledge 9 (2024) 100487
Jou nal o Inno a ion
&Knowledge
h ps://www.jou nals.else ie .com/jou nal-o -inno a ion-and-knowledge
The main con ibu ions o his s udy a e as ollows. Fi s , i p o-
ides new esea ch pe spec i es. The cu en li e a u e mainly dis-
cusses he impac o digi al ans o ma ion on inno a ion
pe o mance (Li e al., 2023;Liu e al., 2023;Peng & Tao, 2022); how-
e e , li le esea ch has in es iga ed he impac o digi al ans o ma-
ion on fi m inno a ion in es men . The e o e, his s udy can
supplemen he exis ing li e a u e and p o ide new explana ions o
he economic e ec s o digi al ans o ma ion. Second, he cu en
s udy explo es new economic mechanisms. The exis ing li e a u e
mainly examines he impac mechanism o digi al ans o ma ion on
fi m inno a ion om he pe spec i es o global inno a ion ne wo k
in eg a ion, knowledge flow, human capi al enhancemen , and inno-
a ion mode op imiza ion (Ak e e al., 2023;Fan e al., 2024;Fo man
& Zeeb oeck, 2019;Li e al., 2024). Meanwhile, his s udy p o ides a
new explana ion o he impac o digi al ans o ma ion on fi m
inno a ion by cla i ying he media ing e ec o TFP. Thi d, while he
exis ing li e a u e mos ly ocuses on he posi i e impac s o digi al
ans o ma ion on fi ms (Vial, 2019;Wang & He, 2024); his s udy is
he fi s o demons a e ha TFP nega i ely media es he impac o
digi al ans o ma ion on fi ms’inno a ion in es men , which is
impo an o comp ehensi e assessmen o he economic e ec s o
digi al ans o ma ion.
Li e a u e e iew
Wha is he impac o digi al ans o ma ion on inno a ion in es -
men s? Wha is he ole o he TFP in his impac ? This sec ion
e iews ele an li e a u e o add ess hese wo ques ions.
Can fi ms’digi al ans o ma ion enhance hei le els o inno a-
ion in es men ? Exis ing li e a u e has no add essed his ques ion
di ec ly. P e ious s udies ha e p ima ily examined he impac o digi-
al ans o ma ion on inno a ion pe o mance om he pe spec i e
o inno a ion ou comes (Li e al., 2023;Liu e al., 2023;Peng & Tao,
2022). These s udies ha e gene ally acknowledged he significan
posi i e e ec o digi al ans o ma ion on inno a ion pe o mance
in e ms o inno a ion ou pu (Ghasemaghaei & Calic, 2020;Vial,
2019;Wu & Li, 2024;Zapa a e al., 2020) and inno a ion e ficiency
(Gu, 2023;Wang & He, 2024). Digi al ans o ma ion can enhance
inno a ion pe o mance by acili a ing global inno a ion ne wo k
in eg a ion, knowledge flow, human capi al enhancemen , and inno-
a ion mode op imiza ion. Fi s , digi al ans o ma ion helps fi ms
access he e ogeneous and di e sified new knowledge and esou ces
h ough global inno a ion ne wo k in eg a ion, ensu ing he esh-
ness, b ead h, and flexibili y o knowledge, which enhances inno a-
ion pe o mance (Ande sson e al., 2005;Fan e al., 2024;Rabbiosi &
San angelo, 2013;Wu e al., 2021). Second, digi al ans o ma ion
s eng hens fi ms’abili y o diges , abso b, and econs uc new ech-
nologies and knowledge, acili a es in e nal and ex e nal da a and
knowledge sha ing, and imp o es inno a ion pe o mance (Fe ei a
e al., 2019;Fo man & Zeeb oeck, 2019;Ning e al., 2023;Sun, 2024;
U bina i e al., 2020). Thi d, digi al ans o ma ion enhances employ-
ees’abili y o acqui e new knowledge and mas e new skills, p o-
mo es he upg ada ion o human capi al wi hin fi ms, and
consequen ly imp o es inno a ion pe o mance (Kohli & Mel ille,
2019;Li e al., 2024;Smi h e al., 2017). Fou h, digi al ans o ma ion
op imizes inno a ion modes and acili a es adap i e adjus men s in
o ganiza ional business and p ocess managemen , he eby enhancing
inno a ion pe o mance (Ak e e al., 2023;Nambisan e al., 2017).
How does digi al ans o ma ion a ec fi ms’inno a ion in es -
men s in e ms o TFP? Exis ing esea ch has p ima ily ocused on
he impac o digi al ans o ma ion on fi ms’TFP and has consis-
en ly ound ha i p omo es TFP by en iching p oduc ion ac o s
(Zhao e al., 2021), imp o ing he di ision o labo e ficiency (Xia e
al., 2022), educing labo cos s (Acemoglu & Res epo, 2020;K aus e
al., 2022), and op imizing human capi al s uc u e (Au o & Do n,
2013;Cheng e al., 2023). Howe e , he impac o TFP on fi ms’inno-
a ion in es men s emains ambiguous.
In o he wo ds, ex an li e a u e does no di ec ly cla i y he
impac o digi al ans o ma ion on fi ms’inno a ion in es men s o
he ole o TFP. This gap can be explained by wo aspec s ha a e
missing om p e ious s udies. Fi s , he di ec impac o digi al ans-
o ma ion on fi ms’inno a ion in es men s has no been explici ly
discussed. Second, he e is a lack o di ec esea ch on he influence
o TFP on fi ms’inno a ion in es men . Hence, his s udy aims o
in es iga e he impac o digi al ans o ma ion on fi ms’inno a ion
in es men s and he media ing ole o TFP, building upon exis ing
heo e ical and empi ical analyses and p o iding addi ional insigh s.
Theo e ical analysis
This sec ion p o ides a heo e ical analysis explo ing he impac
and mechanisms o digi al ans o ma ion on fi ms’inno a ion
in es men s om he TFP pe spec i e.
Theo e ical analysis o he impac o digi al ans o ma ion on fi ms’
inno a ion in es men s
This s udy posi s ha digi al ans o ma ion significan ly p o-
mo es fi ms’inno a ion in es men s, which can be analyzed om
wo aspec s: inno a ion in es men willingness and inno a ion
in es men capabili y.
Fi s , digi al ans o ma ion enhances inno a ion pe o mance,
he eby inc easing a fi m’s willingness o in es in inno a ion. The
ul ima e goal o fi ms engaging in inno a i e ac i i ies is o gain a
compe i i e ad an age and imp o e ope a ional p ofi s. Digi al ans-
o ma ion can p omo e inno a ion pe o mance h ough a ious
means, including global inno a ion ne wo k in eg a ion, knowledge
flow, enhancemen o human capi al, and op imiza ion o inno a ion
modes (Ak e e al., 2023;Fan e al., 2024;Li e al., 2024;Sun, 2024).
This implies ha by empowe ing digi al ans o ma ion, fi ms can
achie e mo e inno a i e ou comes wi h a gi en le el o inno a ion
in es men , ob aining g ea e compe i i e ad an ages and ope a-
ional p ofi s. Digi al ans o ma ion inc eases he expec ed e u ns
on inno a ion in es men , which ine i ably s imula es a fi m’s will-
ingness o in es in inno a ion.
Second, digi al ans o ma ion can alle ia e financial cons ain s,
he eby enhancing fi ms’abili ies o in es in inno a ion. Inno a i e
ac i i ies a e cha ac e ized by high in es men , high isk, and long
cycles, and ely hea ily on he obus ness and scale o financing.
Howe e , he high unce ain y in inno a ion R&D esul s makes i
di ficul o ma ch he secu i y equi emen s o adi ional comme cial
banks’c edi , leading o financial cons ain s o fi ms’ echnological
inno a ions (Haji assiliou & Sa ignac, 2024). Digi al ans o ma ion
can alle ia e hese financial cons ain s. P e ious esea ch has iden i-
fied in o ma ion asymme y as a significan cause o fi ms’financial
cons ain s (S igli z & Weiss, 1981;Xu e al., 2023). Fi ms accumula e
la ge amoun s o da a and in o ma ion du ing hei in e nal p oduc-
ion and ope a ions (Dua e e al., 2012), which a e apped in dis-
c e e p oduc ion models and canno be e ec i ely explo ed and
u ilized. Digi al echnology empowe men enables he collec ion and
p ocessing o accumula ed da a (Wu e al., 2021). Wi h he suppo o
digi al echnology, massi e amoun s o uns uc u ed and non-s an-
da dized da a can be ans o med in o easily unde s andable isual-
ized da a. Fi ms can analyze acqui ed in e nal and ex e nal
in o ma ion and ob ain epo s and o ecas s ega ding hei business
condi ions (Fan e al., 2024). These p ocessed da a can help financial
ins i u ions mo e accu a ely assess he cu en si ua ion and u u e
p ospec s o fi ms, educe he di ficul y and cos o p e-loan examina-
ions, s eng hen posi i e ma ke expec a ions, and consequen ly
alle ia e he financial cons ain s aced by fi ms (Mani a e al., 2020),
he eby enhancing hei capabili y o in es in inno a ion.
J. Yu, Y. Xu, J. Zhou e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100487
2
Based on he abo e analysis, we can conclude ha digi al ans o -
ma ion p omo es inno a ion pe o mance and inc eases fi ms’will-
ingness o in es in inno a ion, while alle ia ing financing
cons ain s and enhancing fi ms’capabili y o in es in inno a ion.
The e o e, we p opose he ollowing hypo hesis:
Hypo hesis 1. Digi al ans o ma ion significan ly enhances fi ms’
inno a ion in es men le els.
The media ing mechanism o o al ac o p oduc i i y
Digi al ans o ma ion p omo es fi ms’ o al ac o p oduc i i y
Digi al ans o ma ion can imp o e fi ms’TFP by en iching p o-
duc ion ac o s, enhancing he di ision o labo e ficiency, educing
labo cos s, and op imizing human capi al s uc u e.
Fi s , in e ms o en iching p oduc ion ac o s, digi al ans o ma-
ion p o ides fi ms wi h aluable, di e se, and he e ogeneous in o -
ma ion (Li e al., 2022). Mo eo e , in o ma ion heo y sugges s ha
e ficien in o ma ion p ocessing enhances a fi m’s s a egic decision-
making e ficiency. Second, in e ms o enhancing he e ficiency o he
di ision o labo , digi al ans o ma ion educes he cos o in o ma-
ion ansmission be ween di e en depa men s wi hin fi ms,
he eby alle ia ing o ganiza ional managemen issues and imp o ing
p oduc ion. Fo example, fi ms’da a analysis capabili ies can op i-
mize o ganiza ional managemen p ocesses and decision ou comes,
he eby imp o ing hei TFP (Xia e al., 2022). Thi d, digi al ans o -
ma ion dec eases labo cos s o fi ms, u he enhancing p oduc ion
and di ision o labo e ficiency. Digi al ans o ma ion enables he
au oma ion o adi ional p oduc ion p ocesses, esul ing in lowe
labo cos s pe uni o ou pu (K aus e al., 2022). Finally, in e ms o
op imizing human capi al s uc u e, as he alue chain in he
manu ac u ing indus y inc eases, he specializa ion o labo ele-
men s inc eases g adually. High-quali y labo and specialized knowl-
edge acili a e he in eg a ion o a ious links in he alue chain,
con ibu ing o business p ocess imp o emen and cos educ ion in
p oduc ion and ansac ions, highligh ing hei ole in he indus ial
di ision o labo (Acemoglu & Res epo, 2020;Au o & Do n, 2013).
The in elligen de elopmen o fi ms will lead o he subs i u ion o
low-skilled labo wi h ad anced machine y and inc eased demand
o highly educa ed labo , op imizing he human capi al s uc u e o
fi ms (Cheng e al., 2023).
To al ac o p oduc i i y c owds ou fi ms’inno a ion in es men
Imp o emen s in TFP can supp ess fi ms’inno a ion in es men s.
TFP ep esen s he esidual alue o o al ou pu a e accoun ing o
he con ibu ions o labo and capi al (Lee & Viale, 2023;Solow,
1957). An inc ease in TFP igge s a p ice e ec , no only educing he
p oduc ion cos and sales p ice pe uni , bu also pushing ou he p o-
duc ion possibili y on ie , encou aging businesses o expand p o-
duc ion scale, and p omo ing he inpu o capi al and labo in he
p oduc ion depa men (Aghion & Howi , 1994;G ego y, 2022).
Howe e , a fi m’s inno a ion in es men also highly depends on he
a ailabili y o capi al and labo (Ca lo e al., 2016;Honjo e al., 2014).
The e o e, unde limi ed fi m esou ces, an inc ease in TFP in ensifies
compe i ion be ween he p oduc ion and inno a ion sec o s o capi-
al and labo in es men , consequen ly educing he le el o fi m
inno a ion in es men .
Based on he abo e analysis, we conclude ha digi al ans o ma-
ion can p omo e he imp o emen o fi ms’TFP, while TFP can
c owd ou he le el o inno a ion in es men . The e o e, we p opose
he ollowing hypo hesis:
Hypo hesis 2. Rega ding he impac o digi al ans o ma ion on
fi ms’inno a ion in es men le els, a nega i e media ing mechanism
om o al ac o exis s. Tha is, digi al ans o ma ion can p omo e a
fi m’s o al ac o p oduc i i y, he eby inhibi ing inno a ion
in es men .
Empi ical esea ch design
This s udy’s empi ical esea ch design comp ises h ee pa s: a i-
able selec ion, model cons uc ion, and da a selec ion.
Va iable selec ion
Explained a iable
The explained a iable in his s udy is a fi m’s inno a ion in es -
men (Y). A fi m’s inno a ion in es men is measu ed based on
esea ch and de elopmen (R&D) in es men . Conside ing he signifi-
can di e ences in he o al amoun o R&D in es men among fi ms
o di e en sizes, he R&D in ensi y index, a he han he scale o
R&D in es men , be e eflec s he le el o R&D in es men app o-
p ia e o a fi m ela i e o i s size (Jeon & Jung, 2024). Following Jeon
and Jung’s (2024) app oach, his s udy adop s he a io o R&D in es -
men o ope a ing e enue mul iplied by 100 as a measu e.
Explana o y a iable
The explana o y a iable in his s udy is a fi m’s digi al ans o -
ma ion (X). A fi m’s digi al ans o ma ion (X) is measu ed using Wu
e al. (2021)) app oach. In e ms o me hodology, annual epo ex-
ual da a o Chinese A-sha e lis ed companies we e collec ed, and he
equency o ele an keywo ds ela ed o “fi m digi al ans o ma-
ion”was coun ed o ob ain he equency o fi m digi al ans o ma-
ion (FX). The specific keywo ds included a ificial-in elligence
echnology, big-da a echnology, cloud-compu ing echnology,
blockchain echnology, and he applica ion o digi al echnology,
o aling fi e dimensions and 76 specific e ms. Fu he mo e, o mi i-
ga e he impac o ou lie s and he e oscedas ici y on he empi ical
esul s, he final measu e o fi m digi al ans o ma ion (X) was sub-
jec ed o a loga i hmic ans o ma ion based on he equency o fi m
digi al ans o ma ion (FX), as shown in Eq. (1).
X¼lnðFX þ1Þð1Þ
Addi ionally, a obus ness es was conduc ed by eplacing he
explana o y a iable ollowing he app oach o Zhao e al. (2021).
Specifically, based on ou dimensions, namely digi al echnology
applica ion, In e ne business models, in elligen manu ac u ing, and
mode n in o ma ion sys ems, 99 ele an e ms we e selec ed. A
new measu e o a fi m’s digi al ans o ma ion ( e e ed o as digi al
ans o ma ion 2 and deno ed as X2 in his s udy) was cons uc ed
based on he equency o hese e ms.
Media ing a iable
The media ing a iable in his s udy is fi m TFP (Z). The es ima ion
o he sample fi ms’TFP (LP TFP o LPTFP) is based on he LP me hod
p oposed by Le insohn and Pe in (2003). Simila o he cons uc ion
o he fi ms’digi al ans o ma ion (X) measu e, he es ima ed LPTFP
measu e ob ained using he LP me hod unde wen a loga i hmic
ans o ma ion o mi iga e he impac o ou lie s and he e oscedas-
ici y on he empi ical esul s, as shown in Eq. (2).
Z¼lnðLPTFP þ1Þð2Þ
Con ol a iables
To ensu e he obus ness and eliabili y o he empi ical findings,
his s udy included a se o con ol a iables a bo h he mic o and
egional le els, as shown in Table 1.
Model cons uc ion
To es Hypo hesis 1, we cons uc ed a linea eg ession model
wi h fi m and yea e ec s. This model is p esen ed by Eq. (3).
Yi ¼aþbXi þxConi þmiþu þei ð3Þ
J. Yu, Y. Xu, J. Zhou e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100487
3
whe e Y ep esen s he explained a iable indica ing a fi m’s le el o
inno a ion in es men ; X ep esen s he explana o y a iable indi-
ca ing he le el o digi al ans o ma ion o fi ms; he subsc ip s iand
deno e he indi idual samples and yea s, espec i ely; Con ep e-
sen s a se ies o con ol a iables a bo h he mic o-and egional le -
els; m ep esen s indi idual e ec s; u ep esen s ime e ec s; and e
ep esen s he andom e o e m. The ela ionship be ween Eq. (3)
and esea ch Hypo hesis 1 is as ollows: I he es ima ed coe ficien o
he explana o y a iable (X) is significan ly posi i e, i indica es a sig-
nifican p omo ing e ec o digi al ans o ma ion on fi m inno a ion
in es men , he eby suppo ing Hypo hesis 1.
To es Hypo hesis 2 and examine whe he he e TFP media es he
impac o digi al ans o ma ion on fi m inno a ion in es men , we
cons uc ed Eqs. (4) and (5):
Zi ¼a0þb0Xi þx0Coni þm0
iþu0
þe0
i ð4Þ
Yi ¼a00 þb00 Xi þλ00 Zi þx00 Coni þm00
iþu00
þe00
i ð5Þ
Eqs. (3)−(5) we e used o es Hypo hesis 2: In he case whe e
Hypo hesis 1 is suppo ed, i he es ima ed coe ficien o he explana-
o y a iable Xin Eq. (4) is significan ly posi i e and he es ima ed
coe ficien o he media ing a iable Zin Eq. (5) is significan ly nega-
i e, hen digi al ans o ma ion inhibi s fi ms’inno a ion in es -
men s by p omo ing TFP; his suppo s Hypo hesis 2.
Da a selec ion
The esea ch sample comp ised A-sha e lis ed manu ac u ing
companies om 2012 o 2021. Financial da a and annual epo ex s
we e ob ained om he Guo ai An da abase, whe eas egional-le el
da a mainly came om he s a is ical yea books o a ious p o inces
in China. A e cleaning he da a, 21,357 mic o-le el samples we e
ob ained. Due o he lagged ea men o he samples in he empi ical
p ocess, 18,513 mic o-le el samples we e e ained. The da a cleaning
p ocess mainly in ol ed he ollowing asks: Fi s , iden i y he ca e-
go y o he samples based on he “Indus y Classifica ion Code o he
China Secu i ies Regula o y Commission (2012 Edi ion),”and selec
samples om he manu ac u ing indus y. Second, o con ol o he
impac o ex eme alues, he co e a iables o his s udy, including
fi m inno a ion in es men , digi al ans o ma ion, and TFP, unde -
wen a 1 % winso iza ion p ocess.
Table 2 p esen s he main desc ip i e s a is ics ela ed o he
explained a iables, explana o y a iable, media ing a iable, and
con ol a iables in Eqs. (3)−(5). The mean, s anda d de ia ion, and
maximum and minimum alues o each a iable we e wi hin he ea-
sonable anges.
Empi ical analysis
Baseline eg ession esul s analysis
To es Hypo hesis 1, a eg ession analysis was conduc ed using
Eq. (3); he esul s a e p esen ed in Table 3. Bo h fixed e ec s (FE)
and andom e ec s (RE) eg ession me hods we e employed. To
add ess po en ial endogenei y issues a ising om simul aneous cau-
sali y, a one-pe iod lag was applied o he explana o y a iable X
(deno ed as L.X). Columns (1) and (2) p esen he eg ession esul s
using only he main explana o y a iable, L.X. To mi iga e he possi-
ble he e oscedas ici y, columns (3) and (4) employed obus s anda d
e o s clus e ed a he fi m le el. Columns (5) and (6) epo he
eg ession esul s a e in oducing he con ol a iables. P(all mi=0)
ep esen s he p- alue o he F- es ega ding he indi idual fixed
e ec s, indica ing whe he FE eg ession is p e e ed o e he pooled
eg ession. The esul s in columns (1), (3), and (5) indica e ha he FE
eg ession ou pe o ms he pooled eg ession a a significance le el
o 1 %. P-Wald ep esen s he p- alue o he Wald es , indica ing
whe he he FE eg ession is p e e ed o e he RE eg ession. The
esul s in columns (1), (3), and (5) show ha he FE eg ession is p e-
e ed o he RE eg ession a a significance le el o 1 %. The e o e,
he empi ical analysis in his s udy ocused p ima ily on he esul s
o he FE eg ession, wi h he RE eg ession esul s used o obus -
ness checks.
Table 1
In oduc ion o con ol a iables.
Va iable name Abb e ia ion Cons uc ion me hod
Fi m age Age Numbe o yea s since he es ablishmen o he fi m
Selling expense a io Se Selling expenses di ided by o al ope a ing e enue
Tobin’s Q a io Tobin Ma ke alue o he fi m a he end o he yea di ided by he asse ese cos
Cu en asse a io La To al cu en asse s a he end o he yea di ided by o al asse s
Cash asse a io Ca To al cash asse s a he end o he yea di ided by o al asse s
Regional dependency a io D Non-wo king age popula ion o wo king age popula ion a io in he p o ince whe e he fi m is loca ed,
mul iplied by 100
Regional e ail p ice index Rpi Re ail p ice index in he p o ince whe e he fi m is loca ed (calcula ed based on he p e ious yea as 100)
P opo ion o e ia y indus y in egional GDP Ind3 P opo ion o he e ia y indus y’s GDP o he o al egional ou pu in he p o ince whe e he fi m is
loca ed, mul iplied by 100
Table 2
Desc ip i e s a is ics o he a iables.
Va iables Sample capaci y Mean alue S anda d de ia ion Minimum alue Maximum alue
Y 21,357 4.592 3.949 0.056 24.22
X 21,357 1.223 1.283 0 6.148
Z 21,357 2.216 0.107 1.565 2.557
Age 21,357 17.50 5.782 2 63
Se 21,357 0.080 0.095 0.002 0.502
Tobin 21,357 2.345 1.597 0.927 10.63
La 21,357 0.588 0.173 0.177 0.932
Ca 21,357 0.164 0.130 0.008 0.631
D 21,357 37.38 7.170 19.27 57.79
Rpi 21,357 92.87 28.61 0 106.0
Ind3 21,357 51.23 10.25 29.70 83.90
J. Yu, Y. Xu, J. Zhou e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100487
4
Columns (1)−(6) o Table 3 show ha he es ima ed coe ficien s o
he explana o y a iable (L.X) a e significan ly posi i e a he 1 %
le el. This esul emains unchanged when in oducing con ol a ia-
bles, al e ing eg ession me hods, o employing obus s anda d
e o s clus e ed a he fi m le el. Mo eo e , he magni ude o he
es ima ed coe ficien s o digi al ans o ma ion (L.X) emains ela-
i ely s able in bo h he FE and RE eg essions. These findings con-
fi med he obus ness o ou empi ical esul s. Thus, based on he
empi ical esul s in Table 3, i can be concluded ha digi al ans o -
ma ion has a significan p omo ing e ec on fi ms’inno a ion in es -
men , which p o ides p elimina y e idence suppo ing Hypo hesis 1.
Robus ness discussion
Al hough he empi ical esul s in Table 3 p o ide a p elimina y
alida ion o Hypo hesis 1, i is essen ial o assess he obus ness
and eliabili y o his conclusion h ough a obus ness discussion. We
examined he obus ness o Hypo hesis 1 based on wo aspec s: con-
olling o endogenei y issues and conduc ing an al e na i e specifi-
ca ion eg ession.
Con olling o endogenei y issues
Ins umen al a iable me hods we e employed o con ol o
po en ial endogenei y issues in he model specified in Eq. (1), and he
empi ical esul s a e p esen ed in Table 4.
Two a iables we e selec ed as ins umen al a iables o digi al
ans o ma ion (X): he New P oduc s Sales Income o Indus ial fi ms
abo e he Designa ed Scale in he egion (IV1) and he Leng h o Op i-
cal Cable Lines in he egion (IV2). As bo h ins umen al a iables a e
egional-le el a iables, hei alues a e unlikely o be influenced by
mic o-le el co po a e ac o s, hus ensu ing hei deg ee o exogene-
i y. Digi al ans o ma ion undamen ally in ol es he applica ion o
new digi al echnologies (Chen e al., 2022;Gilch & Sieweke, 2021).
IV1, which ep esen s he sales income o new p oduc s, e ec i ely
eflec s he ex en o which a company’s en i onmen accep s new
echnology. IV2 indica es he basic condi ions equi ed o a company
o in oduce digi al echnology. In ui i ely, IV1 and IV2 a e likely o
posi i ely influence a company’s digi al ans o ma ion le el, hus
ensu ing he alidi y o he ins umen al a iables. Fu he mo e, s a-
is ical measu es ob ained h ough eg ession we e used o es he
exogenei y and alidi y o he ins umen al a iables om a s a is i-
cal s andpoin . In column (1), he p- alue o he Hansen J s a is ic (p-
HJ) is g ea e han 0.1, sugges ing ha he hypo hesis ha ins u-
men al a iables a e exogenous canno be ejec ed a he 10 % signifi-
cance le el. The CD-F s a is ic was g ea e han 19.93, indica ing ha
he hypo hesis o weak ins umen al a iables could be ejec ed,
e en a he s ic es c i ical alue. The e o e, he selec ion o ins u-
men al a iables was conside ed easonable.
Table 4 epo s he empi ical esul s o he ins umen al a iable-
fixed e ec s (IV-FE) eg ession and ins umen al a iable- andom
e ec s (IV-RE) eg ession using he ins umen al a iables. In col-
umns (1) and (2), he es ima ed coe ficien s o he lagged explana o y
a iable (digi al ans o ma ion, L.X) emain significan ly posi i e a
he 5 % and 1 % le els. The e o e, e en a e con olling o endogene-
i y issues using ins umen al a iable me hods, he conclusion ha
digi al ans o ma ion significan ly p omo es fi m inno a ion in es -
men emains obus . This finding p o ides u he e idence ha
suppo s Hypo hesis 1.
Al e na e specifica ion o he explana o y a iable
To u he examine he obus ness o Hypo hesis 1, we employed
an al e na i e specifica ion by eplacing he co e explana o y a i-
able. Specifically, lagged one-pe iod digi al ans o ma ion 2 (L.X2)
and lagged wo-pe iod digi al ans o ma ion (L2.X) we e used as he
co e explana o y a iables in he eg ession specified in Eq. (3). The
esul s a e epo ed in Table 5. Bo h fixed e ec s (FE) and andom
e ec s (RE) eg essions we e employed, and he he e oscedas ici y
issues we e confi med using fi m-le el clus e - obus s anda d
e o s.
F om he esul s epo ed in columns (1)−(4) o Table 5, i can be
obse ed ha e en wi h he eplacemen o he co e explana o y
Table 3
Empi ical esul s o he impac o digi al ans o ma ion on fi ms’inno a ion in ensi y.
Va iables (1) (2) (3) (4) (5) (6)
FE RE FE RE FE RE
L.X 0.117*** 0.187*** 0.117*** 0.187*** 0.114*** 0.178***
(5.194) (8.716) (3.288) (5.570) (3.514) (5.821)
Con ol a iable No No No No No No
Indi idual e ec Yes Yes Yes Yes Yes Yes
Time e ec Yes Yes Yes Yes Yes Yes
Cons an 4.100*** 4.340*** 4.100*** 4.340*** 6.611 4.501
(75.409) (48.623) (66.679) (48.698) (1.552) (1.056)
Obse a ions 18,513 18,513 18,513 18,513 18,513 18,513
R-squa ed 0.032 —0.032 —0.086 —
Fs a is ics 52.28 —21.35 —20.33 —
P(all mi = 0) 0.000 —0.000 —0.000 —
P-Wald 0.000 —0.000 —0.000 —
Clus e No No Yes Yes Yes Yes
No e: The explained a iable in he able is fi m inno a ion in es men (Y); he clus e deno es
obus s anda d e o s clus e ed a he fi m le el; alues in pa en heses ep esen /zle els.
“***,”“**,”and “*” ep esen s a is ical significance a he 1 %, 5 %, and 10 % le els, espec i ely.
“P-Wald” ep esen s p- alue o Wald es .
Table 4
Empi ical esul s o he ins umen al a i-
able app oach o con olling endogenei y.
Va iables (1) (2)
IV-FE IV-RE
L.X 0.031** 0.069***
(2.463) (5.021)
Con ol a iable Yes Yes
Cons an —4.494
—(1.204)
Obse a ions 18,113 18,113
R-squa ed 0.081 —
Indi idual e ec Yes Yes
Time e ec Yes Yes
CD-Fs a is ics 55.806 —
P-HJ 0.989 —
Fs a is ics 38.05 —
Clus e Yes No
No e:“CD-FS a is ics” ep esen s C agg-
Donald Wald Fs a is ic. “p-KP” ep esen s p-
alue o Hansen J s a is ic.
J. Yu, Y. Xu, J. Zhou e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100487
5
a iable, he coe ficien es ima es o L.X2 and L2.X emain signifi-
can ly posi i e a he 1 % le el. This confi ms he obus ness o
Hypo hesis 1.
Media ing mechanism o o al ac o p oduc i i y
Resea ch Hypo hesis 2 p oposes a nega i e media ing e ec o TFP
in he ela ionship be ween digi al ans o ma ion and fi m’s inno a-
ion in es men . Specifically, digi al ans o ma ion p omo es he
enhancemen o TFP, which leads o a c owding-ou e ec on a fi m’s
inno a ion in es men . To es Hypo hesis 2, Eqs. (3)−(5) we e
eg essed sequen ially o examine he impac o digi al ans o ma-
ion on fi ms’inno a ion in es men s, he impac o digi al ans o -
ma ion on TFP, and he c owding-ou e ec o TFP on fi ms’
inno a ion in es men . The p omo ion e ec o digi al ans o ma-
ion on a fi m’s inno a ion in es men has al eady been alida ed in
he baseline eg ession specified in Eq. (3) and is epo ed in Table 3.
The e o e, in his sec ion, Eqs. (4) and (5) we e u he eg essed in a
s ep-wise manne o di e en ia e be ween he impac o digi al
ans o ma ion on TFP and he c owding-ou e ec o TFP on fi ms’
inno a ion in es men . The esul s a e summa ized in Table 6. Bo h
fixed e ec s (FE) and andom e ec s (RE) eg essions we e employed
o c oss- alida ion, and obus s anda d e o s we e clus e ed a he
fi m le el. To ensu e compa abili y o he empi ical esul s, he
eg ession esul s o Eq. (3) p ima ily e e o columns (5) and (6) o
Table 3.
The eg ession esul s o Eq. (4) a e epo ed in columns (1) and
(3) o Table 6. I is e iden ha in bo h he FE and RE eg essions, he
coe ficien es ima es o he explana o y a iable L.X a e significan ly
posi i e a he 1 % le el, indica ing a significan p omo ional e ec o
digi al ans o ma ion on fi ms’TFP. The eg ession esul s o Eq. (5)
a e epo ed in columns (2) and (4) o Table 6. Consis en esul s
we e obse ed in bo h he FE and RE eg essions, as he coe ficien
es ima es o TFP (Z) we e significan ly nega i e a he 1 % le el when
conside ing he join impac o digi al ans o ma ion and TFP on
fi ms’inno a ion in es men . This finding sugges s he p esence o a
c owding-ou e ec be ween TFP and fi ms’inno a ion in es men .
The e o e, based on he empi ical esul s epo ed in Table 6, com-
bined wi h Hypo hesis 1, i can be concluded ha digi al ans o ma-
ion inhibi s a fi m’s inno a ion in es men by p omo ing TFP. This
finding alida es Hypo hesis 2.
Fu he mo e, a e conside ing he media ing mechanism o TFP,
he coe ficien es ima es o he explana o y a iable L.X in columns
(2) and (4) o Table 6 emain significan ly posi i e a he 1 % le el and
show a no iceable imp o emen compa ed wi h columns (5) and (6)
in Table 3. This indica es ha he media ing mechanism o TFP pa -
ially cons ains he p omo ional e ec o digi al ans o ma ion on a
fi m’s inno a ion in es men s.
He e ogenei y analysis
Based on he analysis o media ing mechanisms, we belie e ha
while digi al ans o ma ion gene ally p omo es fi ms’inno a ion
in es men , he e exis s a nega i e media ing mechanism s emming
om TFP. The heo e ical analysis sugges s ha his is p ima ily due
o compe i ion o labo and capi al esou ces be ween he p oduc-
ion and inno a ion depa men s when a fi m’s esou ce base is lim-
i ed. This also implies ha when a company’s esou ce base is weak,
compe i ion o unds and labo be ween i s p oduc ion and inno a-
ion depa men s becomes mo e in ense. The e o e, we conduc ed a
he e ogenei y analysis om he pe spec i es o financial cons ain s
and human capi al o p o ide mo e empi ical e idence o he nega-
i e media ing mechanism o TFP.
He e ogenei y o financial cons ain
When fi ms ace high financial cons ain s, hei limi ed abili y o
ob ain ex e nal financing may in ensi y esou ce compe i ion
be ween he p oduc ion and inno a ion depa men s. This s eng h-
ens he nega i e media ing mechanism o TFP, leading o a ia ions
in he impac o digi al ans o ma ion on inno a ion in es men s.
This s udy fi s measu ed he le el o financial cons ain using he
SA index and di ided fi ms in o high and low financial cons ain
g oups. The impac o digi al ans o ma ion on a fi m’s inno a ion
in es men and he media ing mechanism o TFP a e hen examined
o he e ogenei y be ween he wo g oups. Rega ding he he e oge-
nei y analysis o he impac o digi al ans o ma ion on fi ms’inno-
a ion in es men , eg essions we e conduc ed sepa a ely on he
high- and low-financing cons ain g oups using Eq. (3). The esul s
a e epo ed in columns (1) and (4) o Table 7. Rega ding he he e o-
genei y analysis o he media ing mechanism o TFP, eg essions
we e pe o med sepa a ely on he high- and low-financial cons ain
g oups using Eqs. (4) and (5). The esul s a e epo ed in columns (2),
(3), (5), and (6) o Table 7. Fixed e ec s (FE) eg ession was employed
and obus s anda d e o s clus e ed a he fi m le el we e u ilized.
Columns (1) and (4) o Table 7 show ha he coe ficien es ima e
o he explana o y a iable L.X is significan ly posi i e a he 1 % le el
in he low financial cons ain g oup, while i is no significan in he
high financial cons ain g oup. This indica es ha he p omo ion
e ec o digi al ans o ma ion on a fi m’s inno a ion in es men
mainly exis s in he low financial cons ain g oup, bu is no signifi-
can in he high financial cons ain g oup. Fu he analysis sugges s
wo easons o his he e ogenei y.
Fi s , he deg ee o he media ing e ec di e s, and he nega i e
media ing e ec o TFP is highe in he high financial cons ain g oup
han in he low financial cons ain g oup. Based on he coe ficien
es ima es o he main a iables in columns (2), (3), (5), and (6), i can
be seen ha he nega i e media ing mechanism o TFP is significan
Table 6
Empi ical esul s o he media ing mechanism o TFP.
Va iables (1) (2) (3) (4)
FE RE
ZY ZY
L.X 0.005*** 0.189*** 0.005*** 0.269***
(5.600) (5.693) (7.209) (8.525)
Z—15.085*** —14.839***
—(15.207) —(18.474)
Con ol a iable Yes Yes Yes Yes
Cons an 1.902*** 23.642*** 1.957*** 25.747***
(20.807) (5.065) (21.948) (5.604)
Obse a ions 16,634 15,908 18,128 17,394
Indi idual e ec Yes Yes Yes Yes
Time e ec Yes Yes Yes Yes
Clus e Yes Yes Yes Yes
Table 5
Empi ical esul s o he obus ness es using al e na e specifica ions o
he explana o y a iable.
Va iables (1) (2) (3) (4)
FE RE FE RE
L.X2 0.165*** 0.225*** ——
(4.392) (6.508) ——
L2.X ——0.104*** 0.166***
——(2.903) (4.928)
Con ol a iable Yes Yes Yes Yes
Cons an 6.703 4.762 5.179 3.207
(1.575) (1.120) (1.215) (0.747)
Obse a ions 18,115 18,115 15,374 15,374
R-squa ed 0.086 —0.090 —
Indi idual e ec Yes Yes Yes Yes
Time e ec Yes Yes Yes Yes
Fs a is ics 20.85 —18.11 —
Clus e Yes Yes Yes Yes
J. Yu, Y. Xu, J. Zhou e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100487
6
in bo h he low and high financial cons ain g oups. E en a e con-
olling o he influence o TFP, he p omo ional e ec o digi al
ans o ma ion on a fi m’s inno a ion in es men emained signifi-
can in he high financial cons ain g oup. This indica es ha in he
high financial cons ain g oup he con adic ion be ween TFP and
inno a ion in es men in e ms o und alloca ion is mo e in ense
because o weake financing capaci y, which leads o a mo e p o-
nounced nega i e media ing e ec o TFP. Specifically, o each uni
inc ease in digi al ans o ma ion, he nega i e impac on inno a ion
in es men h ough he TFP channel is app oxima ely 0.061 (i.e.,
0.004 £15.288) uni s in he low financial cons ain g oup and
app oxima ely 0.087 (i.e., 0.005 £17.370) uni s in he high financial
cons ain g oup.
Second, he deg ee o he di ec p omo ion e ec di e s; a e
con olling o he influence o TFP, he di ec p omo ion e ec o dig-
i al ans o ma ion on a fi m’s inno a ion in es men is lowe in he
high financial cons ain g oup han in he low financial cons ain
g oup. F om he empi ical esul s in Columns (3) and (6), he coe fi-
cien es ima e o he explana o y a iable L.X is 0.186 in he low
financial cons ain g oup and 0.136 in he high financial cons ain
g oup. This indica es ha a e con olling o he influence o TFP,
he di ec p omo ion e ec o digi al ans o ma ion on fi ms’inno a-
ion in es men in he low financial cons ain g oup is app oxi-
ma ely 1.37 imes highe han ha in he high financial cons ain
g oup, which also con ibu es o he insignifican impac o digi al
ans o ma ion on fi ms’inno a ion in es men s in he high financial
cons ain g oup.
The empi ical findings indica e ha om he pe spec i e o finan-
cial cons ain s, he e is compe i ion o capi al be ween he p oduc-
ion and inno a ion depa men s. When he le el o financial
cons ain s is high, his compe i ion in ensifies, s eng hening he
nega i e media ing mechanism o TFP, and ende ing he o e all
impac o digi al ans o ma ion on fi ms’inno a ion in es men
insignifican . By con as , when he le el o financing cons ain s is
low, fi ms ha e su ficien abili y o ob ain financial suppo , which
can alle ia e his compe i ion. The e o e, in he low financial con-
s ain g oup, he o e all impac o digi al ans o ma ion on fi ms’
inno a ion in es men s is significan ly posi i e.
He e ogenei y o human capi al
The e is compe i ion o labo be ween he p oduc ion and inno-
a ion depa men s wi hin fi ms. The dependence o fi m inno a ion
ac i i ies on labo mainly in ol es highly educa ed pe sonnel, specifi-
cally hose wi h highe le els o human capi al, such as employees
wi h a bachelo ’s deg ee o highe (Ca lo e al., 2016). The e o e, his
s udy measu es he le el o a company’s human capi al using he p o-
po ion o employees wi h a bachelo ’s deg ee o highe . Companies
a e di ided in o wo g oups: hose wi h high and hose wi h low
human capi al. This di ision allows us o examine he impac o digi-
al ans o ma ion on fi ms’inno a ion in es men and he he e oge-
neous pe o mance o he media ing mechanism o TFP be ween he
wo g oups o samples. To analyze he he e ogenei y o he e ec o
digi al ans o ma ion on fi m inno a ion in es men , samples om
he high and low human capi al g oups we e subjec ed o eg ession
analysis based on Eq. (3). The esul s a e epo ed in columns (1)−(6)
o Table 8.
Columns (1) and (4) o Table 8 e eal ha in he high human capi-
al g oup, he es ima ed coe ficien o he explana o y a iable L.X is
significan ly posi i e a he 5 % le el. By con as , in he low human
capi al g oup, he es ima ed coe ficien o he explana o y a iable L.
X. was no significan . This indica es ha he p omo ional e ec o
digi al ans o ma ion on fi ms’inno a ion in es men s is p ima ily
Table 7
Empi ical esul s o he he e ogenei y analysis om he pe spec i e o financial cons ain s.
Va iable Low financial cons ain g oup High financial cons ain g oup
(1) (2) (3) (4) (5) (6)
YZY YZY
L.X 0.114*** 0.004*** 0.186*** 0.054 0.005*** 0.136***
(2.929) (3.372) (4.569) (1.188) (3.866) (2.912)
Z——15.288*** —— 17.370***
——(9.858) —— (11.683)
Con ol a iable Yes Yes Yes Yes Yes Yes
Cons an 8.788* 1.760*** 18.803*** 1.324 2.014*** 37.425***
(1.691) (15.596) (3.005) (0.188) (14.329) (5.242)
Obse a ions 9085 9085 9085 9030 9030 9030
Indi idual e ec Yes Yes Yes Yes Yes Yes
Time e ec Yes Yes Yes Yes Yes Yes
Clus e Yes Yes Yes Yes Yes Yes
Table 8
Empi ical esul s o he he e ogenei y analysis om he pe spec i e o human capi al.
Va iable Low human capi al g oup High human capi al g oup
(1) (2) (3) (4) (5) (6)
YZ Y YZ Y
L.X 0.024 0.005*** 0.062** 0.116** 0.004*** 0.136***
(0.895) (4.093) (2.341) (2.330) (3.322) (2.912)
Z—— 8.266*** —— 22.770***
—— (7.741) —— (15.971)
Con ol a iable Yes Yes Yes Yes Yes Yes
Cons an 2.346*** 2.069*** 19.806*** 4.553*** 2.038*** 51.113***
(5.601) (122.713) (8.278) (5.822) (130.073) (16.759)
Obse a ions 8408 8408 8408 8752 8752 8752
Indi idual e ec Yes Yes Yes Yes Yes Yes
Time e ec Yes Yes Yes Yes Yes Yes
Clus e Yes Yes Yes Yes Yes Yes
J. Yu, Y. Xu, J. Zhou e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100487
7
p esen in he high-human-capi al g oup. This di e ence can be
explained in wo ways, simila o financial cons ain s.
Fi s , he e was a di e ence in he deg ee o he media ing e ec s.
The nega i e media ing e ec o TFP is g ea e in he low human cap-
i al g oup han in he high human capi al g oup. Based on he coe fi-
cien es ima ion esul s o he main a iables in columns (2), (3), (5),
and (6), i is e iden ha he nega i e media ing mechanism o TFP is
significan ly alid in bo h he high and low human capi al g oups. In
he low-human capi al g oup, e en a e con olling o he impac o
TFP, he p omo ional e ec o digi al ans o ma ion on fi ms’inno a-
ion in es men s emains significan . Second, he e was a di e ence
in he deg ee o di ec p omo ional e ec s. A e con olling o he
impac o TFP, he di ec p omo ional e ec o digi al ans o ma ion
on fi ms’inno a ion in es men in he low-human capi al g oup was
lowe han ha in he high-human capi al g oup.
The e o e, he empi ical findings sugges ha om he pe spec-
i e o human capi al, he e is compe i ion o esou ces be ween he
p oduc ion and inno a ion depa men s. When he le el o human
capi al is low, owing o he ela i e sca ci y o highe -le el human
capi al, his compe i ion in ensifies, ein o cing he nega i e media -
ing mechanism o TFP, which in u n makes he o e all impac o dig-
i al ans o ma ion on fi m inno a ion in es men insignifican .
When human capi al is high, compe i ion is e ec i ely alle ia ed.
Conclusion and policy ecommenda ions
This s udy uses a sample o manu ac u ing companies lis ed on
China’s A-sha e ma ke om 2012 o 2021 o in es iga e he impac
and mechanisms o digi al ans o ma ion on fi ms’inno a ion
in es men s om he pe spec i e o o al ac o p oduc i i y (TFP).
Based on he heo e ical and empi ical esea ch, his s udy makes he
ollowing h ee findings: Fi s , digi al ans o ma ion p omo es fi ms’
inno a ion in es men s significan ly. This indica es ha digi al ans-
o ma ion has become an impo an d i ing o ce o p omo ing Chi-
na’s economic inno a ion and de elopmen le el. Second, TFP has a
nega i e media ing e ec in he ela ionship be ween digi al ans-
o ma ion and he le el o a fi m’s inno a ion in es men . Specifi-
cally, digi al ans o ma ion can p omo e a fi m’s TFP, which, in u n,
supp esses inno a ion in es men s. The p ima y eason o he phe-
nomena obse ed in he empi ical findings is ha an imp o emen in
TFP in ensifies compe i ion o capi al and labo inpu s be ween he
p oduc ion and inno a ion depa men s o a company. This finding
indica es ha fi ms ace he p oblem o di ec ional decision-making
in he p ocess o digi al ans o ma ion. Thi d, he e ogenei y s udies
ound ha he impac o digi al ans o ma ion on fi ms’inno a ion
in es men a ies significan ly wi h di e en le els o financing con-
s ain s and human capi al. F om he pe spec i e o financing con-
s ain s, he o e all impac o digi al ans o ma ion on a fi m’s
inno a ion in es men is significan ly posi i e in he low financial
cons ain g oup, bu no in he high financial cons ain g oup. This
is explained by wo easons: he highe nega i e media ing e ec o
he TFP in he high financial cons ain g oup compa ed o he low
financial cons ain g oup, and he lowe p omo ion e ec o digi al
ans o ma ion on fi ms’inno a ion in es men in he high financial
cons ain g oup compa ed o he low financial cons ain g oup.
F om he pe spec i e o human capi al, he o e all e ec o digi al
ans o ma ion on fi m inno a ion in es men is significan ly posi-
i e in he high human capi al g oup bu no significan in he low
human capi al g oup. The e a e wo easons o his finding. The neg-
a i e media ing e ec o he TFP media ion mechanism is g ea e in
he low human capi al g oup han in he high human capi al g oup.
The di ec p omo ing e ec o digi al ans o ma ion on fi ms’inno-
a ion in es men was lowe in he low human capi al g oup han in
he high human capi al g oup. The he e ogenei y analysis u he
confi ms he esou ce compe i ion be ween he p oduc ion and inno-
a ion depa men s. Simul aneously, i also demons a es ha
educing financing cons ain s and imp o ing human capi al le el
a e e ec i e app oaches o alle ia ing esou ce compe i ion and
s eng hening he ole o digi al ans o ma ion in p omo ing en e -
p ise inno a ion in es men .
The main con ibu ions o his s udy a e as ollows. Fi s , i p o ide
new esea ch pe spec i es. While he cu en li e a u e mainly exam-
ines he impac o digi al ans o ma ion on inno a ion pe o mance
(Li e al., 2023;Liu e al., 2023;Peng & Tao, 2022), ew s udies ha e
explo ed he impac o digi al ans o ma ion on fi m inno a ion
in es men . The e o e, his s udy can supplemen he ex an li e a u e
and p o ide new explana ions o he economic e ec s o digi al ans-
o ma ion. Second, i in es iga es new economic mechanisms. The
exis ing li e a u e mainly explo es he impac mechanism o digi al
ans o ma ion on fi m inno a ion om he pe spec i es o global
inno a ion ne wo k in eg a ion, knowledge flow, human capi al
enhancemen , and inno a ion mode op imiza ion (Ak e e al., 2023;
Fan e al., 2024;Fo man & Zeeb oeck, 2019;Li e al., 2024). Meanwhile,
his a icle p o ides a new explana ion o he impac o digi al ans-
o ma ion on fi m inno a ion by cla i ying he media ing e ec o TFP.
Thi d, while he exis ing li e a u e mos ly ocuses on he posi i e
impac s o digi al ans o ma ion on fi ms (Vial, 2019;Wang & He,
2024), his s udy is he fi s o find ha TFP nega i ely media es he
e ec o digi al ans o ma ion on fi ms’inno a ion in es men , which
is impo an o comp ehensi e assessmen o he economic e ec s o
digi al ans o ma ion. Ne e heless, his s udy had ce ain limi a ions.
Fi s , he esea ch objec o his s udy is manu ac u ing fi ms; mean-
ing, i does no conside non-manu ac u ing fi ms. The e o e, i does
no compa e he simila i ies and di e ences in he impac o digi al
ans o ma ion on fi m inno a ion in es men ac oss di e en indus-
ies. Second, his s udy explo ed he media ing mechanism o TFP on
he impac o digi al ans o ma ion on fi ms’inno a ion in es men s.
Howe e , digi al ans o ma ion in ol es a ious aspec s o en e p ise
p oduc ion, ope a ions, and managemen and i s economic impac s
a e ine i ably di e se. Conside ing only TFP’s media ing mechanism
may in oduce limi a ions om a esea ch pe spec i e. The e o e,
u u e esea ch could expand on he basis o his s udy in e ms o
esea ch subjec s and impac mechanisms.
Based on he conclusions o his s udy, we p opose h ee policy
sugges ions om he pe spec i e o Chinese go e nmen go e nance
o empowe fi m inno a ion de elopmen h ough digi al ans o -
ma ion. Fi s , he s udy cla ifies he policy o ien a ion and encou ages
fi ms o engage in digi al ans o ma ion. This s udy sugges s ha
digi al ans o ma ion p omo es inno a ion in es men s in fi ms.
The e o e, he go e nmen should cla i y i s policy di ec ion, and he
o mula ion o ela ed policies should be conduci e o fi ms unde -
aking digi al ans o ma ion. Fo fi ms engaged in digi al ans o -
ma ion, i is impo an o s eng hen ela ed in as uc u e, es ablish
a scien ific e alua ion sys em, and in oduce a ge ed subsidy poli-
cies. Second, he go e nmen should in oduce suppo ing financing
policies and imp o e he financing en i onmen o digi al ans o -
ma ion. The in e media y mechanism and he e ogenei y analysis
show ha compe i ion o capi al be ween p oduc ion and inno a-
ion depa men s can weaken he acili a ing e ec o digi al ans o -
ma ion on fi m inno a ion in es men . The go e nmen should
o mula e suppo ing policies om he pe spec i e o co po a e
financing o educe he financial cons ain s fi ms ace du ing digi al
ans o ma ion. Fo example, es ablishing special c edi subsidy
unds o fi ms’digi al ans o ma ion beha io s and accele a ing he
o mula ion o policies o issuing co po a e bonds ela ed o digi al
ans o ma ion. Thi d, policy make s should o mula e alen policies
o enhance he supply o highe human capi al. The in e media y
mechanism and he e ogenei y analysis also highligh ed compe i ion
o labo , especially highe human capi al, be ween p oduc ion and
inno a ion depa men s. The e o e, he go e nmen should also
imp o e he co po a e human capi al s uc u e h ough policy
means, such as s eng hening in es men in highe educa ion in
J. Yu, Y. Xu, J. Zhou e al. Jou nal o Inno a ion & Knowledge 9 (2024) 100487
8
