Vi a elli, Ma co; Pi a, Ma iac is ina; Tani, Massimiliano
Wo king Pape
The ole o business isi s in os e ing R&D in es men
UNU-MERIT Wo king Pape s, No. 2025-010
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: Vi a elli, Ma co; Pi a, Ma iac is ina; Tani, Massimiliano (2025) : The ole o
business isi s in os e ing R&D in es men , UNU-MERIT Wo king Pape s, No. 2025-010, 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/DTMT2556
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/326939
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/
#2025-010
The ole o business isi s in os e ing R&D in es men
Ma co Vi a elli, Ma iac is ina Pi a and Massimiliano Tani
Published 3 Ap il 2025
DOI: h ps://www.doi.o g/10.53330/DTMT2556
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
The ole o business isi s in os e ing R&D in es men
Ma co Vi a elli
1
Depa men o Economic Policy, Uni e si à Ca olica del Sac o Cuo e, Milano, I aly
Maas ich Economic and Social Resea ch Ins i u e on Inno a ion and Technology, UNU-
MERIT, Maas ich , The Ne he lands
IZA, Bonn, Ge many
Email: ma co. i a e[email p o ec ed]
Ma iac is ina Pi a
Depa men o Economic Policy, Uni e si à Ca olica del Sac o Cuo e, Piacenza, I aly
Email: [email p o ec ed]
Massimiliano Tani
School o Business, The Uni e si y o New Sou h Wales, Canbe a, Aus alia
IZA, Bonn, Ge many
Email: [email p o ec ed]
Abs ac
Labo mobili y is conside ed a powe ul channel o acqui e ex e nal knowledge and igge
complemen a i ies in he inno a ion and R&D in es men s a egies; howe e , he ex an
li e a u e has ocused on ei he scien is s’ mobili y o mig a ion o high-skilled wo ke s, while
i ually no a en ion has been de o ed o he possible ole o sho - e m business isi s.
Using a unique and no el da abase o igina ing a coun y/sec o unbalanced panel o e he
pe iod 1998-2019 ( o a o al o 8,316 longi udinal obse a ions), his pape aims o ill his
gap by es ing he impac o BVs on R&D in es men .
Resul s om GMM-SYS es ima es show ha sho - e m mobili y posi i ely and signi ican ly
a ec s R&D in es men s; mo eo e , ou indings indica e - as expec ed - ha he bene icial
impac o BVs is pa icula ly signi ican in less inno a i e coun ies and in less inno a i e
indus ies.
These ou comes jus i y some o m o suppo o BVs wi hin he po olio o he e ec i e
inno a ion policies, bo h a he na ional and local le el.
Keywo ds: Business isi s; labo mobili y; knowledge ans e ; R&D in es men s
Acknowledgemen s
Ma co Vi a elli and Ma iac is ina Pi a acknowledge he suppo by he I alian Minis e o dell’Uni e si à e della
Rice ca (PRIN-2022, p ojec 2022P499ZB: “Inno a ion and labo ma ke dynamics”; p incipal in es iga o :
Ma co Vi a elli; unded by he Eu opean Union – Nex Gene a ion EU, Mission 4, Componen 2, CUP
J53D23004830008). The iews and opinions exp essed a e only hose o he au ho s and do no necessa ily e lec
hose o he Eu opean Union o he Eu opean Commission. Nei he he Eu opean Union no he Eu opean
Commission can be held esponsible o hem.
1
Co esponding au ho
2
1 In oduc ion
Inno a ion policy can be summa ised as he se o s a egies and measu es implemen ed by
go e nmen s o p omo e and suppo inno a ion o he pu pose o enhancing economies’
p oduc i i y and g ow h h ough echnological ad ancemen (Scho and S einmulle , 2018). I s
ele ance has g own o e ime, as he no ion ha scien i ic and echnical knowledge has mo ed
away om i s o iginal cha ac e iza ion as a global public good whose a ailabili y and
ans e abili y ac oss he globe makes i s spa ial loca ion less ele an o accessing i s bene i s.
Unde such app oach (o en adop ed by he mains eam in e na ional economics, based on he
Hecksche -Ohlin-S olpe -Samuelson amewo k), low-income coun ies we e p edic ed no
only o access knowledge de eloped in high-income coun ies wi hou hu dles, bu also o be
able o use i , g ow as e and ‘ca ch up’ wi h hei high-income coun e pa s. His o ically, his
has o en no occu ed, as expe ience has shown ha scien i ic knowledge is ypically localised
and ‘s icky’ (i.e. geog aphically con ained o whe e in es men s in esea ch a e made, see
Heime iks and Boschma, 2014), aci (i.e. no coded, bu embodied in indi iduals, see Polanyi,
1966), pa h-dependen (Da id, 1975, A hu , 1994) and i s ans e is condi ional on he
p esence o ‘abso p i e capaci y’ (Cohen and Le in hal, 1989 and 1990) wi hin knowledge
ecipien s, i.e. p io expe ience in esea ch ac i i ies and adequa e social capabili ies in he
knowledge ans e ’s des ina ion place (see Ab amowi z, 1986; Lee, 2016, 2019 and 2024).
O cou se, as new challenges eme ge alongside he limi a ions o he p e ailing e e ence
model - as is cu en ly he case wi h he Uni ed Na ions’ SDGs
1
- he heo e ical ames
unde pinning inno a ion policy na u ally e ol e. Ye , adap ing he e e ence model may a
1
In mo e ecen imes, inno a ion policy has been asked o ake a mo e p oac i e and expe imen al s ance o shed
ligh on he ans o ma i e changes equi ed o add ess he social and en i onmen al challenges con ained in he
Uni ed Na ions’ Sus ainable De elopmen Goals.
3
imes unde play elemen s ha appea o ha e li le consequence un il an unexpec ed shock may
e eal o he wise.
One such example is he Co id-19 pandemic, which, o a ime, has educed o almos nil
physical in e ac ions be ween people in se e al coun ies. As well-known sou ces o inno a ion
like ela ions be ween p oduce s and supplie s, and p oduce s and inal cus ome s, also occu
h ough people’s in e ac ions, will lowe inno a ion ac i i y ollow ewe in e ac ions due o
he pandemic? I is p obably oo ea ly o obse e such so ecen dynamics in o icial s a is ics,
bu exis ing empi ical e idence suppo s he hypo hesis ha wo k- ela ed mobili y con ibu es
o gene a e and ansmi p oduc i e knowledge, e en when i occu s o e sho pe iods o ime,
as in he case o business isi s ( om now on: BVs). The ele an li e a u e (see nex sec ion)
no es ha sho - e m business isi s a e nei he side e ec s o in e na ional ade and
in es men lows no ‘consump ion’ i ems, as accoun ed o in company books. Can hey
ins ead be iewed, a leas in pa , as a igge ac o in os e ing in es men in knowledge
p oduc ion ac i i ies, such as R&D in es men s? O as a s a egic choice o access undamen al
ex e nal knowledge able o inc ease he expec ed p o i abili y o R&D expendi u es, especially
by o ganisa ions, egions and coun ies cons ained by geog aphy and esou ces a ailabili y?
I so, could some incen i es o BVs be included in inno a ion policies?
Indeed, on he one hand, he ex an ( a e) li e a u e on BVs has in es iga ed hei (indi ec )
impac on p oduc i i y and economic g ow h, bu no (wi h one excep ion, see nex sec ion)
hei di ec e ec on inno a ion ac i i ies (such as R&D in es men s). On he o he hand,
p e ious li e a u e has clea ly unde lined ha inno a ion is cha ac e ized by complemen a i ies
and supe -addi i e e ec s (Milg om and Robe s, 1990 and 1995) which can be conside ed o
be he inne a ionale o o mal coope a i e R&D (Veugele s, 1997; Cassiman and Veugele s,
2002; Piga and Vi a elli, 2003 and 2004) and ‘open inno a ion’ (Chesb ough, 2003;
4
Chesb ough e al. 2006). Unde he same app oach, labo mobili y can be conside ed as an
in es men in accessing in o ma ion and compe ences as i is a powe ul channel o acqui e
ex e nal knowledge and igge complemen a i ies in he inno a ion and R&D in es men
s a egies (B aune hjelm e al., 2020). Ne e heless, p e ious esea ch has de o ed much
a en ion o long- e m labo mobili y - bo h in e ms o scien is s’ mobili y (Geuna, 2015;
Ve gine and Riccaboni, 2021; Lissoni and Miguelez, 2024) and mig a ion o high-skilled
wo ke s (B eschi and Lissoni, 2009; Bose i e al., 2015; Lissoni, 2018; Fassio e al., 2019) -
bu i ually no a en ion has been paid o he possible ole o sho - e m BVs in enhancing
knowledge and p oduc i i y.
We aim o ill hese gaps wi h he p esen s udy, whose pu pose is o empi ically es - in a
c oss-coun y sec o al econome ic amewo k - he e ec o BVs on R&D in es men s by
combining R&D da a om he OECD wi h no el in o ma ion om a p op ie a y da abase
collec ing in e na ional business isi s expendi u es wo ldwide. As be e quali ied in Sec ion
3, in his s udy BVs a e de ined as wo k- ela ed labou mo emen s las ing less han 3 mon hs,
in ol ing no change o esidence and hence gene ally no capped by immig a ion au ho i ies.
Fu he mo e, we will y o assess whe he BVs a e pa icula ly c ucial in coun ies ha a e no
wi hin he club o he R&D wo ldwide champions ( he hypo hesis being ha his channel o
knowledge ans e migh be pa icula ly impo an o coun ies ha - al hough cha ac e ized
by an adequa e abso p i e capaci y – mos need ou side knowledge).
As a p e iew o he indings (see Sec ion 4 o a de ailed discussion), ou esul s highligh ha
BVs do aise R&D in es men s, wi h an elas ici y o abou 4-5% and ha his e ec is
pa icula ly signi ican o coun ies ha a e no R&D-leade s. Howe e , in he cu en policy
deba e BVs a e s ill conside ed as consumable expendi u es a he han an ac i i y mo e akin
o an in es men in accessing and gene a ing p oduc i e knowledge. The e o e, we a gue o a
5
econside a ion o sho - e m labo mobili y as an explici a ge o inno a ion policy unde he
p io ha i s ecogni ion will p omp he de elopmen o sui able na ional and egional
incen i es o p omo e i s occu ence (see Sec ion 5).
The es o he pape is o ganised as ollows. Sec ion 2 p o ides a e iew o he ele an
li e a u e, ahead o he summa y o da a and me hodology (Sec ion 3). The esul s a e discussed
in Sec ion 4, while Sec ion 5 o e s key concluding ema ks and policy conside a ions.
2 The li e a u e
The ex an li e a u e wi hin he domain o he economics o inno a ion ex ensi ely highligh s
he ole o ex e nal knowledge (Sec ion 2.1), bu is ex emely limi ed as a as he ole o BVs
is conce ned (Sec ion 2.2).
2.1 The impo ance o ex e nal knowledge
Inno a ion li e a u e and inno a ion policy ha e e ol ed conside ably o e ime. Ini ially, hey
we e la gely ocused on p omo ing in e nal esea ch and de elopmen (R&D) wi hin na ional
bounda ies, wi h an emphasis on suppo ing speci ic indus ies, companies (“na ional
champions”) o echnologies deemed c i ical o na ional compe i i eness. This app oach was
oo ed in he belie ha go e nmen s could play a c ucial ole in d i ing inno a ion by unding
basic esea ch, p o iding subsidies o R&D ac i i ies, and p o ec ing in ellec ual p ope y
igh s (Nelson, 1959).
Howe e , as he unde s anding o inno a ion p ocesses e ol ed, so did he app oaches o hei
design and implemen a ion. The eme gence o he Na ional Inno a ion Sys ems (NIS)
amewo k in he la e 20 h cen u y ma ked a signi ican shi in he concep ualisa ion o
6
inno a ion policy, as i emphasised he impo ance o in e ac ions be ween he ac o s o an
inno a ion sys em ( i ms, uni e si ies, esea ch ins i u ions, and go e nmen agencies) in
gene a ing new p oduc s and p ocesses. In pa icula , i ecognised ha inno a ion is no a
linea p ocess bu a he a complex, in e ac i e, and sys emic phenomenon
2
ha equi es
coo dina ion and collabo a ion ac oss di e en agen s. Mo eo e , in e nal and ex e nal
knowledge gene a e complemen a i ies ha in u n o igina e supe -addi i e e ec s in e ms o
inno a i e pe o mance (F eeman, 1987; Lund all, 1992; Nelson, 1993).
The de elopmen o inno a ion sys ems a he na ional, egional (Cooke e al., 1997), and
sec o al (Male ba, 2002) le el encapsula es he ene ha scien i ic and echnological
knowledge is cumula i e and pa h-dependen (Da id, 1975; A hu , 1988), con ains impo an
aci elemen s, and does no eely o au oma ically a el o e geog aphical and cul u al
dis ances. Ins ead, i is ‘s icky’ ( on Hippel, 1994), and ypically exis s bo h ou side and wi hin
he success ul inno a o (e.g. Ma ch and Simon, 1958; Mans ield, 1968; Rosenbe g and
S einmulle , 1988).
Howe e , an o ganisa ion’s abili y o ecognise and abso b his ex e nal knowledge, and gain
an edge o e compe i o s as a consequence, depends on i s ‘abso p i e capaci y’ (Cohen and
Le in hal, 1989 and 1990) and on i s ‘dynamic’ capabili ies (Teece e al., 1997): a se o skills,
knowledge, and compe encies ha o ganisa ions de elop and accumula e o e ime. By
in e ac ing, each o ganisa ion lea ns new in o ma ion, p oblems and solu ions, which can be
linked o i s exis ing knowledge s ock. In u n, hese no el linkages expand p oblem-sol ing
capabili ies and skills wi hin indi iduals and o ganisa ions, aising hei e icien abso p ion o
new in o ma ion (Cohen and Le in hal, 1989; Teece e al., 1997), and hei c ea i i y (Shalley
2
The in e disciplina y and mul i- ace ed na u e o inno a ion policy has con inued o da e, encompassing a wide
ange o policy ins umen s and s a egies ha co e bo h supply-side (e.g. R&D unding, ax incen i es, and
suppo o educa ion and skills de elopmen ), and demand-side incen i es (e.g. public p ocu emen , s anda ds,
and egula ions a ge ing inno a i e p oduc s and se ices) (e.g. Edle & Geo ghiou, 2007; Mazzuca o, 2018).
13
and no he numbe o passenge s o ips
10
- hence i ep esen s he agg ega e expendi u e
associa ed wi h BVs o a gi en sec o /coun y/yea and i s e olu ion o e he pe iod co e ed.
Second, hey e e o o e all sho - e m expendi u es, na ional and in e na ional, whe e he
in e na ional componen , due o collec ing p ocedu es, is ypically dominan . The da abase,
which epo s BV expendi u es and he alue o ou pu is, unsu p isingly, comme cially
sensi i e as i is used by ai lines o o ecas and decide hei load capaci y in each coun y o
egional ma ke . The access o his no el and unique da abase was made possible h ough
GBTA's ag eemen o sha e hei in o ma ion a a discoun and inancial suppo om he
Uni e si y o New Sou h Wales awa ded o one o he au ho s.
Ou empi ical s a egy - in a c oss-coun y sec o al amewo k - is based on he me ge o h ee
da ase s: he NBTA/GBTA da abase o BVs desc ibed abo e and he publicly a ailable
OECD-ANBERD and OECD-STAN da abases o economic a iables, including R&D
in es men s. The indus y classi ica ion o bo h da ase s is ISIC Re .4.
The inal sample is an unbalanced (due o OECD missing alues) panel co e ing 30 indus ies
(manu ac u ing and se ices) o 25 coun ies in he 1998-2019 imespan, wi h a o al o 8,316
longi udinal obse a ions. All he mone a y se ies ha e been co ec ed o pu chasing powe
pa i ies, exp essing, a he end, alues in cons an p ices and PPP 2010 US dolla s.
In o de o conside he well-known dynamic and pa h-dependen dimension o R&D
in es men s
11
, we se up a speci ica ion in a Dynamic Panel Da a (DPD) amewo k (1):
10
This means ha a longe a el by a op manage is alued mo e han a sho e a el by a middle-le el manage .
11
See he seminal con ibu ions o A hu (1988 and 1994) and Da id (1975 and 1985) and wha discussed in he
p e ious Sec ions 1 and 2. The pa h-dependen na u e o R&D in es men s calls o he inclusion o he lagged
dependen a iable as in eq.1.
14
𝑙𝑛𝑅𝐷𝑖𝑗𝑡 = 𝛼𝑙𝑛𝑅𝐷𝑖𝑗𝑡−1 + 𝛽1𝑙𝑛𝐵𝑉𝐶𝑖𝑗𝑡 + 𝛽2𝑙𝑛𝐸𝑖𝑗𝑡 + 𝛽3𝑇𝑅𝐴𝐷𝐸𝑗𝑡 + 𝐷𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜈𝑖𝑗𝑡 (1)
wi h:
i (sec o ) = 1,…, 30; j (coun y) = 1,…, 25; ( ime) = 1998,…, 2019; ln = na u al loga i hm
The annual R&D in es men s a he sec o al le el is, he e o e, amed wi hin a dynamic
speci ica ion. The measu e o ou key impac a iable is he en i e business isi s capi al (BVC)
ob ained om he o iginal BVs lows h ough he Pe pe ual In en o y Me hod (PIM)
12
.
Con ols a e E ( he o al numbe o employees pe sec o ) which accoun s o he size o he
indus y, and TRADE in ensi y ( (𝐼𝑚𝑝𝑜𝑟𝑡+𝐸𝑥𝑝𝑜𝑟𝑡)
𝐺𝐷𝑃 ∗100 )
13
, o accoun o in e na ional
exposu e
14
(indeed, his al e na i e channel o echnology ans e migh d i e inno a i e
in es men s po en ially eplacing he ole o sho - e m mobili y o wo ke s; see Acha ya and
Kelle , 2009; B anca i e al., 2024).
BVs a e measu ed as s ocks a he han lows since i is he cumula ed acqui ed knowledge
which may a ec he cu en R&D decisions, a he han he sole con empo aneous low.
Indeed, bo h angible and in angible capi als a e gene ally measu ed in e ms o s ocks in he
ele an inno a ion li e a u e (see Hall and Mai esse, 1995; Pa isi, e al., 2006; O ega e al.,
2014). Mo eo e , using BVs s ock allows emphasizing lea ning and cumula i e pa e ns
ela ed o sho - e m mobili y and so conside ing he mobili y e o a longe and mo e e ec i e
sou ce o inno a ion. Finally, since s ocks inco po a e he accumula ed in es men s in he pas ,
12
𝐵𝑉𝐶0= 𝐵𝑉𝑠0
(𝑔+ 𝛿) ; 𝐵𝑉𝐶𝑡= 𝐵𝑉𝐶𝑡−1(1 − 𝛿)+ 𝐵𝑉𝑠𝑡 whe e BVC is he capi al (s ock); BVs measu e he in es men
low; δ is he dep ecia ion a e o 15% (as BVs - seen as a channel o knowledge acquisi ion - should ha e a as
deg ee o obsolescence, simila in p inciple o he s anda d discoun a e o R&D p oposed by Hall, 2007 and
Hall e al., 2009); inally, g is compu ed as an “ex pos ” 3-yea compound g ow h a e (see Pi a e al., 2023).
13
Ob ained om he Wo ld Bank open da a sou ce.
14
This da a is only a ailable a he coun y-yea le el and so i is epea ed o all indus ies in a gi en coun y in
a gi en yea .
15
he isk o endogenei y is signi ican ly mi iga ed.
The sample composi ion by coun ies is p esen ed in Table 1.
Table 1: Sample composi ion by coun ies
Coun y
Obse a ions
Aus alia
283
Aus ia
477
Belgium
493
Canada
431
Chile
63
Czechia
516
Denma k
232
Finland
412
F ance
332
Ge many
472
G eece
164
Hunga y
324
I eland
205
I aly
510
Japan
319
Ko ea
494
Mexico
432
Ne he lands
129
No way
353
Po ugal
399
Slo akia
147
Spain
440
Sweden
210
Uni ed Kingdom
105
Uni ed S a es
374
To al
8,316
In o de o explo e possible he e ogenei y ac oss na ions, we iden i y wo g oups o coun ies
acco ding o hei inno a ion in ensi y (𝑅&𝐷
𝐺𝐷𝑃) a he mac o le el: he op 5 leade inno a o s
16
(Ko ea, Sweden, Japan, US and Ge many
15
) and non-leade inno a o s ( he emaining 25
coun ies).
The sec o al composi ion o he sample is p esen ed in Table 2.
Table 2: Sample composi ion by indus ies
Indus ies
High (H)
Low (L)
ISIC Re .
4
Obse a ions
Ag icul u e, o es y and ishing
L
01-03
375
Food p oduc s; be e ages
L
10-11
79
Tobacco p oduc s
L
12
69
Tex iles
L
13
217
Wea ing appa el
L
14
195
Lea he and ela ed p oduc s, oo wea
L
15
210
Wood and p oduc s o wood and co k, excep
u ni u e; a icles o s aw and plai ing ma e ials
L
16
344
Pape and pape p oduc s
L
17
341
P in ing and ep oduc ion o eco ded media
L
18
331
Coke and e ined pe oleum p oduc s
L
19
279
Chemicals and chemical p oduc s; pha maceu icals,
medicinal chemical and bo anical p oduc s
H
20-21
323
Rubbe and plas ics p oduc s; o he non-me allic
mine al p oduc s
H
22-23
353
Basic me als
H
24
370
Fab ica ed me al p oduc s, excep machine y and
equipmen
L
25
373
Compu e , elec onic and op ical p oduc s
H
26
346
Elec ical equipmen
H
27
347
Machine y and equipmen n.e.c.
H
28
360
Mo o ehicles, aile s and semi- aile s; o he
anspo equipmen
H
29-30
369
Fu ni u e; o he manu ac u ing; epai and
ins alla ion o machine y and equipmen
H
31-33
362
Cons uc ion
L
41-43
384
Accommoda ion and ood se ice ac i i ies
L
55-56
266
Publishing ac i i ies
L
58
172
Mo ion pic u e, ideo and ele ision p og amme
p oduc ion, sound eco ding and music publishing
ac i i ies; p og amming and b oadcas ing ac i i ies
L
59-60
149
Telecommunica ions
L
61
239
Compu e p og amming, consul ancy and ela ed
ac i i ies; in o ma ion se ice ac i i ies
H
62-63
215
15
The op 5 R&D pe o me s o ou sample we e iden i ied based on he 2019 R&D in ensi y alue, conside ing
ha 2019 is he mos ecen yea o he ime-span o ou analysis.
17
Financial and insu ance ac i i ies
L
64-66
334
Real es a e ac i i ies
L
68
258
P o essional, scien i ic and echnical ac i i ies
H
69-75
256
Adminis a i e and suppo se ice ac i i ies
L
77-82
233
A s, en e ainmen and ec ea ion
L
90-93
167
To al
8,316
No e: ‘High’ R&D in ensi y indus ies (H) include manu ac u ing and non-manu ac u ing indus ies in (High
R&D in ensi y + Medium-high R&D in ensi y + Medium R&D) in ensi y g oups based on OECD axonomy,
while ‘Low’ (L) include manu ac u ing and non-manu ac u ing indus ies in (Medium-low R&D in ensi y +
Low R&D in ensi y) g oups (see Galindo-Rueda and Ve ge , 2016).
The la ge numbe o indus ies, albei wi h an unbalanced dimension, p o ides a
comp ehensi e pic u e o he economic s uc u e o he coun ies analysed. This allows us o
ake in o conside a ion ano he possible sou ce o he e ogenei y, namely he di e en
inno a ion p opensi y ac oss he di e en indus ies (A belo, e al., 2024). The e o e, ollowing
he OECD axonomy (Galindo-Rueda and Ve ge , 2016), we clus e indus ies labelling hem
‘High R&D in ensi y indus ies’ i hey belong o he High R&D in ensi y + Medium-high
R&D in ensi y + Medium R&D in ensi y g oups, while ‘Low R&D in ensi y indus ies’ belong
o he Medium-low R&D in ensi y and Low R&D in ensi y g oups.
Table 3 p esen s desc ip i e s a is ics and co ela ion ma ix o he whole sample. As can be
seen, a posi i e, s a is ically signi ican and ela i ely high in magni ude co ela ion be ween
BVs and R&D eme ges om his e y p elimina y es .
18
Table 3: Desc ip i e s a is ics and co ela ion ma ix
Mean
(S .De ia ion)
ln(RD)
ln(BVC)
ln(E)
ln(RD)
4.30
(2.40)
ln(BVC)
6.42
(1.76)
0.532*
ln(E)
4.73
(1.75)
0.341*
0.627*
TRADE
0.79
(0.43)
-0.204*
-0.202*
0.347*
No es:
- Employees a e exp essed in housands o pe sons engaged, mone a y a iables a e exp essed in millions o cons an
PPP 2010 US dolla s.
- * Signi ican a 95%
4 Resul s
As a as he econome ic me hodology is conce ned, he DPD speci ica ion equi es GMM-
amily es ima o s o gene a e unbiased es ima es. In pa icula , gi en he e y high AR(1)
co ela ion o ou dependen a iable (R&D, equal o 0.97), we op ed o a GMM-SYS as he
bes unbiased es ima o (see Blundell and Bond 1998; Pelleg ino e al., 2019; Damioli e al.,
2021).
In Tables 4 and 5 ou a en ion will ocus on GMM-SYS es ima ed coe icien s, whe e he
lagged R&D is ea ed as endogenous
16
, al hough POLS and FE es ima es a e also epo ed
17
as con ols.
16
A numbe o Hansen es s we e un o assess he po en ial endogenei y o o he eg esso s. Resul s p o ided
e idence o hei exogenei y. Indeed, he BVC a iable is a s ock al eady conside ing - by cons uc ion - BVs
lows in p e ious yea s.
17
POLS is a ec ed by upwa d bias in es ima ing he lagged dependen a iable, meanwhile a downwa d bias is
cha ac e izing he case o he FE es ima o . As i can be seen in Tables 4 and 5, he GMM-SYS es ima o o he
lagged dependen a iable is always wi hin hese uppe and lowe bounds, as equi ed.
19
In Table 4 (column 3) he dependen a iable is con i med o be s ongly pe sis en and au o-
co ela ed wi h a highly signi ican coe icien o abou 0.9. Ou a iable o in e es , BVC,
u ns ou o ha e a posi i e and e y signi ican impac on R&D, wi h an elas ici y equal o
4.4%. Employmen , as size con ol a iable, has an expec ed posi i e and signi ican e ec on
R&D, while TRADE does no seem o a ec in a signi ican way he inno a i e in es men s
a he sec o al le el
18
. Wi h ega d o he s anda d GMM diagnoses, he AR(1) and AR(2) es s
and he non-signi ican Hansen es eassu e us on he p ope choice o he ins umen s ma ix
(see Bond 2002).
Ou key esul - using he whole a ailable sample - suppo s ou hypo hesis ha sho - e m
mobili y (i.e. ideas ci cula ion and ace- o- ace in e ac ions), posi i ely a ec s inno a i e
in es men s. In addi ion, digging in o he sample composi ion and conside ing he wo g oups
o leade and non-leade R&D coun ies, we es he same ela ionship o he sole non-leade
inno a i e coun ies (column 6) o e alua e i his channel migh play a s onge ole in
coun ies ha a e u he away om he inno a ion on ie . He e he hypo hesis is ha weake
coun ies in e ms o domes ic knowledge gene a ion ( hose wi h lowe R&D/GDP a ios) may
bene i mo e om he knowledge ans e associa ed wi h BVs in compa ison wi h inno a i e
leade s . The esul s indica e ha his is he case, as he bene icial impac o BVC is p ima ily
due o i s e ec in less inno a i e coun ies (elas ici y inc easing o 4.7%). This implies ha
he ee exchange o ideas and people could be essen ial o hei inno a i e pe o mance.
18
This ou come may be due o he impe ec measu e we ha e a disposal, ha is he na ional igu e epea ed a
he indus y le el (see abo e).
20
Table 4: Dependen a iable: ln(RD)
(1)
WHOLE
SAMPLE
OLS
(2)
WHOLE
SAMPLE
FE
(3)
WHOLE
SAMPLE
GMM-
SYS
(4)
NON-LEADER
INNOVATORS
POLS
(5)
NON-LEADER
INNOVATORS
FE
(6)
NON-LEADER
INNOVATORS
GMM-SYS
Lagged
ln(RD)
0.979***
(0.002)
0.687***
(0.008)
0.904***
(0.029)
0.976***
(0.002)
0.680***
(0.009)
0.874***
(0.003)
ln(BVC)
0.009**
(0.004)
0.019*
(0.012)
0.044***
(0.015)
0.010***
(0.005)
0.015
(0.021)
0.047***
(0.017)
ln(E)
0.011**
(0.004)
0.043***
(0.014)
0.016***
(0.009)
0.015*
(0.005)
0.035**
(0.015)
0.027*
(0.014)
TRADE
0.045
(0.055)
0.029
(0.055)
0.002
(0.001)
0.041
(0.060)
0.004
(0.065)
0.037
(0.068)
Cons an
-0.034
(0.121)
1.221***
(0.136)
-0.280
(0.279)
0.093
(0.054)
1.216***
(0.147)
0.052
(0.172)
Time-
dummies
Yes
Yes
Yes
Yes
Yes
Yes
Coun y-
dummies
Yes
-
Yes
Yes
-
Yes
Time-
dummies
Wald es
(p- alue)
3.93***
9.30***
4.79***
3.39***
10.10***
5.24***
Coun y-
dummies
Wald es
(p alue)
2.42***
-
0.96
2.10***
-
1.13
Hausman
es
1426.95***
1152.94***
Adj. R2
0.97
0.96
R2 wi hin
0.60
0.61
AR(1)
AR(2)
Hansen
es
-7.84***
0.64
222.18
-7.13***
-0.49
205.25
Numbe
g oups
604
488
Numbe
obs.
8,316
6,447
No es:
- In columns (4), (5), (6), he op RD/GDP pe o me s in 2019 we e excluded (Ko ea, Sweden, Japan, US, Ge many)
- * Signi ican a 90%; ** Signi ican a 95%; *** Signi ican a 99%
21
As a complemen a y exe cise, we un he same es ima ion ocusing on he op 5 R&D in es o s
(Table 5). As ob ious (column 3), he pe sis ence o R&D is highe in leade inno a i e
coun ies (96%), implying ha in hese coun ies indus ies a e keene o in es in R&D in a
s able manne . Tu ning ou a en ion o ou key impac a iable, he BVC, al hough posi i e,
is no longe s a is ically signi ican . Ou in e p e a ion is ha BVs as a channel o knowledge
acquisi ion is no so impo an in hose coun ies ha can ely on an excellen , es ablished, and
con inuous p oduc ion o domes ic knowledge (while i u ns ou o be essen ial o all he o he
coun ies ha a e no wi hin he club o he R&D champions wo ldwide, see Table 4, column
6).
As he di e se indus ial s uc u e o economies could in luence he obse ed di e ences and
shape he ou comes, we classi y indus ies as ei he ‘High’ o ‘Low’ inno a i e (based on he
OECD classi ica ion - see Table 2) o disco e and quali y po en ial di e ences in ou esul s.
We p esen he esul s in Table 6.
The GMM-SYS es ima es e eal ha BVC has he highes and mos signi ican impac on R&D
in es men s in ‘High’ R&D indus ies in non-leade coun ies (column 2). This sugges s ha
ex e nal knowledge ans e can be c ucial in boos ing R&D in indus ies ha a e mo e inclined
owa ds inno a ion bu a e si ua ed in non-leade coun ies. In addi ion, ‘Low’ indus ies, in
bo h leade and non-leade inno a o coun ies (columns 3 and 4, espec i ely), bene i om
he mobili y o people, wi h an elas ici y o abou 3-4%. To summa ize, high- ech indus ies in
leade coun ies do no seem o need BVs as a sou ce o iable knowledge; BVs a e ins ead
c ucial in non-leade coun ies and low- ech indus ies. Howe e , while weake si ua ions
bene i mo e om BVs in gene al, he mos signi ican and la ge coe icien is de ec ed in he
high- ech indus ies in he non-leade coun ies, eminding us o he key ole o he abso p i e
capaci y (see p e ious sec ions).
22
Table 5: Dependen a iable: ln(RD) in Leade Inno a o s
(Ko ea, Sweden, Japan, US, Ge many)
(1)
LEADER
INNOVATORS
POLS
(2)
LEADER
INNOVATORS
FE
(3)
LEADER
INNOVATORS
GMM-SYS
Lagged
ln(RD)
0.987***
(0.004)
0.700***
(0.017)
0.955***
(0.018)
ln(BVC)
0.007
(0.007)
-0.012
(0.039)
0.023
(0.014)
ln(E)
-0.001
(0.010)
0.404***
(0.082)
0.001
(0.010)
TRADE
-0.081
(0.156)
0.033
(0.135)
-0.428
(0.458)
Cons an
0.112
(0.128)
-0.318
(0.489)
0.047
(0.094)
Time-dummies
Yes
Yes
Yes
Coun y-dummies
Yes
-
Yes
Time-dummies
Wald es
(p- alue)
1.32
3.96***
1.78**
Coun y-dummies
Wald es
(p- alue)
3.05**
-
2.18*
Hausman es
310.56***
Adj. R2
0.98
R2 wi hin
0.64
AR(1)
AR(2)
Hansen es
-3.34***
1.59
101.18
Numbe g oups
116
Numbe obs.
1,869
No es:
- To a oid o e -iden i ica ion (see Roodman, 2009), numbe o ins umen s is 115 in columns (3)
- * Signi ican a 90%; ** Signi ican a 95%; *** Signi ican a 99%
29
O ega-A gilés, R., Pi a, M. & Vi a elli, M. (2014). The ansa lan ic p oduc i i y gap: is R&D
he main culp i ? Canadian Jou nal o Economics, 47(4), 1342–1371.
Pa isi, M.L., Schian a elli, F. &. Sembenelli, A. (2006). P oduc i i y, inno a ion c ea ion and
abso p ion, and R&D. Mic oe idence o I aly, Eu opean Economic Re iew, 50(8), 2037–
61.
Pelleg ino, G., Pi a, M. & Vi a elli, M. (2019. Beyond R&D: he ole o embodied
echnological change in a ec ing employmen . Jou nal o E olu iona y Economics, 29,
1151-1171.
Piga, C. & Vi a elli, M. (2003). Sample selec ion in es ima ing he de e minan s o coope a i e
R&D. Applied Economics Le e s, 10(4), 243-246.
Piga, C. & Vi a elli , M. (2004). In e nal and ex e nal R&D: A sample selec ion app oach.
Ox o d Bulle in o Economics and S a is ics, 66(4), 457-482.
Pi a, M., Tani, M. & Vi a elli, M. (2018). Business isi s, knowledge di usion and
p oduc i i y. Jou nal o Popula ion Economics, 31(4), 1321-1338.
Pi a, M., Tani, M. & Vi a elli, M. (2023). The p oduc i i y impac o sho ‑ e m labo mobili y
ac oss indus ies. Small Business Economics, 60(2), 691-705.
Polanyi, M. (1966). The Taci Dimension. Pe e Smi h.
Roge s, M. (1995). In e na ional knowledge spill-o e s: A c oss-coun y s udy. In S. Dow ick
(Ed.), Economic app oaches o inno a ion. Edwa d Elga Publica ions.
Rome , P.M. (1994). The o igins o endogenous g ow h. Jou nal o Economic Pe spec i es,
8(1), 3-22.
Roodman, D. (2009) A no e on he heme o oo many ins umen s. Ox o d Bulle in o
Economics and S a is ics, 71(1), 135–158.
Rosenbe g, N. & S einmuelle , W.E. (1988). Why a e Ame icans such poo imi a o s?.
Ame ican Economic Re iew, 78(2), 229-234.
Sal , J., Single on, A., & Hoga h, J. (1994). Eu ope’s in e na ional mig an s: Da a sou ces,
pa e ns and ends. HMSO.
Scho , J., & S einmuelle , W. E. (2018). Th ee ames o inno a ion policy: R&D, sys ems o
inno a ion and ans o ma i e change. Resea ch Policy, 47(9), 1554-1567.
Shalley, C.E., Zhou, J. & Oldham, G.R. (2004). The e ec s o pe sonal and con ex ual
cha ac e is ics on c ea i i y: Whe e should we go om he e?. Jou nal o Managemen , 30(6),
933-958.
Tani, M. (2014). Business isi s and he ques o ex e nal knowledge. Resea ch in Labo
Economics, 40, 293–324.
Teece, D.J., Pisano, G.P. & Shuen, A. (1997). Dynamic capabili ies and s a egic managemen .
S a egic Managemen Jou nal, 18(7), 509-533.
To e, A. & Ralle , A. (2005). P oximi y and localiza ion. Regional S udies, 39(1), 47–59.
Uni ed Na ions - UN (1998). Recommenda ions on s a is ics o in e na ional mig a ion -Uni ed
Na ions, ST/ESA/STAT/SER.M/58/Re .1.
Ve gine , L. & Riccaboni, M. (2021). Talen goes o global ci ies: he wo ld ne wo k o
scien is s’ mobili y. Resea ch Policy, 50(1), 104127.
Veugele s , R. (1997). In e nal R&D expendi u es and ex e nal echnology sou cing. Resea ch
Policy, 26(3): 303-316.
on Hippel, E. (1987). Coope a ion be ween i als: In o mal know-how ading. Resea ch
Policy, 15(6), 285–305.
30
on Hippel, E. (1994). “S icky in o ma ion” and he locus o p oblem sol ing: Implica ions o
inno a ion. Managemen Science, 40(4), 429-439.
Zellne , C. (2003). The economic e ec s o basic esea ch: e idence o embodied knowledge
ans e ia scien is s’ mig a ion. Resea ch Policy, 32(10),1881–1895.
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
2025-04 Digi al Technologies, Hi ing, T aining, and Fi m Ou comes: E idence om AI and ICT
Adop ion in Indian i mss by Sneha Ma ydas, Nandi ha Ma hew, Gio dano De
Ma zo and Ca lo Pie obelli
2025-05 New echnologies and employmen : he s a e o he a by Ma co Vi a elli and
Guille mo A enas Díaz
2025-06 Fo mal-In o mal Supply Chain Linkages and Fi m P oduc i i y in Sub-Saha an A ica:
The Role o Human Capi al by Gbenoukpo Robe Djidonou, Neil Fos e -McG ego
and Nandi ha Ma hew
2025-07 In e ne Access and You h’s Men al Heal h and Well-being: E idence om E hiopia
by T ang Thi Pham and Pui-Hang Wong
2025-08 A Twin T ansi ion o a policy lagship? Eme gen cons ella ions and dominan blocks
in g een and digi al echnologies by Linnea Nelli, Ma ia En ica Vi gilli o and Ma co
Vi a elli
2025-09 Good o business, no so much o he en i onmen ? En y in o impo ing and he
ene gy in ensi y o Indian plan s by Michele Dele a, Nandi ha Ma hew and Tania
T eibich
2025-10 The ole o business isi s in os e ing R&D in es men by Ma co Vi a elli,
Ma iac is ina Pi a and Massimiliano Tani
