Mohnen, Pie e; Mai esse, Jacques; No en, Ad
Wo king Pape
Inno a ion and p oduc i i y: The ecen empi ical
li e a u e and he s a e o he a
UNU-MERIT Wo king Pape s, No. 2025-003
P o ided in Coope a ion wi h:
Maas ich Economic and Social Resea ch Ins i u e on Inno a ion and Technology (UNU-MERIT),
Uni ed Na ions Uni e si y (UNU)
Sugges ed Ci a ion: Mohnen, Pie e; Mai esse, Jacques; No en, Ad (2025) : Inno a ion and
p oduc i i y: The ecen empi ical li e a u e and he s a e o he a , UNU-MERIT Wo king Pape s,
No. 2025-003, Uni ed Na ions Uni e si y (UNU), Maas ich Economic and Social Resea ch Ins i u e
on Inno a ion and Technology (UNU-MERIT), Maas ich ,
h ps://doi.o g/10.53330/AEUN4217
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/326932
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-sa/4.0/
#2025-003
Inno a ion and p oduc i i y: he ecen empi ical li e a u e and
he s a e o he a
Jacques Mai esse, Pie e Mohnen and Ad No en
Published 3 Feb ua i 2025
DOI: h ps://www.doi.o g/10.53330/AEUN4217
Maas ich Economic and social Resea ch ins i u e on Inno a ion and Technology (UNU-MERIT)
email: in o@me i .unu.edu | websi e: h p://www.me i .unu.edu
Boschs aa 24, 6211 AX Maas ich , The Ne he lands
Tel: (31) (43) 388 44 00
UNU-MERIT Wo king Pape s
ISSN 1871-9872
Maas ich Economic and social Resea ch Ins i u e on Inno a ion and Technology
UNU-MERIT
UNU-MERIT Wo king Pape s in end o dissemina e p elimina y esul s o esea ch ca ied
ou a UNU-MERIT o s imula e discussion on he issues aised.
h ps://c ea i ecommons.o g/licenses/by-nc-sa/4.0/
1
Inno a ion and p oduc i i y: he ecen empi ical li e a u e and he s a e
o he a
Jacques Mai esse (Ins i u Poly echnique de Pa is, CREST, NBER)
Jacques.Mai esse@ensae.
Pie e Mohnen (Maas ich Uni e si y and CIRANO)
p.mohnen@maas ich uni e si y.nl
Ad No en (Maas ich Uni e si y)
ad.no en@maas ich uni e si y.nl
Janua y 2025
Abs ac
This pape e iews he empi ical wo k ha has been done o e he pe iod 2013-2023 on he opic
o inno a ion and p oduc i i y. A isual g aph based on keywo ds shows he main a eas ha ha e
been in es iga ed. The li e a u e e iew is o ganized a ound he way he link be ween inno a ion
and p oduc i i y has been analyzed, he da a ha ha e been used, and he e idence ha has been
ob ained. The pape ends wi h sugges ions o u u e esea ch on he opic.
Keywo ds: li e a u e e iew, p oduc i i y, inno a ion, mic o da a
JEL codes: D24, O30, O31, O32
Acknowledgemen : We hank Ma co Vi a elli and wo e e ees o hei cons uc i e commen s.
This pape has been accep ed o publica ion in he Eu asian Business Re iew.
2
1. In oduc ion
This pape akes s ock o wha has been done o e he pe iod 2013-2023 on he opic o inno a ion
and p oduc i i y emphasizing he di ec ion in which esea ch has been going, wha has been ound
and wha could be done in he u u e. I is by p esen ing wha has been ound and how i has been
done ha we ge a clea e pic u e o wha needs o be done in he u u e.
We shall build up on he e iews by Hall e al. (2010), Mai esse and Mohnen (2010), Hall (2011),
Mohnen and Hall (2013), Mohnen (2019) and Ugu and Vi a elli (2021). We shall ci cumsc ibe
he una oidable o e lap wi h he p e ious e iews by es ic ing his one o pape s published o e
he pe iod 2013-2023, by no e iewing he li e a u e ha ela es R&D o p oduc i i y wi hou
conside ing inno a ion ou pu , and by no co e ing any heo e ical pape s. To clea ly delimi he
scope o his li e a u e e iew, we only co e empi ical wo k using mic o da a. As such, we do no
co e pape s dealing essen ially wi h he e ec i eness o policy in e en ions like en i onmen al
egula ions, subsidies o ax incen i es, no pape s looking a de e minan s o p oduc i i y o
inno a ion wi hou making he link be ween he wo a iables, no agen -based models eplica ing
s ylized ac s o indus ial dynamics. We ha e also delibe a ely elimina ed pape s on pa icula
echnologies, like ag icul u al echnologies, digi iza ion, au oma ion, o a i icial in elligence,
unless hey a e ela ed o he link be ween inno a ion and p oduc i i y.
We ha e ce ainly missed some con ibu ions o his li e a u e. We apologize o ha . The main
poin was no o include each and e e y pape w i en in his a ea, bu mainly o indica e he
di ec ions in which he li e a u e has been de eloping in he las en yea s. We s a ed by collec ing
pape s in a sys ema ic way using a mul i-p onged app oach. Ou sea ch s a egy i s ocused on
pee - e iewed pape s ha ci e he 1998 pape by C épon, Dugue and Mai esse (C épon e
al.,1998), a pape also known by he ac onym CDM. In pa allel we also collec ed pee - e iewed
pape s s udying inno a ion and p oduc i i y using mic o-da a h ough a gene al keywo d sea ch
in he i le, abs ac and keywo d ields. This sea ch was delimi ed by a selec ion o pee - e iewed
jou nals and wo king pape se ies, i.e. a “jou nal se ”1, which is a well-accep ed way o delinea ing
a esea ch a ea (Miloje ic, 2020, p.184). Then, in a hi d s age we looked a speci ic ci a ions in
he lis o e e ences o he pape s we e iewed, aking on boa d hose pape s ha ma ched he
ea lie sea ch c i e ia. In all, mo e han a housand pape s we e collec ed o ini ial e iew.
P oduc i i y essen ially means p oducing mo e ou pu wi h less inpu , bu i can be measu ed in
di e en ways: single ac o p oduc i i y like ou pu pe hou s wo ked, o al ac o p oduc i i y,
which is he ou pu no explained by he inse o all ac o s o p oduc ion, e enue p oduc i i y,
whe e ou pu is exp essed in nominal p ices and he e o e cap u es bo h he o al ac o
1 Jou nals and (wo king) pape se ies usually se e as an ou le o specific esea ch communi ies. Au ho s, edi o s,
e iewe s and eade s o hese ou le s a e an in eg al pa o hese communi ies and iew pa icula jou nals as
“co e” o hei communi y, esea ch a ea and discipline.
3
p oduc i i y and he p ice ma kup, and e iciency, which measu es he dis ance o he bes p ac ice
on ie . Because o he pauci y o good s a is ical da a, like he una ailabili y o capi al s ock,
ma e ials o ou pu p ices, p oduc i i y is o en measu ed by labou p oduc i i y o e enue
p oduc i i y and he e o e measu es only impe ec ly o al ac o p oduc i i y. Besides he
di icul ies ela ed o he da a needed o measu e p oduc i i y, ano he majo challenge is o
es ima e he ma ginal p oduc i i ies o ou pu elas ici ies o he a ious inpu s. Unde some
s ingen assump ions like pe ec ly compe i i e ma ke s, cons an e u ns o scale and op imal
ac o holdings, o al ac o p oduc i i y can be compu ed by he index me hod. Al e na i ely, i
can be es ima ed econome ically om a p oduc ion unc ion, especially i some o hese
assump ions a e no imposed. Ano he p oblem ha many s udies ha e ied o ackle ecen ly is
he endogenei y o ce ain inpu s. Recen s udies ha e ackled his di icul y by assuming ha he
p oduc i i y shock ha is known o he i m bu unobse ed by he econome ician can be p oxied
by ano he obse ed a iable and ollows a i s -o de Ma ko p ocess ha can be app oxima ed
by a polynomial unc ion, see Olley and Pakes (1996), Le insohn and Pe in (2003) and Acke be g
e al. (2015). Gene ally, le els o p oduc i i y a e conside ed, bu some imes also g ow h a es,
i m le els ela i e o sec o means o quin iles o he p oduc i i y dis ibu ion (Aud e sch and
Beli ski, 2020; Coad e al., 2016).
Inno a ion can be measu ed in di e en ways as well. On he inpu side, you ha e R&D
expendi u e o he pu chase o pa en s and licenses. On he ou pu side, you ha e pa en s, new
p oduc s/se ices, new p ocesses, new ways o o ganizing p oduc ion o o ma ke ing he p oduc s.
Gi en he huge li e a u e on R&D and p oduc i i y, which has been e iewed many imes al eady
(see Hall e al., 2010; Kokko e al., 2015; and Ugu e al., 2016), we shall dis ega d hese s udies
and concen a e on he li e a u e o inno a ion ou pu and p oduc i i y. Inno a ion inpu s a e only
examined in so a as hey a ec inno a ion ou pu .
The e has been a p oli e a ion o new su ey da a ha allow o mo e con ol a iables, mo e
sou ces o he e ogenei y, mo e coun ies, longe ime se ies and he examina ion o inno a ion in
pa icula echnologies like ICT (In o ma ion and Communica ion Technologies), obo iza ion,
digi iza ion, a i icial in elligence o en i onmen al inno a ions. New wa es o he Inno a ion
Su eys, which ollow he guidelines o he Oslo Manual (OECD-Eu os a , 2018), ha e been
conduc ed in many coun ies, including de eloping coun ies. The Wo ld Bank En e p ise Su eys
(WBES) conduc ed by he Wo ld Bank in many de eloped and de eloping coun ies, and he
Business En i onmen and En e p ise Pe o mance Su eys (BEEPS) conduc ed by he Eu opean
Bank o Recons uc ion and De elopmen and he Wo ld Bank in coun ies om Eas e n and
Cen al Eu ope and om Cen al Asia and he Caucasus, allow in e na ional compa isons and links
wi h many o he a iables besides inno a ion and p oduc i i y. The su ey da a ha e he
disad an age o be sel - epo ed and subjec i e (see Ci e a and Muzzi, 2016 o a c i ical iew o
he inno a ion su eys in de eloping coun ies). The e is ongoing wo k a emp ing o measu e
inno a ion in a mo e imely ashion using web-based da a. As an example, Na han and Rosso
(2022) use e en da a on he launch o a new p oduc o se ice as a new me ic o inno a ion and
4
combine hese da a wi h da a om he UK companies egis e . Thei new p oduc /se ice launch
da a a e ob ained wi h he help o machine lea ning ou ines om e en s epo ed in 3740 news
sou ces.
On he modeling side, he e a e s ill some s udies, bu ew o hem, ha eg ess p oduc i i y on
inno a ion conside ing he la e o be exogenous (Ca alho and de A ella (2017) o B azil, Long
and Ahn (2017) o Vie nam, Liao (2020) o Spain, Na han and Rosso (2022) o he UK). The
CDM model has clea ly been he wo kho se o he empi ical li e a u e on inno a ion and
p oduc i i y in he las 10 yea s. The model has he pa icula i y ha (1) i handles he endogenei y
o R&D and inno a ion in he p oduc i i y equa ion due o measu emen e o s, e e se causali y
o common dependence wi h hi d ac o s; (2) i deals wi h he selec ion in o R&D and inno a ion,
i.e. i models he non-occu ence o hose wo a iables; and (3) i p oposes a causali y unning
om R&D o inno a ion o p oduc i i y (mo e on his la e ).2 Ac ually, he au ho s o he o iginal
CDM pape had in mind a amewo k mo e han a model in which o inco po a e R&D, inno a ion
and p oduc i i y. As we shall documen , he o iginal CDM model has indeed been ex ended in
di e en di ec ions. A new app oach in oduces s ochas ici y in he ela ionships be ween R&D,
inno a ion and p oduc i i y: no all R&D p ojec s lead o inno a ion, and new p oduc s o
p ocesses do no necessa ily inc ease p oduc i i y. Finally, some s udies ha e es ed o a causal
ela ionship be ween inno a ion and p oduc i i y by using andomized o quasi- andom
expe imen s.
As a way o scoping he cu en “inno a ion and p oduc i i y landscape”, we ha e collec ed all
he keywo ds con ained in he inal selec ion o pape s and classi ied hem, somewha a bi a ily,
in o wel e classes as shown in Table 1 below. Fo ins ance, “collabo a ion” would be classi ied
unde inno a ion cha ac e is ics, “ i m size” unde i m cha ac e is ics and “gene al p opensi y
sco e” unde econome ic echniques. “Eme ging coun ies” eg oups all he coun ies ha a e no
classi ied as ad anced economies in he IMF classi ica ion.
Table 1. Twel e Classes o keywo ds
1. Da a
7. Inno a ion Cha ac e is ics
2. Econome ic Techniques
8. Modeling
3. Eme ging Coun ies
9. Pe o mance
4. Fi m Cha ac e is ics
10. Policy Measu es
5. Indus ial Coun ies
11. Technological Cha ac e is ics
6. Indus ies
12. Technologies
No e: Au ho ’s own in e p e a ion
2 A special issue o he Economics o Inno a ion and New Technology was de o ed o he 20 h anni e sa y o he
CDM model (see Löö e al., 2017).
5
Using his scoping exe cise, we y o unde s and which a e he a eas o he li e a u e ha a e well
esea ched, and which a eas a e less popula in e ms o esea ch conduc ed. The main bene i o
his me hod is ha i has gi en us a eeling o “ he lay o he land”.
In o de o isualize his classi ica ion, we ha e used ne wo k analy ical ools such as Pajek
(Ba agelj and M a , 2002) and VOS iewe 3. The isualiza ion in Figu e 1 shows ha he li e a u e
ocuses o emos on Inno a ion Cha ac e is ics and Pe o mance, as well as Fi m Cha ac e is ics.
The li e a u e also ocuses on Eme ging Coun ies and, ob iously, on Modeling and Econome ic
Techniques. Al hough we ha e no used his classi ica ion as a guidance ool o ou li e a u e
e iew, some o he classes a e consis en wi h some o he e iew’s headings. Fo ins ance, he
class “Eme ging Coun ies” eappea s unde heading 2.1 “Ex ension o mo e coun ies”, he class
“Econome ic Techniques” unde heading 2.2 “Al e na i e es ima ion me hods”, and “Inno a ion
Cha ac e is ics” unde heading 2.3 “Addi ional endogenous inno a ion inpu s”.
Figu e 1.Inno a ion and P oduc i i y: A isualiza ion o he classifica ion
No e: Ne wo k calcula ions using Pajek and isualized using VOS iewe .
In he end, we ha e chosen o o ganize he li e a u e along he models used o analyze he
ela ionship be ween inno a ion and p oduc i i y. Fi s , we go o e he ex ensions b ough o he
3 See: VOS iewe : Visualizing Scien ific Landscapes, h ps://www. os iewe .com/
6
o iginal CDM model and p esen he esul s ob ained ega ding complemen a i y be ween ypes o
inno a ion, causali y, links wi h o he endogenous a iables on he inpu o he ou pu side o
inno a ion, and he e ogenei y ega ding inno a ion ypes and business en i onmen s. Secondly,
we e iew he s udies ha ha e in oduced s ochas ic elemen s in he CDM model. Thi dly, we
p esen some es ima es o he a e age ea men e ec o inno a ion om s udies ha look a
causali y h ough o he me hods han ins umen al a iables. We close by discussing a enues o
u u e esea ch.
2. CDM model
The o iginal pu pose o he CDM model was o en ich he es ima ion o he a e o e u n o R&D
– a e u n in e ms o p oduc i i y - by es ima ing a ecu si e model whe e R&D leads o
inno a ion ou pu (using pa en coun da a o da a on sales o inno a i e p oduc s om Inno a ion
Su eys), and inno a ion ou pu a ec s p oduc i i y. O e he las en yea s he o iginal CDM
model was gene alized in di e en di ec ions: ex ension o mo e coun ies and o panel da a (2.1),
al e na i e es ima ion me hods (2.2), addi ional endogenous inno a ion inpu s besides R&D (2.3),
in oduc ion o di e en ypes o inno a ion ou pu (2.4), addi ional endogenous de e minan s o
p oduc i i y besides inno a ion ou pu (2.5), he e ogenei y in he links be ween inno a ion ou pu
and p oduc i i y (2.6), and he inclusion o spillo e s (2.7).
2.1 Ex ension o mo e coun ies and o panel da a
Thanks o new inno a ion su eys, especially in de eloping coun ies, he model has been
es ima ed on coun ies o egions hi he o no examined. Mo eo e , he Wo ld Bank En e p ise
Su eys and he BEEPS4 da ase s join ly se up by he Wo ld Bank and he Eu opean Bank o
Recons uc ion and De elopmen con ain many addi ional s a is ics on he business en i onmen ,
which can be used as ins umen al o con ol a iables and allow o la ge economic models (wi h
addi ional endogenous a iables). These da a a e, howe e , essen ially c oss-sec ions o epea ed
c oss-sec ions o i m da a in di e en coun ies. In o he coun ies, addi ional inno a ion su eys
ha e allowed o panel da a analysis o he es ima ion o dynamic models (Raymond e al., 2015;
Demmel e al., 2017; Van Leeuwen and Mohnen, 2017; Mo is, 2018; Ta ei a e al., 2019;
Aud e sch and Beli ski, 2020; Aud e sch e al., 2020; Edeh and Acedo, 2021; Hoang e al., 2021;
Ga cía-Pozo e al. (2021); B oome e al., 2023).
In gene al, he posi i e link be ween inno a ion and p oduc i i y has been con i med (mo e on his
la e ). Howe e , in s udies based on c oss-sec ional da a i is ha d o speak o causali y. As Mo is
(2018) illus a es, c oss-sec ional da a, which do no allow o co ec o indi idual e ec s, end o
o e es ima e he link be ween inno a ion and p oduc i i y. A ew s udies ha e used lagged
inno a ion and cu en p oduc i i y o es whe he he e is a leas a G ange causali y om
4 Business En i onmen and En e p ise Pe o mance Su eys
13
as an inno a ion inpu . Pe e s e al. (2017b) ind no complemen a i y be ween p oduc and p ocess
inno a ion in Ge many. Aldie i e al. (2021) conside he 4 ypes o inno a ion (p oduc , p ocess,
o ganiza ional and ma ke ing) and a ious combina ions o hem, bu do no eally es o
complemen a i y be ween hem. Ins ead, hey concen a e on possible complemen a i ies be ween
inno a ion ypes and human capi al, physical capi al o ICT. Mohnen e al. (2021) es ima e a model
ha simul aneously de e mines p oduc i i y g ow h and he bina y choices o R&D, o ganiza ional
inno a ion and ICT ha lead o ha p oduc i i y g ow h on Du ch i m panel da a. They ind signs
o complemen a i y be ween R&D and o ganiza ional inno a ion and be ween ICT and R&D, bu
no be ween ICT and o ganiza ional inno a ion. Va ious easons can explain hese di e gen
esul s. Fi s , he ac ha o en di e en ypes o inno a ion a e conduc ed simul aneously does
no necessa ily imply ha he join inno a ions lead o be e pe o mance. Second,
complemen a i ies may di e depending on he pe o mance one is ocusing on. Thi d,
complemen a i ies a e o en examined on he basis o occu ence o non-occu ence o an
inno a ion ins ead o inno a ion in ensi ies. Fou h, he analysis o complemen a i y ge s
complica ed by he ac ha he inno a ion choices a e hemsel es endogenous. In he u u e i
would ce ainly help o ha e la ge da ase s wi h su icien a ia ions in he combina ions o
inno a ion ypes o each mo e signi ican conclusions.
2.5 Addi ional endogenous de e minan s o p oduc i i y besides inno a ion ou pu
Besides hese di e en inno a ion ou pu s, some o he endogenous a iables (i.e. wi h a sepa a e
equa ion explaining hei de e minan s) ha e been in oduced in he p oduc i i y equa ion.8
F iesenbichle and Penede (2016) add o he CDM model a compe i ion equa ion, wi h
compe i ion depending on inno a ion and a ec ing p oduc i i y and inno a ion expendi u es.
They ind ha he e is an in e ed-U ela ionship be ween inno a ion and R&D, ha inno a ion
educes compe i ion and ha bo h compe i ion and inno a ion a e posi i ely ela ed o
p oduc i i y.
Sega a-Blasco e al. (2022) add wo expo equa ions o he CDM model, one ha explains he
sha e o expo s in o al sales and he o he one he numbe o ma ke s in which he i m expo s.
On da a om 7 Eu opean coun ies, he au ho s ind ha i ms wi h a mo e ex ensi e and in ensi e
p esence on o eign ma ke s a e mo e likely o do R&D and o inno a e and a e mo e R&D-
in ensi e. The au ho s also con i m he selec ion hypo hesis acco ding o which mo e p oduc i e
i ms a e he ones ha a e mo e in ensi ely and ex ensi ely ac i e on o eign ma ke s.
8 Ba elsman e al. (2019) ha e added he p opo ion o employees wi h b oadband in e ne connec ion o he
a ious inno a ion ou pu s in explaining o al ac o p oduc i i y. On e idence om 10 Eu opean coun ies, hey find
ha ICT is an impo an con ibu o o p oduc i i y and diminishes he significance o he co ela ion o inno a ions
and p oduc i i y. Bu hey do no handle he endogenei y o ei he inno a ion ou pu s o o b oadband connec ions.
Gogokhia and Be ula a (2021) cons uc a business en i onmen index on he basis o pe cei ed cons ain s by fi ms
o doing business in 28 ansi ion economies. An imp o ed business en i onmen is ound o be co ela ed
significan ly wi h R&D, inno a ion ou pu and p oduc i i y, bu he index is ea ed as exogenous.
14
Mason e al. (2020) es ima e by 3SLS on a panel da ase o se en indus ies in 8 coun ies a
ecu si e sys em o h ee equa ions: openness o o eign ade and FDI, he g ow h o pa en s pe
hou wo ked, and mul i ac o p oduc i i y g ow h. Skills and hei in e ac ions wi h openness and
p oximi y o he on ie en e esp. he second and hi d equa ion. In gene al, skills a e posi i ely
ela ed o pa en lows and di ec ly and indi ec ly o p oduc i i y g ow h, while pa en lows a e
posi i ely ela ed o p oduc i i y g ow h. Siedschlag and Zhang (2015) compa e o I eland he
impac on inno a ion and on p oduc i i y o domes ic non-expo e , domes ic expo ing i ms and
o eign-owned i ms. Fi ms ha ope a e in e na ionally a e mo e inno a i e han domes ic non-
expo ing i ms. Fo eign-owned i ms a e mo e p oduc i e han domes ic i ms.
Ramí ez e al. (2020) enla ge he CDM model by adding an endogenous human capi al a iable,
measu ed by he numbe o R&D wo ke s in he o al numbe o wo ke s, in he R&D and
inno a ion equa ions. On Colombian da a hey ob ain signi ican ma ginal e ec s o human capi al
on R&D and inno a ion. They also use wo speci ic kinds o human capi al, namely he pe cen age
o echnicians and p o essionals, espec i ely pos g adua es, wo king in R&D. Whe eas he
p opo ion o echnicians and p o essionals has a g ea e impac han he p opo ion o
pos g adua es on R&D and inno a ion, inno a ion has a g ea e e ec on p oduc i i y wi h he
pos g adua es.
C espi e al. (2016) in oduce IPR no among he condi ions leading o an inno a ion, bu di ec ly
as a measu e o inno a ion ou pu . On La in-Ame ican i m da a hey ind ha i ms ha apply o
IPR ha e a highe p oduc i i y han hose ha do no .
These addi ional links be ween he endogenous a iables en ich he models by in oducing di ec
and indi ec e ec s and eedback loops. The di icul y is o ind exogenous a iables ha d i e he
endogenous a iables and exclusion es ic ions o o he ways o iden i y he s uc u al o m
pa ame e s. Mo e wo k in his di ec ion would p o ide a deepe unde s anding o he economic
o ces a play, a possible es ing o di e en causali y s uc u es and a ool o conduc ing
simula ions o he e ec s o policy measu es o unexpec ed sys emic shocks on inno a ion and
p oduc i i y.
2.6 He e ogenei y in he links be ween inno a ion and p oduc i i y
The CDM model has also been es ima ed on di e en kinds o speci ic da ase s such as mic o i ms
(Aud e sch e al., 2020; Gaglio e al., 2022), small and medium-sized en e p ises (Edeh and Acedo,
2021; Hoang e al., 2021), se ice sec o s (Kasongo e al., 2023), high- ech/low ech sec o s,
manu ac u ing/se ices (Ál a ez e al., 2015; Busom and Vélez-Ospina, 2017), eme ging and
ansi ion coun ies (Ba z-Zuccala e al., 2018), as well as speci ic egions (Ga cia-Pozo e al.,
2021) and indus ies (Wadho and Chaudh y, 2018; F ick e al., 2019). As is o en he case when
wo king wi h i m da a, he e is a lo o he e ogenei y in he link be ween inno a ion and
p oduc i i y, which is no allowed o i he same coe icien s hold o all obse a ions. Baum e
al. (2017) es o mally he homogenei y ac oss di e en indus ies on Swedish panel i m da a. In
15
all h ee pa s o he CDM model hey ejec he homogenei y o key coe icien s. The same is
ound by F ick e al. (2019) on da a om ou Eu opean coun ies and h ee di e en indus ies.
He e a e a ew examples o allowed o he e ogenei ies.
Hashi and S ojčić (2013) compa e he inno a ion-p oduc i i y link be ween Wes e n Eu opean and
Cen al and Eas Eu opean coun ies and do no ind a s a is ically signi ican di e ence, excep
o he in luence o o ganiza ional inno a ion, which is posi i ely ela ed o p oduc i i y in
Wes e n Eu ope and no signi ican in Cen al and Eas Eu opean coun ies. Demmel e al. (2017)
ind a sel -selec ion in o inno a ion on he basis o pas p oduc i i y only o he wo uppe -middle-
income coun ies in hei sample and only o p oduc inno a ion and an inno a ion o p oduc i i y
e ec again d i en by p oduc inno a ion and only o he uppe -middle-income coun ies. They
conclude ha he le el o de elopmen plays a media ing ole in he link be ween inno a ion and
p oduc i i y. C owley and McCann (2018) sepa a e hei sample o i ms o hi een Eu opean
coun ies in o inno a ing coun ies and coun ies s ill in ansi ion be ween e iciency seeking and
inno a ing. They ind in gene al a posi i e link be ween inno a ion and p oduc i i y excep in he
manu ac u ing sec o s o coun ies in ansi ion be ween e iciency seeking and inno a ion. Ba z-
Zuccala e al. (2018) compa e low-income coun ies (in e ms o g oss na ional income pe capi a)
wi h high-income coun ies. In bo h g oups inno a ion and managemen p ac ices a e posi i ely
ela ed o p oduc i i y, bu in he o me g oup he ma ginal e ec o managemen quali y is a
leas h ee imes as high as he ma ginal e ec o inno a ion, whe eas in he la e g oup inno a ion
has a la ge ma ginal e ec on p oduc i i y han managemen . R&D has a signi ican indi ec e ec
only in he la e g oup. In de eloping coun ies, a g ea deal o GDP and employmen occu s in
he in o mal sec o . Fu e al. (2018) compa e he ole o inno a ion on p oduc i i y in o mal and
in o mal i ms in Ghana. They ind ha o mal i ms do no end o be mo e p oduc i e han
in o mal i ms a e, bu inno a ion ends o be mo e impo an o p oduc i i y in o mal i ms.
Busom and Vélez-Ospina (2017) on Colombian i m da a un a quan ile eg ession o p oduc i i y.
They ind ha inno a ion a ec s p oduc i i y below he median o he p oduc i i y dis ibu ion in
se ice indus ies and mo e pe asi ely ac oss he p oduc i i y dis ibu ion in manu ac u ing.
Acco ding o hei esul s, inno a ion a ec s p oduc i i y mo e in he bo om o he p oduc i i y
dis ibu ion. C espi e al. (2016) come o he opposi e conclusion on pooled da a om 16 La in
Ame ican coun ies. Liao (2020) on i m panel da a om Spain epo s ha o i ms loca ed in
he lowe quan iles o hei labo p oduc i i y dis ibu ion imi a i e sales (i.e. sales o new- o- he-
i m p oduc s) ha e a g ea e ma ginal e ec on p oduc i i y han inno a i e sales (sales o new-
o- he-ma ke p oduc s) and he opposi e holds o i ms in he uppe quan iles o hei p oduc i i y
dis ibu ion.
Baumann and K i ikos (2016) conclude om hei s udy on Ge man mic o, small and medium-
sized i ms, using a CDM model, ha “o e all, he link be ween R&D, inno a ion, and
p oduc i i y in mic o i ms does no la gely di e om hei la ge coun e pa s”. Hall and Sena
16
(2017) ind ha i ms ha inno a e and use o mal IP a e mo e p oduc i e, and his is he case
mo e in se ices, ade and u ili y sec o s han in manu ac u ing. Hoang e al. (2021) epo ha in
Vie namese SMEs inno a o s a e 23% mo e p oduc i e han non-inno a o s. T an e al. (2022)
add nine indica o s o open inno a ion as de e minan s o inno a ion inpu s and ou pu s, and wo
measu es o economic pe o mance ( o al ac o p oduc i i y and p o i abili y). On Vie namese
i m da a hey con i m ha open inno a ion knowledge managemen has a posi i e e ec on
inno a ion capabili ies, which in u n a ec posi i ely p oduc i i y and p o i abili y. Fo SMEs
hough he e a e hu dles in he acquisi ion o capabili ies o enhance economic pe o mance.
Ramí ez e al. (2020) also ob ain a much highe impac o inno a ion on p oduc i i y o la ge
compa ed o small i ms in Colombia. Piekkola and Rahko (2020) dis inguish be ween high-
ma ke -sha e and low-ma ke -sha e i ms in Finland. P oduc and p ocess inno a ions inc ease
p oduc i i y mo e in low-ma ke -sha e i ms han in high-ma ke -sha e i ms, bu hey a e mo e
p o i able in high-ma ke sha e i ms. The impac o he inancial c isis on inno a ion and
p oduc i i y was also di e en o small and la ge i ms. F om a ield su ey conduc ed in G eece
in 2011 and 2013, Gio opoulos e al. (2023) epo ha he c isis discou aged small i ms o engage
in R&D, while i inc eased he willingness o engage in R&D ac i i ies among he la ge ones.
Knowledge p oduc ion was no a ec ed bu he inno a ion ou pu seemed o imp o e he
p oduc i i y o la ge i ms only.
Despi e di e ences in esul s ac oss s udies and samples, p oduc i i y is ela ed o inno a ion in
de eloping and eme ging coun ies as i is in ad anced coun ies, and in mic o i ms as in la ge
and middle-sized i ms. Di e ences in signi icance o magni ude can be due o di e ences in
sampling, he sample composi ions, he numbe o obse a ions, o he en i onmen s in which
i ms ope a e. I would be ad isable o in es iga e wha he di e ences a e due o, when hey occu ,
and conduc sensi i i y analyses o ind ou whe he he di e ences a e due o he da a, econome ic
p oblems like selec i i y o endogenei y (e o s in a iables, simul anei y o omi ed a iables), o
di e ences in he en i onmen in which i ms ope a e, maybe by pe o ming a me a-analysis.
2.7 Inclusion o spillo e s
A couple o s udies ha e inco po a ed spillo e s in he CDM model. Spillo e s a e conside ed as
exogenous. They can be spillo e s in inno a ion, aining, R&D and FDI, hey can be in e -
indus y o in e na ional spillo e s.
A posi i e ole o inno a ion spillo e s is epo ed in he s udy o C espi e al. (2016): i ms bene i
om p oduc (bu no p ocess) inno a ions and om IPR applica ions o o he i ms in he same
coun y/indus y. Resea ch by Gio anne i and Piga (2017), on UK i m da a, ind ha in e sec o al
R&D and aining spillo e s, in wha hey call passi e sou ces o collabo a ion as opposed o ac i e
sou ces o collabo a ion wi h cus ome s, uni e si ies, supplie s, compe i o s and so on, in luence
posi i ely mos ly p ocess inno a ion, sligh ly o ganiza ional inno a ion and no signi ican ly
p oduc inno a ion, which all h ee ha e a posi i e in luence on he g oss alue added and g oss
17
p o i ma gins. The spillo e s ha e ha dly any e ec on inno a ion inpu s (in e nal R&D, ex e nal
R&D, aining and ad e ising expendi u e o e sales). Howell (2018) adds lea ning a iables
(lea ning by doing, lea ning by expo ing, abso p ion capaci y and media ing ins i u ions) a he
a ious s ages o he CDM model and inds ha o Chinese i ms lea ning spillo e s educe he
incen i es o do R&D bu inc ease he inno a ion ou pu and p oduc i i y.
Aud e sch and Beli ski (2020) look a complemen a i y be ween R&D and R&D spillo e s in
p oducing inno a ion. Knowledge spillo e s measu ed using sec o al R&D and inpu -ou pu
weigh s inc ease co-c ea ed inno a ion, bu no hose made in e nally and hose bough ex e nally,
con ibu ing posi i ely o p oduc i i y. R&D spillo e s a e a be e p edic o o p oduc i i y han
inno a ion. Vujano ić e al. (2022) add FDI spillo e s as addi ional sou ces o knowledge and
di ec de e minan s o inno a ion and p oduc i i y. They es ima e hei enla ged CDM model on
Se bian i m da a. They ind posi i e o eign spillo e e ec s om o eign i ms in he same
indus y and om indus y sales o o eign supplie s mos ly a he inno a ion le el and mo e so
o i ms ha ely on ex e nal knowledge (knowledge use s) han o i ms ha do R&D hemsel es
(knowledge c ea o s).
3 S ochas ic e u ns o R&D and inno a ion
In he adi ion o Olley and Pakes (1996), an al e na i e o he ex ended p oduc ion unc ion
app oach, whe e R&D and/o some measu e o inno a ion is inse ed as an addi ional inpu , was
p oposed by Do aszelski and Jaumand eu (2013). They model p oduc i i y as a s ochas ic shock
known o he i m bu no obse able by he econome ician. I ollows a i s -o de Ma ko
p ocess and depends on he R&D in es men in he p e ious pe iod. In his way, R&D no longe
a ec s p oduc i i y in a linea and de e minis ic manne . Applying hei model o Spanish da a,
he au ho s ind ha he e u n o R&D is highe , he highe is pas p oduc i i y; be ween 25% and
75% o p oduc i i y is unp edic able; he mean expec ed p oduc i i y is highe o R&D-
pe o ming han o non-R&D-pe o ming i ms; p oduc i i y is mo e pe sis en in indus ies
whe e he e is less unce ain y; he ne (o dep ecia ion) a e o e u n o R&D is highe in
indus ies whe e he unce ain y ( he a iance o he s ochas ic pa o he p oduc i i y) is highe ,
hence pa o he e u n o R&D compensa es o he inhe en unce ain y.
Pe e s e al. (2017b) combine he CDM model and he Do aszelski and Jaumand eu (2013) model.
Ins ead o ha ing R&D a ec ing p oduc i i y di ec ly, hey le R&D a ec p oduc i i y h ough
inno a ion ou pu bu in oduce unce ain y in he e ec s o R&D on inno a ion and o inno a ion
on p oduc i i y. The e a e h ee sou ces o s ochas ici y in hei model: one ela es o he cos o
doing R&D, one ega ds he p obabili y o ha ing combina ions o p oduc and p ocess
inno a ions depending on he choice o R&D, and one ela es o he e olu ion o p oduc i i y.
Thei model allows i ms o be inno a i e wi hou doing R&D, and ac ually on Ge man da a hey
ind ha his is he case o 22% o he i ms. The decision o in es in R&D o no is endogenous.
Fi ms in es in R&D i he di e ence in expec ed u u e p o i s by in es ing in R&D o no is
18
highe han he s a up o main enance cos s o doing R&D. They ind ha i ms ha do R&D a e
mo e likely o be inno a i e, bu R&D is no a su icien condi ion o being inno a i e: he
p obabili y o no u ning ou o be inno a i e is 10% in low- ech indus ies and 20% in high- ech
indus ies. The success a e is highe o p oduc han o p ocess inno a ions. The mos likely
ou come is he simul aneous in oduc ion o new p oduc s and new p ocesses. In high- ech
indus ies i is p oduc inno a ion ha inc eases p oduc i i y, whe eas in low- ech indus ies i is
p ocess inno a ion ha ma e s. The e is no sign ha he simul aneous in oduc ion o p oduc and
p ocess inno a ion has any addi ional e ec on p oduc i i y. The e is a high pe sis ence in
p oduc i i y, which implies ha p oduc i e i ms a e mo e likely o in es in R&D. The e is a lo
o he e ogenei y be ween i ms.
Chen, Zhang and Zie (2021) gene alize Pe e s e al. (2017b) in ha ing inno a ion depend on bo h
R&D (a con inuous a iable ep esen ing aci knowledge) and pa en s (bina y a iables
ep esen ing codi ied knowledge). R&D a ec s p oduc i i y di ec ly and in combina ion wi h wo
ypes o pa en s: in en ions and u ili y models. They es ima e hei model on Chinese pa en and
i m da a o high- ech manu ac u ing i ms. They ind ha o hei sample o Chinese i ms; 1)
he a e o e u n on R&D in e ms o inc ease in i m alue is low: 0.22%; 2) he la ges pa o
ha e u n comes om aci knowledge, i.e. no h ough pa en ing; 3) on a e age an in en ion
pa en causes a 0.76 pe cen inc ease in i m alue, a u ili y model a 0.66 pe cen inc ease; 4) he
s a -up cos s o R&D a e a ound en imes as la ge as he main enance R&D cos s.
Pe e s e al. (2022) apply he Pe e s e al. (2017b) s uc u al R&D/p oduc and p ocess
inno a ions/p oduc i i y model o Ge man expo ing and non-expo ing i ms sepa a ely. They
ob ain a highe a e o e u n on R&D o expo ing i ms, which may explain he highe R&D
in es men s o expo ing i ms. The pe sis en ly highe impac o inno a ion on p oduc i i y o
expo ing i ms explains hei highe a e o e u n o R&D. In ano he applica ion o he Pe e s e
al. (2017b) model, Pe e s e al. (2017a) make he cos o doing R&D depend on he inancial
s eng h o he i m. The ind ha “ inancially s ong i ms ha e a highe p obabili y o gene a ing
inno a ions om hei R&D in es men , and he inno a ions ha e a la ge impac on p oduc i i y
and p o i s”.9
4 Causali y o inno a ion based on andomized o quasi- andomized expe imen s
In many a eas o economics, quasi-expe imen s a e un o assess he causali y o ce ain ea men s
on measu es o economic pe o mance. This app oach has also s a ed o be ollowed in he
empi ical li e a u e on inno a ion and p oduc i i y.
9 Ro h e al. (2023) do no qui e es ima e he Pe e s e al. (2017b) model bu add a ious in angible in es men s in
he Ma ko p ocess de e mining s ochas ic p oduc i i y. They show ha in Ge many in angibles ha e a posi i e
effec on p oduc i i y, o en a abo e he effec o only R&D. In angibles include, besides R&D, aining, ad e ising
and ma ke ing, design and licenses, and so wa e and da abases.
19
Bloom e al. (2013) un a andomized expe imen s wi h 17 la ge Indian ex ile i ms p o iding
consul ing on managemen p ac ices o 6 andomly chosen i ms, he o he ones being he con ol
g oup. The au ho s ind ha he i ms ha adop ed he managemen p ac ices, a o m o
o ganiza ional inno a ion, a e 17% mo e p oduc i e in he i s yea han hei coun e pa s in he
con ol g oup and inc ease in size h ee yea s a e he expe imen . Mul idimensional p opensi y
sco e ma chings a e used by Coad e al. (2016) o in es iga e he non-inno a ion and p oduc i i y
link. They ma ch i ms by he p obabili y o encoun e mul iple ba ie s o inno a ion and ind ha
he a e age ea men e ec on he ea ed in e ms o p oduc i i y a e nega i e and signi ican o
mos ba ie s. In o he wo ds, ba ie s o inno a ion p e en i ms om inno a ing and a ec
nega i ely hei p oduc i i y. D’A oma and Pacei (2018) use he EU-EFIGE/B uegel-Unic edi
da ase o 7 Eu opean coun ies and a gene alized p opensi y sco e app oach o a es he e ec
o p oduc inno a ion in ensi y on p oduc i i y. They do no ind a s a is ically signi ican e ec
o inno a ion in ensi ies on p oduc i i y. Dai and Sun (2021) ob ain an a e age ea men e ec o
inno a ion on p oduc i i y o i ms in China on he basis o a p opensi y sco e ma ching wi h
di e ence in di e ence. Inno a ing i ms a e ma ched o non-inno a ing i ms wi h simila
p opensi y sco es o being inno a i e and hen i s di e ences o p oduc i i y a e compa ed o
he ea ed i ms and he con ol g oup. The au ho s ind ha inno a ing i ms a e 11% mo e
p oduc i e han compa able non-inno a ing i ms. Asiedu e al. (2023) use a c oss-sec ion o i e
Cen al Ame ican coun ies (Ecuado , El Sal ado , Gua emala, Hondu as, and Nica agua) and
compa e he p oduc i i y o i ms wi h and wi hou p oduc and p ocess inno a ions on he basis
o a p opensi y sco e ma ching. P oduc i i y is he ou come a iable, p oduc and p ocess
inno a ion a e he ea men a iables. They epo ha inno a i e i ms a e mo e p oduc i e:
“ i ms ha did no in oduce signi ican ly new p oduc s in he las 3 yea s a e 23.3% less
p oduc i e han i ms ha did. Also, i ms ha did no in oduce new echnology in he pas 3 yea s
a e 25.2% less p oduc i e han i ms ha in oduced signi ican ly imp o ed echnology”. Fo a
pa icula ype o inno a ion, namely AI pa en s, Alde ucci e al. (2020) ind ha AI-pa en ing
i ms a e mo e p oduc i e. They each his conclusion a e ma ching AI pa en ing i ms wi h non-
AI pa en ing i ms in he U.S. (by age, mul i-uni s a us, p ima y indus y, p ima y s a e, size) and
pe o ming an e en s udy.
5 Discussion and conclusion
The e idence ga he ed in he s udies published o e he ime span 2013-2023 eaches he ollowing
conclusions. P oduc i i y is posi i ely ela ed o inno a ion, be i in small o in la ge i ms, o mal
o in o mal i ms, ad anced and eme ging coun ies. Besides R&D, o he inno a ion expendi u es,
ICT, AI, managemen p ac ices, public suppo o inno a ion, collabo a ions, and p o ec ion o
in ellec ual p ope y ha e been shown o s imula e inno a ion a he i m le el. P oduc
inno a ions a e no always posi i ely associa ed wi h p oduc i i y, excep in he mos ad anced
coun ies, high- ech sec o s and p oduc i e i ms. Many s udies conclude ha p oduc i i y depends
20
mo e on p ocess han on p oduc inno a ions. Bu he ela i e e ec s o di e en ypes o
inno a ion a e also sec o -speci ic. I seems ha in se ices non- echnological inno a ions ha e
mo e o an e ec on p oduc i i y han echnological inno a ions and p ocess inno a ions mo e
han p oduc inno a ions. En i onmen al inno a ions do no always inc ease p oduc i i y, bu AI
seems o be an impo an d i e no jus o inno a ion bu also o p oduc i i y. The e is no much
sign o complemen a i y be ween he di e en kinds o inno a ion, bu mo e be ween inno a ion
and ou side condi ions such as human capi al, a a o able in es men clima e and compe i ion.
I is ha d o ank he s eng h o he ela ionship be ween inno a ion and p oduc i i y ac oss i m
size, le els o de elopmen , he o mal and in o mal na u e o i ms, and indus ial sec o s,
especially when hese cha ac e is ics in e ac . Ye , i seems ha p oduc inno a ion, especially
d as ic p oduc inno a ion, and o ganiza ional inno a ion a e mo e s ongly linked o p oduc i i y
in ad anced coun ies, high- ech indus ies and in i ms o he o mal sec o . In low- ech indus ies
p ocess inno a ions a e mo e impo an han p oduc inno a ions and in eme ging coun ies
imi a i e inno a ions mo e han d as ic inno a ions.
The s udies we ha e examined in his su ey we e no cen e ed on he e ec i eness and e icacy
o pa icula inno a ion policies. I any policy conclusion can be d awn om he e idence we ha e
ga he ed i would be o encou age inno a ion since i is ela ed o p oduc i i y (possibly in a bi-
causal way), de elop he capaci ies o inno a e, emo e ba ie s o inno a ion such as egula ions
and access o inance, accep he possibili y ha inno a ions can ail, and pu sue an open-doo
policy, as o eign ade ansmi s knowledge and a he same ime c ea es a compe i i e
en i onmen , whe e only he bes can su i e.
On he modeling side, i has been a pe iod o consolida ion o he CDM model as oughly h ee
qua e s o he wo k has adop ed o ex ended he CDM model. Ex ensions include o he inno a ion
inpu s besides R&D, o he de e minan s o p oduc i i y besides inno a ion, g anula i y in he ypes
o inno a ion and hei complemen a i y and ex ension o eme ging coun ies and mic o i ms. An
impo an gene aliza ion o he CDM model has been he inco po a ion o unce ain y in ge ing
inno a ion ou pu om R&D and in ge ing p oduc i i y imp o emen s om inno a ion. Finally,
some ecen pape s ha e used expe imen s o quasi-expe imen s ins ead o ins umen s o unco e
a possible causali y be ween inno a ion and p oduc i i y.
To w ap up ou e iew o he s a e o he a o he li e a u e on inno a ion and p oduc i i y we
discuss a ew di ec ions in which we hink p og ess could be made.
Inno a ion Su ey da a a e s ill use ul o benchma king, bu o he ad ancemen o ou
unde s anding o inno a ion new su eys, maybe su eys speci ic o ce ain sec o s o he
economy, ough o be conduc ed. As mo e da a become a ailable o pa icula indus ies,
geog aphical egions and echnological ajec o ies, s udies can be conduc ed a a mo e g anula
le el. In his way, we could unco e he easons o he e ogenei y be ween inno a ion and
21
p oduc i i y ac oss sec o s, echnologies, egions and ypes o i ms. The links be ween inno a ion
and p oduc i i y could be di e en o pa icula echnologies.
As he Inno a ion Su eys a e epea ed, la ge panel da ase s will become a ailable. The use o
panel da a allows o accoun o unobse ed indi idual e ec s - wi hou which he p oduc i i y
e ec s o inno a ion a e p obably o e es ima ed -, o so en some o he endogenei y p oblems by
using lagged explana o y a iables - al hough he use o lagged a iables is no he ubiqui ous
solu ion o endogenei y - and o allow o dynamics in he link be ween inno a ion and
p oduc i i y.
Mo e use could be made o a ailable web-based da a. Mo e and mo e in o ma ion on esea ch
endea o s, new echnologies, echniques and codi ied know-how is a ailable on he web. These
da a would ha e o be o ma ed, s anda dized in some way, bu wi h he ad en o a i icial
in elligence his should be inc easingly easy o do. These da a would be mo e imely han su ey
da a om s a is ical agencies.
As we ha e men ioned, he CDM model has been ex ended o o he measu es o economic
pe o mance. I has, o ins ance, been no iced ha expo s, he use o ICT, AI, and inno a ion
spillo e s a e acili a o s o inno a ion and indi ec ly o p oduc i i y. Especially in eme ging
coun ies, inno a ion ope a es no jus h ough R&D bu e y much h ough impo ed
echnologies, imi a ion and echnology adop ion, in o he wo ds he “buy” besides he “make”.
Inno a ion may go hand in hand wi h a shi in he composi ion o employmen , ou sou cing and
o sho ing o ce ain ac i i ies. Only la ge models can ake hese nuances in o accoun .
Adding unce ain y in he ela ionship be ween inno a ion and p oduc i i y has been maybe he
mos inno a i e aspec o he li e a u e in he las 10 yea s. Mo e wo k could, and p obably will,
be done ex ending he Pe e s e al. (2017b) model o con inuous a iables and o o he dimensions
o inno a ion. Especially in he a ea o inno a ion, i is impo an o inco po a e he s ochas ic
elemen and i s e ec on decision-making. Linked o he impe ec in o ma ion is he ole o
compe i ion be ween incumben s and newcome s, especially in he wo ld o digi al echnologies.
How much u he we can build up s uc u al models depends on he da a a ailable.
A majo di icul y wi h he mic o da a a ailable om R&D, p oduc ion and inno a ion su eys is
he endogenei y o many, i no mos , o he a iables, such as expo ing, collabo a ion, aining,
and public suppo o inno a ion. Consequen ly, i is di icul o ind uly exogenous explana o y
a iables and sound ins umen al a iables o ea he endogenei y p oblem. An al e na i e ha
has s a ed o be de eloped also in his li e a u e is he use o quasi-expe imen s o e en na u al
expe imen s o be able o assess causali ies. A las wo d o cau ion o end his su ey. I has o en
been s a ed ha inno a ion leads o highe p oduc i i y. S ic ly speaking i is da ing o make such
a claim when wo king wi h c oss-sec ional da a o e en wi h panel da a and s a ic models.
22
Re e ences
Aboal, D. & Ga da, P. (2016). Technological and non- echnological inno a ion and p oduc i i y
in se ices is-à- is manu ac u ing sec o s. Economics o Inno a ion and New Technology,
25(5), 435-454. h ps://doi.o g/10.1080/10438599.2015.1073478
Aboal, D. & Tacsi , E. (2018). Inno a ion and p oduc i i y in se ices and manu ac u ing: he
ole o ICT. Indus ial and Co po a e Change, 27 (2), 221–241.
h ps://doi.o g/10.1093/icc/d x030
Acke be g, D. A., Ca es, K. & F aze , G. (2015). Iden i ica ion p ope ies o ecen p oduc ion
unc ion es ima o s. Econome ica, 83(6), 2411–2451. h ps://doi.o g/10.3982/ec a13408
Alde ucci, D., B ans e e , L., Ho y, E., Runge, A. & Zolas, N. (2020). Quan i ying he impac o
AI on p oduc i i y and labo demand: E idence om U.S. census mic oda a. In Pape p esen ed
a he Allied Social Science Associa ions—ASSA 2020 Annual Mee ing.
Aldie i, L., Ba a, C., Vinci, C.P., & Zo i, R. (2021). The join impac o di e en ypes o
inno a ion on i m's p oduc i i y: E idence om I aly. Economics o Inno a ion and New
Technology, 30(2), 151-182. h ps://doi.o g/10.1080/10438599.2019.1685211
Ál a ez R., B a o-O ega, C., & Zahle , A. (2015). Inno a ion and P oduc i i y in Se ices:
E idence om Chile. Eme ging Ma ke s Finance and T ade, 51(3), 593-611.
h ps://doi.o g/10.1080/1540496X.2015.1026696
Asiedu, M., E ah, N.A.A., & A hu , B. (2023). Inno a ion and Fi m P oduc i i y in Cen al
Ame ica. Jou nal o he Knowledge Economy, o hcoming. h ps://doi.o g/10.1007/s13132-023-
01321-w
Aud e sch, D. B., & Beli ski, M. (2020). The ole o R&D and knowledge spillo e s in
inno a ion and p oduc i i y. Eu opean Economic Re iew, 123, 103391.
h ps://doi.o g/10.1016/j.eu oeco e .2020.103391
Aud e sch, D.B., K i ikos, A.S., & Schie sch, A. (2020), Mic o i ms and inno a ion in he
se ice sec o . Small Business Economics, ol. 55(4), 997-1018. h ps://doi.o g/10.1007/s11187-
020-00366-4
Ba elsman, E., Falk, M., Hags en E., & Polde M. (2019), P oduc i i y, echnological
inno a ions and b oadband connec i i y: i m-le el e idence om 10 Eu opean coun ies.
Eu asian Business Re iew, 9, 25-48. h ps://doi.o g/10.1007/s40821-018-0113-0
Ba z-Zuccala, W., Mohnen, P., & Schweige , H. (2018). The ole o inno a ion and managemen
p ac ices in de e mining i m p oduc i i y g ow h in de eloping economies. Compa a i e
Economic S udies, 60(4), 502-530. h ps:// doi.o g/10.1057/s41294-018-0075-3
29
OECD/Eu os a (2018). Oslo Manual 2018: Guidelines o Collec ing, Repo ing and Using
Da a on Inno a ion, 4 h Ed., OECD. h ps://doi.o g/10.1787/9789264304604-en
Olley, S., & Pakes, A. (1996). The Dynamics o p oduc i i y in he elecommunica ions
equipmen indus y. Econome ica, 64, 1263-1297. h ps://doi.o g/10.2307/2171831
Pe e s, B., Robe s, M.J., & Vuong, V. A. (2017a). Dynamic R&D choice and he impac o he
i m’s inancial s eng h. Economics o Inno a ion and New Technology, 26 (1–2), 134–149.
h ps://doi.o g/10.1080/10438599.2016.1202516
Pe e s, B., Robe s, M.J., & Vuong, V. A. (2022). Fi m R&D in es men and expo ma ke
exposu e. Resea ch Policy, 51(10), 104601. h ps://doi.o g/10.1016/j. espol.2022.104601
Pe e s, B., Robe s, M.J., Vuong, V.A., & F yges, H. (2017b). Es ima ing dynamic R&D choice:
an analysis o cos s and long- un bene i s. RAND Jou nal o Economics, 48(2), 409–437.
h ps://doi.o g/10.1111/1756-2171.12181
Piekkola, H., & Rahko, J. (2020). Inno a i e g ow h: he ole o ma ke powe and nega i e
selec ion. Economics o Inno a ion and New Technology, 29(6), 603-624.
h ps://doi.o g/10.1080/10438599.2019.1655878
Ramí ez S., Gallego, J., & Tamayo, M. (2020). Human capi al, inno a ion and p oduc i i y in
Colombian en e p ises: a s uc u al app oach using ins umen al a iables. Economics o
Inno a ion and New Technology, 29(6), 625-642.
h ps://doi.o g/10.1080/10438599.2019.1664700
Ramme , C., Fe nández, G., & Cza ni zki, D. (2022). A i icial in elligence and indus ial
inno a ion. E idence om Ge man i m-le el da a. Resea ch Policy, 51, 104555.
h ps://doi.o g/10.1016/j. espol.2022.104555
Raymond, W., Mai esse, J., Mohnen, P., & Palm, F. (2015). Dynamic models o R&D,
inno a ion and p oduc i i y: panel da a e idence o Du ch and F ench manu ac u ing. Eu opean
Economic Re iew, 78, 285–306. h ps://doi.o g/10.1016/j.eu oeco e .2015.06.002
Rochina-Ba achina, M. E., Máñez, J.A., & Sanchis-Llopis, J.A. (2010). P ocess inno a ions and
i m p oduc i i y g ow h. Small Business Economics, 34(2), 147–166.
h ps://doi.o g/10.1007/s11187-008-9110-5
Ro h F., Sen, A., & Ramme , C. (2023). The ole o in angibles in i m-le el p oduc i i y –
e idence om Ge many. Indus y and Inno a ion, 30(2), 263-285.
h ps://doi.o g/10.1080/13662716.2022.2138280
Sega a-Blasco, A., Te uel, M., & Ca a uzzo, S. (2022). Inno a ion, p oduc i i y and lea ning
induced by expo ac oss Eu opean manu ac u ing i ms. Economics o Inno a ion and New
Technology, 31(5), 387-415, h ps://doi.o g/10.1080/10438599.2020.1823673
30
Siedschlag I., & Zhang, X. (2015). In e na ionalisa ion o i ms and hei inno a ion and
p oduc i i y, Economics o Inno a ion and New Technology, 24(3), 183–203.
h ps://doi.o g/10.1080/10438599.2014.918439
Ta ei a, J. G., Gonçal es, E., & F eguglia, R.D.S. (2019). The missing link be ween inno a ion
and pe o mance in B azilian i ms: A panel da a app oach. Applied Economics, 51(33), 3632–
3649. h ps://doi.o g/10.1080/00036846.2019.1584374.
Tello, M. D. (2015). Fi ms’ inno a ion, public inancial suppo , and o al ac o p oduc i i y:
The case o manu ac u es in Pe u. Re iew o De elopmen Economics, 19(2), 358–374.
h ps://doi.o g/10.1111/ ode.12147
Te do ski, D., Tose ska-T pce ska, K., & Disoska, E.M. (2017). Wha is he ole o inno a ion
in p oduc i i y g ow h in Cen al and Eas e n Eu opean coun ies? The Economics o T ansi ion,
25(3), 527–551. h ps://doi.o g/10.1111/eco .12121
Toshe ska-T pche ska, K., Disoska, E., Te do ski, D., & S ojkoski, V. (2019). The impac o a
c isis on he inno a ion sys ems in Eu ope: E idence om he CIS10 inno a ion su ey.
Eu opean Re iew, 27(4), 543–562. h ps ://doi.o g/10.1017/S1062 79871 90002 18.
T an H.T., San a elli, E., & Wei, W.X. (2022). Open inno a ion knowledge managemen in
ansi ion o ma ke economy: in eg a ing dynamic capabili y and ins i u ional heo y. Economics
o Inno a ion and New Technology, 31(7), 575-603,
h ps://doi.o g/10.1080/10438599.2020.1841942
Tuncel, C.O., & Ok ay, D. (2022). Inno a ion and p oduc i i y in Tu kish manu ac u ing i ms.
Applied Economics Le e s, 29(17), 1610-1614. h ps://doi.o g/10.1080/13504851.2021.1950904
Ugu , M., T ushin, E., Solomon, E., & Guidi, F. (2016). R&D and p oduc i i y in OECD i ms
and indus ies: A hie a chical me a- eg ession analysis. Resea ch Policy, 45, 2069-2086.
h ps://doi.o g/10.1016/j. espol.2016.08.001
Ugu , M., & Vi a elli, M. (2021). Inno a ion, i m su i al and p oduc i i y: he s a e o he a ,
Economics o Inno a ion and New Technology, 30(5), 433-467.
h ps://doi.o g/10.1080/10438599.2020.1828509
Véganzonès-Va oudakis, M., & Plane, P. (2019). Inno a ion, expo s, p oduc i i y and
in es men clima e; a s udy based on Indian manu ac u ing i m-le el da a. Applied Economics,
51(41), 4455–4476. h ps://doi.o g/10.1080/00036846.2019.1591606
Vujano ić, N., Radoše ić, S., S ojčić, N., Hisa ciklila , M., & Hashi, I. (2022). FDI spillo e
e ec s on inno a ion ac i i ies o knowledge using and knowledge c ea ing i ms: E idence
om an eme ging economy. Techno a ion, 118, 102512.
h ps://doi.o g/10.1016/j. echno a ion.2022.102512
31
Wadho, W., & Chaudh y, A. (2018). Inno a ion and i m pe o mance in de eloping coun ies:
he case o Pakis ani ex ile and appa el manu ac u e s. Resea ch Policy, 47(7), 1283–1294.
h ps://doi.o g/10.1016/j. espol.2018.04.007
Zhang, M., & Mohnen, P. (2022). R&D, inno a ion and su i al o i ms in Chinese
manu ac u ing, 2000-2006. Eu asian Business Re iew, 12, 59-95.
h ps://doi.o g/10.1007/s40821-021-00200-1
32
Annex 1. Sea ch s a egy and Jou nal Se
Sea ch s a egy:
1s s age: Sea ch o Re e ence i le “Resea ch, Inno a ion and P oduc i i y” AND
Re e ence publica ion yea “1998”.
2nd s age: Sea ch Keywo ds “inno a ion” AND “p oduc i i y” AND Publica ion yea >
2019 AND Sou ce i le (see jou nal se below).
Sea ches conduc ed in Scopus and Web o Science.
Jou nal se de ini ion:
Jou nals
• Ame ican Economic Jou nal: Applied Economics
• Ame ican Economic Re iew
• Camb idge Jou nal o Economics
• Economics o Inno a ion and New Technology
• Economic Jou nal
• Eu opean Economic Re iew
• Eu asian Business Re iew
• Indus y and Inno a ion
• Indus ial and Co po a e Change
• In e na ional Jou nal o Indus ial O ganiza ion
• Jou nal o De elopmen Economics
• Jou nal o Economic Li e a u e
• Jou nal o Economic Pe spec i es
• Jou nal o Economic Su eys
• Jou nal o Economics and S a is ics
• Jou nal o E olu iona y Economics
• Jou nal o Indus ial Economics
• Jou nal o Poli ical Economy
• Jou nal o P oduc i i y Analysis
• Jou nal o Technology T ans e
• Jou nal o he Eu opean Economic Associa ion
• Managemen Science
• Qua e ly Jou nal o Economics
• RAND Jou nal o Economics
• Resea ch Policy
• Re iew o Income and Weal h
33
• Small Business Economics
• S uc u al Change and Economic Dynamics
• Technological Fo ecas ing and Social Change
• Techno a ion
• Wo ld De elopmen
Wo king pape se ies
• CEPR
• NBER
• ZEW
The UNU-MERIT WORKING Pape Se ies
2025-01 De elopmen s a egies o he g een hyd ogen economy in eme ging economies
by Fabianna Bacil, An hony Black, Ma ina Domingues, Jun Jin, Rasmus Lema, Glen
Robbins and Sö en Schol in
2025-02 Do global alue chains and local capabili ies ma e o economic complexi y in EU
egions? by R. Boschma, E. He nández-Rod íguez, A. Mo ison and C. Pie obelli
2025-03 Inno a ion and p oduc i i y: he ecen empi ical li e a u e and he s a e o he a
by Jacques Mai esse, Pie e Mohnen and Ad No en
