Calda ola, Be na do; Mazzilli, Da io; Pa elli, Au elio; Sba della, Angelica
Wo king Pape
S uc u al Change, Employmen , and Inequali y in Eu ope:
An Economic Complexi y App oach
UNU-MERIT Wo king Pape s, No. 2024-033
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: Calda ola, Be na do; Mazzilli, Da io; Pa elli, Au elio; Sba della, Angelica (2024) :
S uc u al Change, Employmen , and Inequali y in Eu ope: An Economic Complexi y App oach,
UNU-MERIT Wo king Pape s, No. 2024-033, 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
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/326929
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/
#2024-033
S uc u al Change, Employmen , and Inequali y in Eu ope:
an Economic Complexi y App oach
Be na do Calda ola, Da io Mazzilli, Au elio Pa elli and Angelica Sba della
Published 25 No embe 2024
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.
S uc u al Change, Employmen , and Inequali y in
Eu ope: an Economic Complexi y App oach
Be na do Calda ola∗1,2,3, Da io Mazzilli1, Au elio Pa elli1, and Angelica
Sba della1
1En ico Fe mi Resea ch Cen e , Rome, I aly
2UNU-MERIT, Maas ich , he Ne he lands
3School o Business and Economics, Maas ich Uni e si y, Maas ich , he Ne he lands
No embe 25, 2024
Abs ac
S uc u al change consis s o indus ial di e si ica ion owa ds mo e p oduc i e,
knowledge-in ensi e ac i i ies. Howe e , changes in he p oduc i e s uc u e bea in-
he en links wi h job c ea ion and income dis ibu ion. In his pape , by aking an
economic complexi y app oach, we in es iga e he consequences o s uc u al change
de ined in e ms o labou shi s owa ds mo e complex indus ies on employmen
g ow h, wage inequali y and unc ional dis ibu ion o income. The analysis is con-
duc ed o Eu opean coun ies using da a on disagg ega ed indus ial employmen
sha es o e he pe iod 2010 – 2018. Fi s , we iden i y pa e ns o indus ial specialisa-
ion by alida ing a coun y-indus y indus ial employmen ma ix using a bipa i e
weigh ed con igu a ion model (BiWCM). Secondly, we in oduce a coun y-le el mea-
su e o labou -weigh ed Economic Fi ness, which can be decomposed in such a way as
o isola e a componen ha iden i ies he mo emen o labou owa ds mo e complex
indus ies he s uc u al change componen . Thi dly, we link s uc u al change o i)
employmen g ow h, ii) wage inequali y, and iii) he labou sha e o he economy. Ou
indings indica e ha he s uc u al change measu e we p opose is associa ed nega i ely
wi h employmen g ow h. Howe e , i is also associa ed wi h lowe income inequali y:
as coun ies mo e o mo e complex indus ies, hey d op he leas complex ones, so
he (low-paid) jobs in he leas complex sec o s disappea . Finally, s uc u al change
p edic s a highe labou a io o he economy; howe e , his is likely o be due o he
inc ease in wages a he han o job c ea ion.
∗Co esponding au ho . Email: [email p o ec ed]
1
Keywo ds: S uc u al Change, Economic Complexi y, Inequali y, Employmen
JEL Codes: D63, E24, J31, O11, O52
2
1 In oduc ion
S uc u al change can be de ined by he ealloca ion o economic ac i i y ac oss b oad sec o s
o he economy (Lewis 1954,Sy quin 1988). The shi o he ela i e impo ance o indus ial
employmen o ou pu owa ds inc easingly mo e p oduc i e ac i i ies ha de ines he p ocess
o s uc u al change is also an engine o long- e m, sus ained economic g ow h (He endo
e al. 2014,2022). Howe e , changes in he p oduc i e s uc u e a e also inhe en ly associa ed
wi h he dis ibu ion o income, as highligh ed in he seminal wo k by Kuzne s (1955). Fo
ins ance, when coun ies upg ade hei p oduc i e s uc u e, hey also mo e o sec o s wi h
highe knowledge in ensi y, and lowe labou equi emen s. This may ha e consequences on
he wage and unc ional dis ibu ion o income, as well on he c ea ion o new jobs and he
des uc ion o old ones.
In his pape , we use he case o Eu ope o s udy he ela ionship be ween he mo emen
o labou o inc easingly knowledge in ensi e ac i i ies – de ined he e as s uc u al change
– wi h employmen g ow h, wage inequali ies, and he unc ional dis ibu ion o income
be ween labou and capi al.
As a ma e o ac , in he las decades wi hin coun y inequali y has been g owing in
Eu ope and in high-income coun ies. A he same ime, hese coun ies ha e de-indus ialised
du ing he las wa e o ICT-led globalisa ion, leading o he subs i u ion o manu ac u ing
wi h jobs in complex, knowledge in ensi e indus ies – such as business se ices – con ibu ing
o inc easing inequali ies (An onelli & Tubiana 2020). The mechanism unde lying his end
has been explo ed u he in he empi ical li e a u e. The Skill-Biased Technological Change
(SBTC) li e a u e (Au o e al. 2003,Au o 2022,Acemoglu & Res epo 2022) has examined
he impac o he main d i e o s uc u al change – echnological change – on employmen
and on he dis ibu ion o ea nings, sugges ing ha wa es o echnological change end o
complemen only high-skilled wo ke s in knowledge in ensi e indus ies, lea ing low-skilled
wo ke s behind.
Mo eo e , s uc u al change bea s a s ong linkage wi h employmen c ea ion. As economies
ans o m in hei p oduc i e s uc u es, hey c ea e winne s and lose s (Schumpe e 1934).
The cu en wa e o echnological change may no educe o al employmen (Dau h e al.
2021), bu may induce labou ma ke adjus men s a he disad an age o low-skilled wo ke s,
whose sha e o e o al wo ke s is educed (G ae z & Michaels 2018). In he case o o he
echnologies, such as ICTs, hese may ha e been he culp i s o jobless g ow h (B ynjol sson
& McA ee 2011,F ey & Osbo ne 2017) and con ibu ed o he sh inking o he numbe o
middle-skill jobs (Goos e al. 2009,2014).
In o de o iden i y he deg ee o knowledge in ensi y o Eu opean indus ies we employ
1
an Economic Complexi y (EC) pe spec i e. Complexi y me ics aim a quan i ying he ca-
pabili y s uc u e o coun ies by looking a hei p oduc i e baske , p ese ing in o ma ion
on hei specialisa ion and di e si ica ion pa e ns a he han by simply agg ega ing hei
economic ou pu (Hausmann & Hidalgo 2009,Tacchella e al. 2012). By he same oken, EC
measu es allow o iden i y he capabili y (knowledge) in ensi y (Lo Tu co & Maggioni 2022)
o indi idual p oduc s, indus ies, o echnologies by conside ing hei ubiqui y – a p oxy
o he di usion o he capabili ies ha such ac i i ies equi e – weigh ed by he capabili y
endowmen o he coun ies specialised in such ac i i ies – i.e., he coun ies’ Fi ness. Gi en
he s ong link o EC me ics wi h economic di e si ica ion, and in pa icula o he Economic
Fi ness and Complexi y (EFC) app oach (Tacchella e al. 2012), hey ha e p o ed o be a
use ul ool o desc ibe a long- e m dynamic such a s uc u al change (F ei e 2021,Cas añeda
e al. 2022,McNe ney e al. 2023), being also highly p edic i e o GDP g ow h (Tacchella
e al. 2018). Following he analogy be ween s uc u al change and complexi y, he la e can
be used o explain he inequali y ends ha esul o changes in he sec o al composi ion o
he economy. The li e a u e on complexi y and inequali ies (Sba della e al. 2017,Ha mann
e al. 2017,Ha mann & Pinhei o 2022) indica es ha highe complexi y o he p oduc i e
s uc u e is associa ed o lowe inequali ies a he coun y le el. I is also well known ha
knowledge concen a es in space (Balland e al. 2020,2022), u ning geog aphy in o a use ul
lens h ough which he income dis ibu ion can be analysed. In ac , a he egional le el
he sign o his ela ionship lips (Sba della e al. 2017), as only egions wi h he s onges
knowledge basis a e able o di e si y in o highly complex indus ies, lea ing lagga ds behind
(Pinhei o e al. 2022). Looking a he ela ionship be ween complexi y and employmen ,
Adam e al. (2023) inds ha highe complexi y is linked o job c ea ion. Whe e exis ing
jobs a e made edundan , i has been shown ha e-employmen is easie in p esence o
ela ed indus ial a ie y a he egional le el (Hane-Weijman e al. 2018).
One impo an ea u e o EC me hods is ha hese allow o quan i y he capabili y
equi emen s o inely disagg ega ed indus ies, o e ing a g anula desc ip ion o he p ocess
o s uc u al change bo h ac oss and wi hin b oad sec o s o he economy. In o de o o e
an encompassing assessmen o he complexi y o Eu opean indus ies and o he labou
lows ac oss hem, we depa by adi ional app oach o he EC li e a u e – which in e s he
p oduc i e s uc u e o coun ies using expo da a – by using da a on indus ial employmen
in Eu opean coun ies be ween 2010 and 2019. Fi s , we conside he use o employmen da a
o measu e indus ial Complexi y and coun ies’ employmen Fi ness as his allows o include
also non- adeable ac i i ies, mos o which a e in he se ice sec o . This is pa icula ly
ele an conside ing ha Eu opean economies ha e hea ily shi ed hei p oduc i e s uc u e
owa ds he e ia y indus ies along wi h he p ocess o globalisa ion. Secondly, he ocus
2
on employmen ep esen s a c ucial elemen in he concep ualisa ion o s uc u al change as
he change in ela i e sha es o he economy. Using employmen da a o measu e Economic
Fi ness and Complexi y espec i ely a he coun y and indus y le el esona es wi h his
concep ual amewo k, and allows o ou line he s uc u e o capabili ies o labou as a
p oduc i e inpu .
In he p esen s udy, he analysis o he ela ionship be ween s uc u al change, inequali y,
and employmen c ea ion is conduc ed o Eu opean coun ies o e he pe iod 2010 - 2018.
The empi ical app oach elies on in o ma ion on Eu opean indus ial employmen p o ided
by Eu os a s S uc u al Business Su ey da a, measu ed a a ine le el o disagg ega ion
(NACE 4 digi s) ac oss manu ac u ing and se ice indus ies. This in o ma ion is ma ched
wi h da a on he dis ibu ion o wages a he coun y le el, p o ided by he ILO, and wi h
da a on he labou sha e o he economy and employmen g ow h, ob ained by ARDECO.
Using his newly cons uc ed da ase , we s a by iden i ying pa e ns o indus ial special-
isa ion by alida ing he coun y-indus y indus ial employmen ma ix using a bipa i e
weigh ed con igu a ion model (BiWCM) (?) , o e coming some limi a ions imposed by he
Balassa (1965) me hod. Secondly, we in oduce a measu e o labou -weigh ed Fi ness, which
sums up he complexi y o indus ies weigh ed by hei employmen sha e. This measu e can
be decomposed in such a way as o iden i y he con ibu ion o changes in labou -weigh ed
Fi ness coming om he mo emen o labou owa ds mo e complex indus ies. We iden i y
such componen as he one linked o s uc u al change om a labou iewpoin he s uc-
u al change componen . Thi dly, we link he s uc u al change componen o a numbe o
economic ou comes a he coun y le el: i) employmen g ow h, ii) wage inequali y, and iii)
he unc ional dis ibu ion o income (labou sha e o he economy). In his way, we analyse
he dis ibu ional consequences o s uc u al ans o ma ion – amed as he mo emen o
labou o mo e complex indus ies – and i s job-c ea ing po en ial.
An analysis o he specialisa ion pa e ns in Eu opean coun ies, conduc ed by inspec ing
he signi ican links o he coun y-indus y specialisa ion ma ix gene a ed by he BiWCM,
e eals ha he mos di e si ied coun ies ha e p og essi ely mo ed away om specialisa ion
in low-complexi y, labou -in ensi e indus ies. Mo eo e , ou empi ical analysis explo es
he associa ion be ween he s uc u al change componen o ou labou -Fi ness measu e -
de i ed om he mo emen o labou owa ds mo e complex indus ies - o employmen
g ow h, wage inequali y, and he labou sha e o he economy). The p elimina y esul s o
OLS panel eg essions wi h coun y and ime ixed e ec s indica e ha ou s uc u al change
measu e is associa ed nega i ely wi h employmen g ow h, co obo a ing he e idence ha
highly complex indus ies ha e lowe labou equi emen s. Howe e , i is also associa ed wi h
lowe income inequali y measu ed in e ms o he a io o a e age wages in he nin h and
3
i s deciles o he wage dis ibu ion. As coun ies mo e o mo e complex indus ies, hey
d op he leas complex ones, so he (low-paid) jobs in he leas complex sec o s disappea ,
making he i s decile o he sala y dis ibu ion go up. Finally, s uc u al change p edic s a
highe labou a io o he economy; howe e , his is likely o be due o he inc ease in sala ies
in highly complex indus ies a he han by job c ea ion. Ve y in e es ingly, he s uc u al
change componen is he only one o be signi ican in he eg essions, while he inc ease in
wi hin-sec o complexi y, and he labou -weigh ed Fi ness o coun ies do no explain ou
ou come a iables. Resul s a e obus o a ba e y o obus ness checks, including lea e-
one-ou eg essions and a speci ica ion whe e he s uc u al change componen o he labou -
weigh ed Fi ness is eplaced by a simila componen ob ained decomposing a labou -weigh ed
en opy measu e.
The emainde o his pape is o ganised as ollows. In Sec ion 2we del e in o he
Economic Fi ness and Complexi y amewo k, and we in oduce a new measu e – he Labou -
Weigh ed Fi ness (LWF) he a ia ion o which can be decomposed o ex ac a s uc u al
change componen . In Sec ion 3we ou lined he da a sou ces used in he analysis, which in
u n is desc ibed in he ollowing Sec ion (4). Sec ion 5illus a es he p elimina y esul s o
he analysis, and Sec ion 6concludes.
2 Analy ical F amewo k
In his sec ion, we will ou line he analy ical amewo k used o de i e a measu e o s uc u al
change based on he ealloca ion o labou o complex indus ies. Fi s , we will de ine coun-
ies’ Fi ness and indus ial complexi y acco ding o he EFC amewo k (Tacchella e al.
2012). Second, we in oduce a new measu e – he Labou Weigh ed Fi ness – om which we
de i e a s uc u al change componen .
2.1 S uc u al change, Fi ness and Complexi y
Economic Complexi y is a amewo k ha builds on ea lie e olu iona y and ins i u ional
li e a u e (Cimoli & Dosi 1995,Hi schman 1958,Teece e al. 1994) o ackle he complexi y
o economic sys ems. I desc ibes he economy as an e olu iona y p ocess o globally in-
e connec ed ecosys ems. The main ecen ad ance wi h espec o he ea lie li e a u e is
he use o newly de eloped ne wo k science and o he me hods o in es iga e complex and
dynamical sys ems (Hausmann & Klinge 2006,Hausmann & Hidalgo 2009,Tacchella e al.
2012) o sepa a e he andom noise om he unde lying signal. The Economic Complexi y
amewo k shi s he ocus om agg ega e quan i ies (Wha is he GDP o he coun y?) o a
4
Whe e be weenc, ,wi hinc, and ∆LWFc, a e espec i ely he be ween e m, wi hin e m,
and sum o he p e ious wo, as desc ibed by equa ion 6; in all cases, hese quan i ies a e
compu ed o k= 1.Yc, is a placeholde o measu es o employmen g ow h, wage inequali y
and labou sha e o he economy. Employmen g ow h is compu ed as he sha e o employed,
wo king age popula ion in each coun y a ime o e he same sha e o employmen in he
p e ious ime pe iod, as ob ained om he ARDECO da ase :
gc, =empc,
empc, −1
(9)
Wage inequali y is cons uc ed using he ILO S a is ics on Labou Income and Inequali y
desc ibed in Sec ion 3.3. In o de o cap u e he e ec o s uc u al change on he dis ibu ion
o income, we ake h ee di e en measu es o wage inequali y: 9 h o 1s pe cen ile, 9 h o
5 h pe cen ile, and 5 h o 1s pe cen ile:
ineqp h/q h
c, =wagep h
c,
wageq h
c,
(10)
Wi h he supe sc ip s pand qindica ing ei he he 9 h, 5 h o 1s pe cen ile. Finally, we
use he ARDECO da a o compu e a measu e o he labou sha e o he economy in each
coun y and yea , aking he a io o he o al wages paid o wo ke s o e he o al alue
added, all measu ed a cons an p ices (2015):
labsha ec, =wagec,
V Ac,
(11)
Finally, Xc, is a ec o o ime a ying con ols a he coun y le el, including GDP pe
capi a, popula ion ( om ARDECO), sha e o R&D in es men s o e GDP, sha e o expo s
o e GDP ( om he Wo ld De elopmen Indica o s).
All equa ions a e es ima ed using OLS coun y (σc) and ime (τ ) ixed e ec s, wi h
e o s clus e ed by coun y and yea . P elimina y esul s o he desc ip i e econome ic
model a e p o ided in he nex sec ion, along wi h an illus a ion o he esul s p oduced by
he cons uc ion o he specialisa ion ma ix and s uc u al decomposi ions.
5 Resul s
5.1 Desc ip i e esul s
We begin by commen ing on he pa e ns exhibi ed by he specialisa ion ma ix, which has
been ob ained by il e ing he coun y-indus y employmen ma ix o he yea 2010 as
11
desc ibed in Sec ion 4.1. Figu e 1con ains in o ma ion on he s a is ical signi icance o
indi idual coun y-indus y linkages, wi h alues going om mo e signi ican ( ed) o less
signi ican (blue). The ows o he ma ix ep esen coun ies, while he columns iden i y
indus ies. Coun ies ha e been o de ed om highe Fi ness ( op) o lowe Fi ness (bo om),
and indus ies a e so ed om he leas (le ) o he mos ( igh ) complex.
0 25 50 75 100 125 150 175 200
0
5
10
15
20
25
0.0
0.2
0.4
0.6
0.8
1.0
Figu e 1: BIWCM Specialisa ion Ma ix
A a i s isual inspec ion, he ma ix displays a pa e n asc ibable o he nes ed pa e n
(Bus os e al. 2012) ha ypically cha ac e ises coun y-p oduc RCA ma ices, il e ed using
he Balassa index (Pa elli e al. 2023). P ecisely, he dis ibu ion o he s a is ically signi ican
linkages be ween coun ies and indus ies ha desc ibes coun ies’ di e si ica ion pa e ns
has a iangula shape, which can be used as a b oad app oxima ion o he nes ed na u e o
he sys em being obse ed. While he concep o nes edness has o igina ed in e olu iona y
biology, i has ound as applica ion in he economic complexi y li e a u e: p oduc i e
sys ems ha exhibi a nes ed pa e n e eal he p esence o co ela ions wi hin he sys em
ha canno be accoun ed o using linea models o a e age app oxima ions. In economic
e ms, nes edness indica es ha indus ial ac i i ies ha a e pu sued by low-Fi ness coun ies
a e also pu sued by high-Fi ness coun ies, which a e howe e also mo e di e si ied.
In he speci ic case desc ibed in igu e 1, he economic in e p e a ion o nes edness holds
only in pa ; as a ma e o ac , he op le co ne – whe e high-Fi ness coun ies and low-
Complexi y indus ies a e loca ed – appea s o be less popula ed ha one would expec ed in
a ully nes ed ma ix. While a simila pa e n eme ges also in o he sys ems – such as he
12
one o scien i ic p oduc ion (Pa elli e al. 2023) – in his case we ind he obse ed esul con-
sis en wi h he p ocess o s uc u al change. When i comes o he employmen dimension
o he indus ial p oduc ion, high-Fi ness coun ies end o ha e highe di e si ica ion and o
de elop a compa a i e ad an age in capi al- o knowledge-in ensi e indus ies, a he expen-
si e o mo e labou -in ensi e (and low-complexi y) ones. As a esul , indus ialising coun ies
wi h a compa a i e ad an age in labou in ensi e ac i i ies (o en due o lowe wages) end
o specialise in he la e indus ies, and o ha e lowe di e si ica ion. In subs ance, he
sample o Eu opean coun ies seems o be di ided in wo subsamples o coun ies a di e en
s ages o s uc u al ans o ma ion, wi h a mino i y o mo e ma u e coun ies and a majo i y
o indus ialising coun ies. The dynamic o Fi ness ankings shows clea ly which coun ies
belong o each g oup, wi h he mo e ma u e economies being Ge many, F ance, I aly, Uni ed
Kingdom and Spain, ollowed by eme ging coun ies such as Poland and Czechia.
AT
BE
BG
CH
CY
CZ
DE
DK
EE
EL
ES
FI
FR
HR
HU
IE
IT
LT
LU
LV
MT
NL
NO
PL
PT
RO
SE
SI
SK
UK
AT
BE
BG
CH
CY
CZ
DE
DK
EE
EL
ES
FI
FR
HR
HU
IE
IT
LT
LU
LV
MT
NL
NO
PL
PT
RO
SE
SI
SK
UK
Fi ness wi h BiWCM: anking dynamics
Figu e 2: Fi ness ankings: 2010-2018
We now mo e owa ds an analysis o he a ia ion in LWF o e he ime pe iod 2010–2018.
13
Figu e 3shows he di e ence in LWF be ween 2010 and 2018, compu ed as pe equa ion 6,
wi h i s espec i e be ween and wi hin componen . We shall ecall ha he o me measu es
he con ibu ion o s uc u al change – de ined he e as he mo emen o labou o mo e
complex indus ies – o he a ia ion in LWF o e he ime pe iod unde conside a ion; while
he la e picks up he change in LWF due o inc easing complexi y o he indus ies in which
coun ies de elop an (in e ed) compa a i e ad an age. O e all, igu e 3indica es ha he
be ween and wi hin componen a e o en an ico ela ed. This can be ead in economic e ms
by conside ing he easonable assump ion ha as indus ies become mo e complex, hey also
become less labou in ensi e. The wi hin componen cap u es i he indus ies in which a
coun y is specialised become mo e complex – o ins ance, as a esul o echnical change
wi hin he indus y. By doing so, labou is shed by such complex indus ies, lowing in o
indus ies wi h lowe complexi y – a pa e n associa ed wi h a be ween componen o he
opposi e sign o he wi hin componen .
To ha e a clea e idea o wha happened in Eu opean coun ies o e he 2010–2018, igu e
4desc ibes he same quan i ies as igu e 3, bu o wo di e en ime pe iods: 2010–2014
(le panel) and 2014–2018 ( igh panel). The igu e indica es ha he i s ime pe iod he
a ia ion in LWF i ness was nega i e o mos coun ies, mos ly d i en by he dec easing
wi hin-indus y complexi y; i mus be no ed ha his pe iod coincides wi h he a e ma hs
o he 2008 inancial c ises, which has se e ely comp omised economic ac i i y in Eu ope
ac oss he boa d. In he ollowing pe iod, howe e , Eu opean coun ies show a much be e
pe o mance, wi h mos coun ies inc easing hei LWF, in many case due o wi hin indus y
complexi y and be ween indus y ealloca ion o labou . One in e es ing case is he one
o Poland, which is an ou lie in bo h pe iod: he coun y shows a huge wi hin indus y
g ow h o complexi y in 2010–2014, which is howe e ollowed by a nega i e be ween indus y
ealloca ion o labou in he ollowing ime pe iod – hypo he ically because o he lowe labou
in ensi y o i s p oduc i e s uc u e ha has shi ed owa ds inc easingly knowledge-in ensi e
ac i i ies.
14
UK
BE
SE
EE
LU
FI
CH
NL
CZ
DE
IT
LV
HU
AT
SK
FR
IE
DK
LT
HR
MT
NO
CY
ES
EL
RO
BG
SI
PL
PT
−2e−04 0e+00 2e−04 4e−04
Labou −weigh ed Fi ness decomposi ion
Coun y
Decomposi ion Wi hin e m Be ween e m Labou −weigh ed Fi ness
Figu e 3: Labou -weigh ed Fi ness decomposi ion: 2010-2018
15
CZ
UK
DE
BE
FR
FI
EE
HU
CH
NL
SE
AT
IT
SK
SI
LV
NO
RO
LT
IE
HR
DK
LU
BG
EL
ES
PT
MT
CY
PL
−2e−04 0e+00 2e−04 4e−04
Labou −weigh ed Fi ness decomposi ion
Coun y
PL
LU
UK
CY
MT
SE
BE
EE
DK
IE
HR
LV
NL
CH
LT
ES
IT
FI
EL
NO
SK
BG
AT
HU
RO
DE
PT
CZ
FR
SI
−2e−04 −1e−04 0e+00 1e−04 2e−04
Labou −weigh ed Fi ness decomposi ion
Coun y
Decomposi ion Wi hin e m Be ween e m Labou −weigh ed Fi ness
Figu e 4: Labou -weigh ed Fi ness decomposi ion: 2010-2014 and 2014-2018
16
5.2 S uc u al change and employmen g ow h, wage inequali y,
and unc ional dis ibu ion o income
In his sec ion we epo he esul s o he es ima ion o equa ion 8using he di e en ou come
a iables o employmen g ow h, wage inequali y and unc ional dis ibu ion o income. I
should bo ne in mind ha , a he p esen s age, he esul s p esen ed below aim exclusi ely
a desc ibe he co ela ion be ween s uc u al change and he ou come a iables, wi hou
any implica ion o hei causal ela ionships.
Table 1summa ises he ela ionship be ween s uc u al change (be ween) and employ-
men g ow h. Looking a columns (1) and (4), he be ween componen o he LWF s uc u al
decomposi ion shows a nega i e co ela ion wi h employmen g ow h a he coun y le el.
This indica es ha when coun ies expe ience a labou low om low- o high-complexi y
indus ies, hey also expe ience slowe a es o employmen g ow h. This inding seems o
co obo a e he hypo hesis ha high-complexi y indus ies con ibu e less o agg ega e em-
ploymen , due o hei highe le els o knowledge in ensi y, and lowe labou -in ensi y, as
hey equi e ewe (and likely mo e skilled) wo ke s. I is wo h men ioning ha nei he
he wi hin componen o he a ia ion in LWF (column 2), no he sum o bo h componen s
(column 3) do no show any signi ican co ela ion wi h employmen g ow h, highligh ing he
impo ance o s uc u al change in explaining ou comes ela ed o he c ea ion o employ-
men . The coe icien on he be ween componen emains signi ican also when he wi hin
componen is also included in he eg ession.
We now mo e o he ela ionship be ween s uc u al change and wage inequali y. Fi s ,
we look a he link be ween s uc u al change and he wage gap be ween op (9 h decile)
and bo om (1s decile) wages. The wage gaps appea s o con ac as a esul o labou
ealloca ion owa ds knowledge-in ensi e indus ies, as indica ed by columns (1) and (4) in
Table 2. Howe e , i is no clea whe he his esul is d i en by lowe wages in he op
o he dis ibu ion, o highe wages a he bo om. In o de o u he explo e his, Tables
3and 4show espec i ely he esul s o he wage gap be ween 5 h/1s and 9 h/5 h deciles.
In e es ingly, he wage gap be ween median and bo om sala ies na ows (columns 1 and 4
in Table 3) as a esul o s uc u al change, while he a io be ween sala ies a he 9 h and
1s decile doesn’ show any associa ion wi h s uc u al change (Table 4, columns 1 and 4).
We conclude ha he educ ion in he wage gap be ween op and bo om wages esul s om
bo om wages g owing a a as e a e han op wages. In o de o a ionalise his esul , i
is wo h going back o he specialisa ion ma ix desc ibed in igu e 1. As coun ies mo e o
mo e complex indus ies, hey d op he leas complex ones, as indica ed by he "emp ie " op-
le co ne o he ma ix. Assuming a posi i e co ela ion be ween complexi y and a e age
17
wages, i can be in e ed ha he educ ion in wage inequali y is d i en by he ac h a
he low-paid jobs in he leas complex sec o s disappea , making he 1s decile o he sala y
dis ibu ion go up. Al hough his conjec u e is ye o be es ed a he p esen s age o he
esea ch, i ep esen s one o he nex s eps in he analysis conduc ed in his pape .
Finally, Table 5displays he esul s o eg essing labou sha e ( he a io be ween agg ega e
wages and alue added in he economy) on he be ween componen o LWF. Columns (1)
and (4) show a posi i e ela ionship be ween he wo, indica ing ha as wo ke s mo e o
mo e complex indus ies, ha con ibu es posi i ely o he inc ease o he labou componen
o p oduc ion, which appea s o become mo e in ensi e. In o de o unde s and his esul s,
we should ely once mo e on he assump ion ha high-complexi y indus ies pay, on a e age,
highe wages. Bea ing in mind he di e si ica ion pa e ns displayed by igu e 1, coupled wi h
he esul s on employmen g ow h in Table 1. I employmen g ow h slows down as a esul
o s uc u al change owa ds knowledge-in ensi e indus ies, he inc ease in he labou sha e
o he economy is due o be d i en by highe a e age wages a he han highe pa icipa ion
o he popula ion o he wo k o ce. This hypo hesis will be es ed in he u u e s eps ha
we will ake o ca y on wi h he analysis.
As a inal no e on he p elimina y esul s p esen ed in his sec ion, i is impo an o
s ess ha he s uc u al change componen is he only one o be signi ican ac oss he
eg ession speci ica ions summa ised by equa ion 8. As a ma e o ac , nei he he wi hin
componen no he change in labou -weigh ed i ness explain any o he ou comes obse ed,
indica ing ha aming s uc u al change as he mo emen o labou ac oss indus ies o
di e en complexi y – ha is, knowledge in ensi y – can con ibu e signi ican ly o explain
he cu en ends in employmen g ow h, wage inequali y and unc ional dis ibu ion o
income in Eu opean coun ies.
18
Table 1: Employmen g ow h eg essions
Dependen Va iable: Yea ly employmen a e g ow h (%)
FE OLS
Model: (1) (2) (3) (4)
Va iables
Be ween -7,313.359∗∗ -7,527.217∗∗
(3,229.347) (3,248.861)
Wi hin 1,266.110 1,412.623
(1,121.540) (1,123.097)
∆C279.081
(993.907)
Popula ion (log) -19.942 -19.817 -19.927 -19.785
(12.878) (13.098) (13.057) (12.968)
GDPpc (log) -6.179∗-5.824 -6.046 -5.859
(3.563) (3.572) (3.556) (3.570)
R&D (%GDP) 0.169∗∗ 0.175∗∗∗ 0.174∗∗ 0.170∗∗
(0.064) (0.063) (0.063) (0.063)
Expo s (%GDP) 0.142 0.232 0.177 0.222
(0.761) (0.736) (0.756) (0.718)
Fixed-e ec s
Coun y Yes Yes Yes Yes
Yea Yes Yes Yes Yes
Fi s a is ics
Obse a ions 224 224 224 224
R20.526 0.522 0.520 0.528
Wi hin R20.233 0.225 0.223 0.236
Clus e ed (Coun y-Yea ) s anda d-e o s in pa en heses. Signi . Codes: ***: 0.01, **: 0.05, *: 0.1
19
Table 2: Wage inequali y eg essions: a io 9 h/1s dec.
Dependen Va iable: Ra io 9 h/1s dec.
FE OLS
Model: (1) (2) (3) (4)
Va iables
Be ween -8,201.860∗∗∗ -8,142.939∗∗∗
(2,884.048) (2,893.616)
Wi hin -547.693 -389.195
(671.490) (744.401)
∆C-1,372.345
(956.596)
GDPpc (log) -3.781 -3.831 -4.031 -3.870
(2.770) (2.863) (2.861) (2.778)
Popula ion (log) -5.596 -5.675 -5.763 -5.640
(7.401) (7.552) (7.514) (7.424)
R&D (%GDP) -0.013 -0.008 -0.010 -0.014
(0.024) (0.025) (0.024) (0.024)
Expo s (%GDP) -1.080 -1.091 -1.141 -1.102
(0.995) (1.000) (1.000) (1.002)
Fixed-e ec s
Coun y Yes Yes Yes Yes
Yea Yes Yes Yes Yes
Fi s a is ics
Obse a ions 224 224 224 224
R20.906 0.903 0.904 0.906
Wi hin R20.073 0.044 0.051 0.074
Clus e ed (Coun y-Yea ) s anda d-e o s in pa en heses. Signi . Codes: ***: 0.01, **: 0.05, *: 0.1
20
esul s, we a e planning o de ise an iden i ica ion s a egy based on a shi -sha e ins umen ,
which appea s a he momen he mos sui able iden i ica ion s a egy gi en he na u e o
ou (po en ially) endogenous explana o y a iable o in e es .
Ou esul s bea impo an consequences o policy making, as hey highligh a ade-o
be ween indus ial upg ading and labou c ea ion, while also dissec ing he na u e o he
educ ion o wage and unc ional inequali y esul ing om he mo emen o labou owa ds
complex indus ies. Based on hese esul s, policy-make s will ha e o de ice policy ools
ha s ee s uc u al change owa ds hose indus ies ha , a he same ime, con la e job
c ea ion wi h highe echnological sophis ica ion, in o de o sus ain a p ocess o inclusi e
economic g ow h. Such indus ies can be iden i ied using he EFC amewo k wi h a ocus
on labou , a he han expo s alone. Coupling he geog aphy o p oduc i e capabili ies wi h
labou and inequali y, enables us o ad ance wi h espec o he ex an li e a u e no only
in p o iding a policy- ele an and da a-d i en oolki o emb ace socio-economic complexi y,
bu also in o e ing a no el analy ical and empi ical unde s anding o a pa hway no only
gea ed owa ds boos ing g ow h, bu also o s ee ing inclusi e and sus ainable de elopmen .
27
Re e ences
Acemoglu, D. & Res epo, P. (2022), ‘Tasks, Au oma ion, and he Rise in U.S. Wage In-
equali y’, Econome ica 90(5), 1973–2016.
URL: h ps://www.econome icsocie y.o g/doi/10.3982/ECTA19815
Adam, A., Ga as, A., Ka sai i, M.-S. & Lapa inas, A. (2023), ‘Economic complexi y and
jobs: an empi ical analysis’, Economics o Inno a ion and New Technology 32(1), 25–52.
URL: h ps://doi.o g/10.1080/10438599.2020.1859751
An onelli, C. & Tubiana, M. (2020), ‘Income inequali y in he knowledge economy’, S uc u al
Change and Economic Dynamics 55, 153–164.
URL: h ps://www.sciencedi ec .com/science/a icle/pii/S0954349X2030374X
Au o , D. (2022), The labo ma ke impac s o echnological change: F om unb idled en hu-
siasm o quali ied op imism o as unce ain y, Wo king Pape 30074, Na ional Bu eau o
Economic Resea ch.
URL: h p://www.nbe .o g/pape s/w30074
Au o , D. H., Le y, F. & Mu nane, R. J. (2003), ‘The skill con en o ecen ech-
nological change: An empi ical explo a ion’, The Qua e ly Jou nal o Economics
118(4), 1279–1333.
Balassa, B. (1965), ‘T ade Libe alisa ion and Re ealed Compa a i e Ad an age’, The Manch-
es e School 33(2), 99–123.
Balland, P.-A., B oekel, T., Dioda o, D., Giuliani, E., Hausmann, R., O’Cle y, N. & Rigby,
D. (2022), ‘The new pa adigm o economic complexi y’, Resea ch Policy 51(3), 104450.
URL: h ps://doi.o g/10.1016/j. espol.2021.104450
Balland, P.-A., Ja a-Figue oa, C., Pe alia, S. G., S eijn, M. P. A., Rigby, D. L. & Hidalgo,
C. A. (2020), ‘Complex economic ac i i ies concen a ein la ge ci ies’, Na u e Human Be-
ha iou 4, 248–254.
B uno, M., Mazzilli, D., Pa elli, A., Squa ini, T. & Sa acco, F. (2023), ‘In e ing compa a i e
ad an age ia en opy maximiza ion’. _ep in : 2304.12245.
B ynjol sson, E. & McA ee, A. (2011), Race agains he machine: How he digi al e olu ion
is accele a ing inno a ion, d i ing p oduc i i y, and i e e sibly ans o ming employmen
and he economy, B ynjol sson and McA ee.
28
Bus os, S., Gomez, C., Hausmann, R. & Hidalgo, C. A. (2012), ‘The Dynamics o Nes edness
P edic s he E olu ion o Indus ial Ecosys ems’, PLoS ONE 7(11), 1–8.
Calda ola, B., Mazzilli, D., Napoli ano, L., Pa elli, A. & Sba della, A. (2023), ‘Economic
complexi y and he sus ainabili y ansi ion: A e iew o da a, me hods, and li e a u e’.
Cas añeda, G., Pie one o, L., Rome o-Padilla, J. & Zacca ia, A. (2022), ‘The complex
dynamic o g ow h: Fi ness and he di e en pa e ns o economic ac i i y in he medium
and long e ms’, S uc u al Change and Economic Dynamics 62, 231–246.
URL: h ps://www.sciencedi ec .com/science/a icle/pii/S0954349X22000601
Cimoli, M. & Dosi, G. (1995), ‘Technological pa adigms, pa e ns o lea ning and de elop-
men : An in oduc o y oadmap’, Jou nal o E olu iona y Economics 5(3), 243–268.
URL: h ps://doi.o g/10.1007/BF01198306
C is elli, M., Gab ielli, A., Tacchella, A., Calda elli, G. & Pie one o, L. (2013), ‘Measu ing
he in angibles: A me ics o he economic complexi y o coun ies and p oduc s’, PLoS
ONE 8(8).
Dau h, W., Findeisen, S., Suedekum, J. & Woessne , N. (2021), ‘The Adjus men o Labo
Ma ke s o Robo s’, Jou nal o he Eu opean Economic Associa ion 19(6), 3104–3153.
URL: h ps://doi.o g/10.1093/jeea/j ab012
de V ies, G., Timme , M. & de V ies, K. (2015), ‘S uc u al T ans o ma ion in A ica: S a ic
Gains, Dynamic Losses’, The Jou nal o De elopmen S udies 51(6), 674–688.
Fab ican , S. (1942), Employmen in manu ac u ing, 1899-1939: An analysis o i s ela ion
o he olume o p oduc ion, Na ional Bu eau o Economic Resea ch, Inc.
F ei e, C. (2021), Economic Complexi y Pe spec i es on S uc u al Change, in ‘New Pe -
spec i es on S uc u al Change’, Ox o d Uni e si y P ess, pp. 188–214.
F ey, C. B. & Osbo ne, M. A. (2017), ‘The u u e o employmen : How suscep ible a e jobs
o compu e isa ion?’, Technological Fo ecas ing and Social Change 114, 254–280.
Ga laschelli, D. & Lo edo, M. I. (2008), ‘Maximum likelihood: Ex ac ing unbiased in-
o ma ion om complex ne wo ks’, Physical Re iew E - S a is ical, Nonlinea , and So
Ma e Physics 78, 1–5.
Goos, M., Manning, A. & Salomons, A. (2009), ‘Job pola iza ion in eu ope’, Ame ican Eco-
nomic Re iew 99(2), 58–63.
URL: h ps://www.aeaweb.o g/a icles?id=10.1257/ae .99.2.58
29
Goos, M., Manning, A. & Salomons, A. (2014), ‘Explaining job pola iza ion: Rou ine-biased
echnological change and o sho ing’, Ame ican Economic Re iew 104(8), 2509–26.
URL: h ps://www.aeaweb.o g/a icles?id=10.1257/ae .104.8.2509
G ae z, G. & Michaels, G. (2018), ‘Robo s a Wo k’, 100(5), 753–768.
URL: h ps://doi.o g/10.1162/ es _a_00754
Hane-Weijman, E., E iksson, R. H. & Henning, M. (2018), ‘Re u ning o wo k: egional
de e minan s o e-employmen a e majo edundancies’, Regional S udies 52, 768 – 780.
URL: h ps://api.seman icschola .o g/Co pusID:53615559
Ha mann, D., Gue a a, M. R., Ja a-Figue oa, C., A is a án, M. & Hidalgo, C. A. (2017),
‘Linking economic complexi y, ins i u ions, and income inequali y’, Wo ld De elopmen
93, 75–93.
URL: h ps://www.sciencedi ec .com/science/a icle/pii/S0305750X15309876
Ha mann, D. & Pinhei o, F. L. (2022), ‘Economic complexi y and inequali y a he na ional
and egional le el’.
Hausmann, R. & Hidalgo, C. (2009), ‘The building blocks o economic complexi y’,
P oceedings o he Na ional Academy o Sciences o he Uni ed S a es o Ame ica
106(26), 10570–10575.
Hausmann, R. & Klinge , B. (2006), S uc u al ans o ma ion and pa e ns o compa a i e
ad an age in he p oduc space, CID Wo king Pape RWP06-041, Cen e o In e na ional
De elopmen , Ha a d Uni e si y.
He endo , B., Roge son, R. & Valen inyi, A. (2014), G ow h and s uc u al ans o ma ion,
in ‘Handbook o economic g ow h’, Vol. 2, Else ie , pp. 855–941.
He endo , B., Roge son, R. & Ákos Valen inyi (2022), New e idence on sec o al labou
p oduc i i y: implica ions o indus ializa ion and de elopmen , Wo king Pape 29834,
NBER.
Hi schman, A. O. (1958), The S a egy o Economic De elopmen , Yale Uni e si y P ess,
New Ha en.
K an z, R., Gemme o, V. & Ga laschelli, D. (2018), ‘Maximum-en opy ools o economic
i ness and complexi y’, En opy 20(10).
URL: h ps://www.mdpi.com/1099-4300/20/10/743
30
Kuzne s, S. (1955), ‘Economic g ow h and income inequali y’, The Ame ican Economic Re-
iew 45(1), 1–28.
URL: h p://www.js o .o g/s able/1811581
Lewis, A. (1954), ‘Economic De elopmen wi h Unlimi ed Supplies o Labou ’, The Manch-
es e School 22(8), 139–191.
Lo Tu co, A. & Maggioni, D. (2022), ‘The knowledge and skill con en o p oduc ion com-
plexi y’, Resea ch Policy 51(8), 104059. Special Issue on Economic Complexi y.
URL: h ps://www.sciencedi ec .com/science/a icle/pii/S0048733320301372
Mazzilli, D., Ma iani, M. S., Mo one, F. & Pa elli, A. (2024), ‘Equi alence be ween he
i ness-complexi y and he sinkho n-knopp algo i hms’, Jou nal o Physics: Complexi y .
McMillan, M., Rod ik, D. & Sepúl eda, C., eds (2017), S uc u al change, undamen als,
and g ow h: A amewo k and case s udies, In e na ional Food Policy Resea ch Ins i u e
(IFPRI), Washing on, D.C.
McMillan, M., Rod ik, D. & Íñigo Ve duzco-Gallo (2014), ‘Globaliza ion, S uc u al Change,
and P oduc i i y G ow h, wi h an Upda e on A ica’, Wo ld De elopmen 63, 11–32.
McNe ney, J., Li, Y., Gomez-Lie ano, A. & Ne ke, F. (2023), ‘B idging he sho - e m and
long- e m dynamics o economic s uc u al change’.
Pa elli, A., Napoli ano, L., Cimini, G., Pugliese, E. & Gab ielli, A. (2023), ‘Capabili y
accumula ion pa e ns ac oss economic, inno a ion, and knowledge-p oduc ion ac i i ies’,
Scien i ic Repo s 13(1), 12988.
URL: h ps://doi.o g/10.1038/s41598-023-29979-x
Pinhei o, F. L., Balland, P.-A., Boschma, R. & Ha mann, D. (2022), ‘The da k side o
he geog aphy o inno a ion: ela edness, complexi y and egional inequali y in eu ope’,
Regional S udies Fo hcoming special issue: The da k side o inno a ion and i s
geog aphy(0), 1–16.
URL: h ps://doi.o g/10.1080/00343404.2022.2106362
Sba della, A., Pugliese, E. & Pie one o, L. (2017), ‘Economic de elopmen and wage in-
equali y: A complex sys em analysis’, PLOS ONE 12(9), 1–26.
Schumpe e , J. A. (1934), The Theo y o Economic De elopmen : An Inqui y in o P o i s,
Capi al, C edi , In e es , and he Business Cycle, Ha a d Uni e si y P ess, Camb idge,
MA.
31
Sy quin, M. (1988), Pa e ns o s uc u al change, in H. Chene y & T. S ini asan, eds,
‘Handbook o De elopmen Economics’, 1 edn, Vol. 1, Else ie , chap e 07, pp. 203–273.
URL: h ps://EconPape s. epec.o g/RePEc:eee:de chp:1-07
Tacchella, A., C is elli, M., Calda elli, G., Gab ielli, A. & Pie one o, L. (2012), ‘A New
Me ics o Coun ies Fi ness and P oduc s Complexi y’, Scien i ic Repo s 2(723), 1–7.
Tacchella, A., Mazzilli, D. & Pie one o, L. (2018), ‘A dynamical sys ems app oach o g oss
domes ic p oduc o ecas ing’, Na u e Physics 14(8), 861–865.
Teece, D. J., Rumel , R., Dosi, G. & Win e , S. (1994), ‘Unde s anding co po a e cohe ence:
Theo y and e idence’, Jou nal o Economic Beha io & O ganiza ion 23(1), 1–30.
Voll a h, T. L. (1991), ‘A heo e ical e alua ion o al e na i e ade in ensi y measu es o
e ealed compa a i e ad an age’, Wel wi scha liches A chi 127(2), 265–280.
URL: h p://www.js o .o g/s able/40439943
32
A Da a econs uc ion
The Eu os a ’s S uc u al Business Su ey da a used o compu e he employmen based EFC
measu es used in his pape a e one o he mos comp ehensi e and de ailed da a sou ces
o employmen in Eu pean indus ies, classi ied a a ine le el o disagg ega ion (NACE 4
digi ). Howe e , he da ase p esen s a la ge numbe o missing alues, which hampe he
cons uc ion o a specialisa ion ma ix as explained in Sec ion 4.1. In ac , he BiWCM and
EFC algo i hm equi e bo h a ma ix wi h comple e in o ma ion.
In o de o add ess his sho coming in he SBS da a, we ha e pu sued, compa ed and
alida ed se en di e en in e pola ion s a egies, in o de o e ie e a good app oxima ion
o he missing da a. The i s s ep has been o a emp he econs uc ion o he SBS
da a ollowing 7 al e na i e s a egies, applied o he coun y-indus y ime se ies ha ing a
maximum 7 missing da a poin s5:
1. In e nal missing alues ( hose a e he i s a ailable da a poin , and be o e he las ) a e
in e pola ed using linea in e pola ion, while ex e nal missing alues (be o e and a e ,
espec i ely, he i s and las da a poin a ailable) ha e been ex apola ed backwa ds
aking he i s a ailable alue as cons an ;
2. In e nal missing alues a e in e pola ed using linea in e pola ion, while ex e nal miss-
ing alues a e ex apola ed using he closes g ow h a e a ailable, ob ained by a e
in e pola ing in e nal missing alues;
3. In e nal missing alues a e in e pola ed using linea in e pola ion, while ex e nal miss-
ing alues a e ex apola ed using a e age g ow h a e applied o he is and las da a
poin s a ailable a e in e pola ing. This s a egy p oduced nega i e alues, which we e
cons ained o 0;
4. In e nal missing alues a e in e pola ed using linea in e pola ion, while ex e nal miss-
ing alues a e ex apola ed using he i s a ailable g ow h a e by illing up/downwa ds
a ailable g ow h a es;
5. In e nal missing alues a e in e pola ed using linea in e pola ion, while ex e nal miss-
ing alues a e ex apola ed using mo ing a e age o g ow h a es, cons uc ed using a
olling window o 3 yea s. The i s wo yea s a e always ‘NA‘; second yea is illed wi h
mo ing a e age o =2, and i s yea is jus he g ow h wi h espec o he p e ious
yea ;
5In he case o ully missing se ies, he coun y-indus y couple has been excluded. In case o 8 ou o
9 missing da a poin s, he alue has been assumed o be cons an ac oss he ime se ies, ex apola ing a
cons an alues om he only a ailable da a poin in he se ies.
33
6. In e pola e and ex apola e wi h linea i compu ed using he a ailable da a poin s in
each coun y-indus y se ies;
7. In e nal missing alues a e in e pola ed using linea in e pola ion, while ex e nal miss-
ing alues a e ex apola ed using linea i .
Each s a egy is compa ed and alida ed using in o ma ion om he comple e coun y-
indus y ime se ies. A e il e ing only he comple e se ies om he da ase , we ha e c ea ed
i e di e en alida ion subsamples by andomising missing alues in he comple e da ase ,
imposing he ollowing condi ions:
•Re lec NA equency in o iginal da a (14% o da a missing)
•Re lec equency in o iginal da a ś 5%,
•Re lec equency in o iginal da a + 10%
•Impose 50% o missing alues.
In o de o pick he da a econs uc ion s a egy ha bes app oxima es he obse ed
da a, we compa e he Mean Absolu e E o o each s a egy ac oss he 5 alida ion samples.
As shown by able 7, he s a egy ha minimises he p edic ion e o ac oss he di e en
alida ion subsamples is he i s – i.e. he one ha in e pola es in e nal alues linea ly, and
ex apola es ex e nal alues aking he i s and las a ailable da a poin as cons an , bo h
backwa d and o wa d.
Table 7: Mean Absolu e E o s
Recons uc ion s a egy
Subsample 1 2 3 4 5 6 7
NA 9% 174.96 189.05 179.47 192.23 189.44 222.55 197.73
NA 14% 214.95 227.60 215.10 230.20 227.45 273.22 240.86
NA 19% 254.61 294.14 267.06 297.80 292.47 323.64 286.27
NA 24% 311.79 329.90 315.13 331.72 325.46 395.41 355.26
NA 50% 692.86 788.71 749.16 796.20 784.38 819.18 777.39
34
The UNU-MERIT WORKING Pape Se ies
2024-01 The g een ans o ma ion as a new di ec ion o echno-economic de elopmen by
Rasmus Lema and Ca lo a Pe ez
2024-02 E idence on an endogenous g ow h model wi h public R&D by Thomas H.W.
Zieseme
2024-03 Highe Educa ional Sys ems and E- esilience by Mindel an de Laa , Julie a Ma o a
and Lisa de G aa
2024-04 Es ima ing he wage p emia o e ugee immig an s: Lessons om Sweden by
Ch is ophe F. Baum, Hans Löö , And eas S ephan and Klaus F. Zimme mann
2024-05 G een window o oppo uni y h ough global alue chains o c i ical mine als: An
empi ical es o e ining coppe and li hium indus ies By Jo ge Val e de-
Ca bonell and Alejand o Micco A.
2024-06 Mig a ion and Consump ion by Robe a Misu aca and Klaus F. Zimme mann
2024-07 The powe o na a i es: How aming e ugee mig a ion impac s a i udes and
poli ical suppo owa ds e ugees by Tobias Hillenb and, B uno Ma o ano, Lau a
Me zge and Melissa Siegel
2024-08 S a egies o sea ch and pa en ing unde di e en IPR egimes by Robin Cowan,
Nicolas Jona d and Ru h Samson
2024-09 F om d ains o b idges: The ole o in e na ionally mobile PhD s uden s in linking
non-mobile wi h o eign scien is s by Rod igo I o, Diego Cha a o, Tommaso Cia li,
Robin Cowan and Fabiana Visen in
2024-10 Expanding ho izons: The impac o pos -doc o al mobili y g an s on esea ch
di e si ica ion by Yağmu Yıldız, Diego Cha a o, Fabiana Visen in and Tommaso
Cia li
2024-11 Sco ching hea and sh inking ho izons: The impac o ising empe a u es on
ma iages and mig a ion in u al India by Manisha Mukhe jee, B uno Ma o ano
and Melissa Siegel
2024-12 Labou ma ke in eg a ion o e ugees: A social jus ice amewo k by Tama a Kool
and Zina Nimeh
2024-13 The e olu ion o de elopmen wi h ade in global alue chains by Önde Nomale
and Ba Ve spagen
2024-14 The impac s o he Mic o inance Mul iplied app oach on seasonal ood insecu i y:
E idence om a high- equency panel su ey in Uganda by Rica do Mo el, F anziska
Gassmann, B uno Ma o ano, Nyasha Ti i ayi and John Kamau
2024-15 Building CoPS Capabili y o Ca ching Up Du ing T ansi ions: The Case o ACWA
Powe in Saudi A abia by Kao i Mi a and Michiko Iizuka
2024-16 The KSTE+I app oach and he AI echnologies: e idence om he Eu opean egions
by F ancesco D’Allesand o, En ico San a elli and Ma co Vi a elli
2024-17 Theo y and measu emen in SFC models: he ole o he inancial sec o by Huub
Meije s and Joan Muysken
2024-18 Is a i icial in elligence gene a ing a new pa adigm? E idence om he eme ging
phase by Giacomo Damioli, Vincen Van Roy, Daniel Ve esy and Ma co Vi a elli
2024-19 Indus ializing A ica - a science, echnology, and inno a ion policy pe spec i e by
Be ha Vallejo
2024-20 Inno a ion in he wind ene gy sec o by Rasmus Lema, Da ide Bonaglia and Ul ich
Elme Hansen
2024-21 The In es men Landscape in he Li hium T iangle: Implica ions o he EU’s C i ical
Raw Ma e ials Ac Agenda by Bea iz Calzada Ol e a and Melissa Ve ga a-
Fe nández
2024-22 3G in e ne di usion and seconda y educa ion a ainmen : E idence o oppo uni y
cos in Vie Nam by T ang Thi Pham and Be na do Calda ola
2024-23 In luence o he Sec o al Composi ion o P oduc ion on Occupa ional and Sec o al
Wage S uc u es: E idence om Mul iple Economies by Nelson Ma coni, Danilo
Spinola, Tiago Po o and Eliane A aújo
2024-24 T ansi ion om In o ma ion and Communica ions Technology o Educa ional
Technologies and Game-Based Lea ning in he Republic o Moldo a: The Case o
The Mobili y E a Game by I ina Bu lacu de Paoli
2024-25 When one eaches, wo lea n: The bidi ec ional lea ning p ocess supe iso -PhD
s uden by Pie e Bou os, Michele Pezzoni, So a o Shibayama and Fabiana Visen in
2024-26 Cloud echnologies, i m g ow h and indus y concen a ion: E idence om F ance
by Be na do Calda ola and Luca Fon anelli
2024-27 The Inno a ion and S uc u al T ans o ma ion Da abase: A Guide by Neil Fos e -
McG ego , Önde Nomale & Ba Ve spagen
2024-28 The Poli ical Economy o Ta ge ing: An Empi ical Assessmen o Poland’s Family
500+ P og amme by S e anie Roos and F anziska Gassmann
2024-29 Melke Jobs, a unique expe imen by Kees Mosselman and Joan Muysken
2024-30 A pos -mo em o in e es a e policy: bewa e o inancial agili y by Huub Meije s
and Joan Muysken
2024-31 Rankings and Job Ma ke Dynamics: A Model o Academic Science by Robin Cowan
and Nicolas Jona d
2024-32 Twin ansi ion ade based on mul i-dimensional economic complexi y by Ma ía
de las Me cedes Menéndez
2024-33 S uc u al Change, Employmen , and Inequali y in Eu ope: an Economic Complexi y
App oach by Be na do Calda ola, Da io Mazzilli, Au elio Pa elli and Angelica
Sba della
