The role of between- and within-occupation differences in wage inequality trends in Europe (2002-2018)

O ao e Vale Tabe ne o, Guille mo; Fe nández-Macías, En ique; Malo Ocaña,
Miguel Angel
Wo king Pape
The ole o be ween- and wi hin-occupa ion di e ences in
wage inequali y ends in Eu ope (2002-2018)
JRC Wo king Pape s Se ies on Labou , Educa ion and Technology, No. 2024/01
P o ided in Coope a ion wi h:
Join Resea ch Cen e (JRC), Eu opean Commission
Sugges ed Ci a ion: O ao e Vale Tabe ne o, Guille mo; Fe nández-Macías, En ique; Malo Ocaña,
Miguel Angel (2024) : The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y
ends in Eu ope (2002-2018), JRC Wo king Pape s Se ies on Labou , Educa ion and Technology, No.
2024/01, Eu opean Commission, Join Resea ch Cen e (JRC), Se ille
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/306583
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by/4.0/
Cen e
The ole o be ween- and wi hin-occupa ion
di e ences in wage inequali y ends in Eu ope
(2002-2018)
JRC Wo king Pape s Se ies on
Labou , educa ion and Technology
2024/01
G. O ao, E. Fe nández-Macías and M. A. Malo
Au ho 1, Au ho 2, Au ho 3 (Co e )
This publica ion is pa o a wo king pape se ies on Labou , Educa ion and Technology by he Join
Resea ch Cen e (JRC). The JRC is he Eu opean Commission’s science and knowledge se ice.
I aims o p o ide e idence-based scien i ic suppo o he Eu opean policymaking p ocess. The
con en s o his publica ion do no necessa ily e lec he posi ion o opinion o he Eu opean
Commission. Nei he he Eu opean Commission no any pe son ac ing on behal o he Commission
is esponsible o he use ha migh be made o his publica ion. Fo in o ma ion on he
me hodology and quali y unde lying he da a used in his publica ion o which he sou ce is nei he
Eu os a no o he Commission se ices, use s should con ac he e e enced sou ce. The
designa ions employed and he p esen a ion o ma e ial on he maps do no imply he exp ession
o any opinion wha soe e on he pa o he Eu opean Union conce ning he legal s a us o any
coun y, e i o y, ci y o a ea o o i s au ho i ies, o conce ning he delimi a ion o i s on ie s o
bounda ies.
Con ac in o ma ion
Name: En ique Fe nández-Macías
Add ess: Join Resea ch Cen e, Eu opean Commission (Se ille, Spain)
Email: En [email protected] opa.eu
EU Science Hub
h ps://join - esea ch-cen e.ec.eu opa.eu
JRC137944
Se ille: Eu opean Commission, 2024
© Eu opean Union, 2024
The euse policy o he Eu opean Commission is implemen ed by he Commission Decision
2011/833/EU o 12 Decembe 2011 on he euse o Commission documen s (OJ L 330, 14.12.2011,
p. 39). Excep o he wise no ed, he euse o his documen is au ho ised unde he C ea i e Com-
mons A ibu ion 4.0 In e na ional (CC BY 4.0) licence (h ps://c ea i ecommons.o g/licenses/by/4.0/).
This means ha euse is allowed p o ided app op ia e c edi is gi en and any changes a e indica ed.
Fo any use o ep oduc ion o pho os o o he ma e ial ha is no owned by he EU, pe mission
mus be sough di ec ly om he copy igh holde s.
All con en © Eu opean Union 2024
How o ci e his epo : O ao, G., Fe nández-Macías, E. and Malo, M.A., The Role o Be ween- and
Wi hin-Occupa ion Di e ences in Wage Inequali y T ends in EU-28 Coun ies (2002-2018), JRC
Wo king Pape s on Labou , Educa ion and Technology 2024/01, Eu opean Commission, Se ille,
2024, JRC137944.
Con en s
Execu i e summa y ............................................................................................................................................................ 2
1 In oduc ion ................................................................................................................................................................... 3
2 Li e a u e Re iew ....................................................................................................................................................... 3
3 Da a and me hodology app oach ..................................................................................................................... 5
3.1 Da a ..................................................................................................................................................................... 5
3.2 Me hodology app oach ............................................................................................................................. 7
4 Resul s .............................................................................................................................................................................. 8
4.1 Wage inequali y and occupa ional s uc u e change in Eu ope ........................................ 8
4.2 Wage inequali y be ween- and wi hin-jobs ............................................................................... 14
5 Conclusions ................................................................................................................................................................. 21
Re e ences ............................................................................................................................................................................ 23
Lis o abb e ia ions and de ini ions ..................................................................................................................... 25
Lis o igu es...................................................................................................................................................................... 26
Lis o ables ....................................................................................................................................................................... 27
Appendix ................................................................................................................................................................................ 28
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
1
The ole o be ween- and wi hin-occupa ion
di e ences in wage inequali y ends in Eu ope
(2002-2018)
O ao, G. (Uni e si y o Salamanca), Fe nández-Macías, E. (Join Resea ch Cen e o he Eu opean
Commission) and Malo, M.A. (Uni e si y o Salamanca)
Abs ac
This wo king pape p esen s a compa a i e analysis o he ole played by occupa ional changes in
ecen wage inequali y ends in six Eu opean coun ies be ween 2002 and 2018. Using he Eu opean
Union S uc u e o Ea nings Su ey, he analysis shows wo pa e ns in he sha e o wage inequali y
explained by be ween-occupa ion di e en ials: while he ela i e impo ance o be ween-occupa ion
ends has g own in Finland and he UK, i has diminished in Spain, F ance, Poland and Romania.
Al hough be ween-occupa ion di e en ials accoun o a g ea sha e o o al wages’ a iance,
changes in he occupa ional s uc u e (in pa icula , he pa e ns o job pola isa ion and upg ading
widely discussed in he li e a u e) ha e no d i en ecen wage inequali y ends in Eu ope. Wage
inequali y, ins ead, has been mos ly d i en by changes in wage di e en ials wi hin occupa ions.
Finally, we ound ha occupa ions e ec i ely accoun o he dis ibu ion o wages, ye hei
explana o y signi icance ma kedly declines a he highes wage ie s. This wo k con ibu es o a be e
unde s anding o how wi hin- and be ween-occupa ion di e ences ha e in luenced wage inequali y
ends in Eu ope. Consequen ly, ou esul s add signi ican alue o he deba e abou ecen
s a i ica ion heo y, which has challenged he idea ha occupa ions s uc u e economic dispa i ies
and wage inequali y as impo an ly as hey once did.
Keywo ds: Eu ope, jobs, occupa ional s uc u e, occupa ions, wage inequali y.
Join Resea ch Cen e e e ence numbe : JRC137944
Rela ed publica ions and epo s:
Ma iscal-De-Gan e, Á., Palencia Es eban, A., G ubano -Bosko ic, S. and Fe nandez Macias, E.,
Feminiza ion, ageing and occupa ional change in Eu ope in he las 25 yea s, JRC Wo king Pape s
Se ies on Labou , educa ion and Technology 2023/04, Eu opean Commission, 2023, JRC132328.
h ps://publica ions.j c.ec.eu opa.eu/ eposi o y/handle/JRC132328
To ejon Pe ez, S., Hu ley, J., Fe nandez Macias, E. and S a a, E., Employmen shi s in Eu ope om
1997 o 2021: om job upg ading o pola isa ion, JRC Wo king Pape s Se ies on Labou , Educa ion
and Technology 2023/05, Eu opean Commission, 2023, JRC132678.
h ps://publica ions.j c.ec.eu opa.eu/ eposi o y/handle/JRC132678
Acknowledgemen s
This wo k has been suppo ed by he Join Resea ch Cen e o he Eu opean Commission
[JRC/SVQ/2023/VLPV/2903]; and he Jun a de Cas illa & León co- unded by he Eu opean Social
Fund unde g an [EDU/601/2020 o G.O.]. We also acknowledge Eu os a o he da a p o ided o
de elop he esea ch p oposal RPP 328/2023-SES.

The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
2
Execu i e summa y
This wo king pape p esen s a compa a i e analysis o he ole played by occupa ional changes in
ecen wage inequali y ends in six Eu opean coun ies be ween 2002 and 2018. In pa icula , he
analysis conside s he changes in he occupa ional s uc u e, he e olu ion o main inequali y indices
and he a ia ions in he ela i e weigh o bo h wi hin- and be ween-job di e en ials in explaining
wage inequali y.
Policy con ex
Eu opean labou ma ke s a e con inuously unde going shi s in hei occupa ional s uc u es.
Technological changes and in e na ional ade ha e d i en a di e si y o pa e ns o occupa ional
change in Eu ope, including he pola isa ion, upg ading and downg ading p ocesses. A he same ime,
wage inequali y has g own in many o hese economies, especially since he 1980s. This aises he
ques ion o whe he he end in wage inequali y can be a leas pa ly d i en by occupa ional
changes.
Key conclusions
The e has been a dec ease o s agna ion o wage inequali y in all Eu opean coun ies wi h he
excep ion o Finland be ween 2002 and 2018. The ole o occupa ions g ew signi ican ly in Finland
and he UK, while hei ela i e weigh dec eased in Spain, F ance, Poland and Romania. Howe e , we
conclude ha , in gene al, occupa ional dynamics did no d i e ecen ends in wage inequali y in
Eu ope since wi hin-job di e en ials o e shadowed hei e ec . Ou indings suppo he idea ha
occupa ions no longe s uc u e economic dispa i ies and wage inequali y as impo an ly as hey once
did. E en conside ing ha occupa ions ha e gained ela i ely mo e weigh in wage inequali y ends
in some EU coun ies, such as Finland and he UK, ou esul s a e mo e in line wi h he a gumen ha
ends in wage inequali y a e mos ly d i en by di e ences wi hin-occupa ion o he bulk o Eu opean
coun ies ollowed by changes in he mean wages o occupa ions.
Main indings
The Theil decomposi ion esul s e eal ha he e a e di e en pa e ns in he sha e o wage
inequali y explained by be ween-occupa ion di e ences. While he ole o occupa ions g ew
signi ican ly om 2002 o 2018 in Finland and he UK, hei ela i e weigh dec eased in Spain,
F ance, Poland and Romania. These indings con i m ha he e is no unique end in e ms o he ole
o occupa ions in he de elopmen o wage inequali y in Eu ope. The e olu ion o bo h wi hin and
be ween componen s sugges s ha occupa ional change, which di e s conside ably ac oss EU
coun ies, had a posi i e e ec on wage inequali y ends. Howe e , he nega i e e ec o changes
in wi hin-occupa ion di e en ials and he mean wages o occupa ions we e mo e subs an ial, and
mo e ele an o explaining ecen wage inequali y ends.
Be ween-occupa ion di e en ials accoun o a ound 35-60 pe cen o o al wages a iance.
Howe e , signi ican di e ences a e ound by coun y. While in Spain and F ance occupa ions explain
abou a 35-40% o his a iance, in Poland and Romania, hey explain up o a 45-50%. Mo eo e , in
Finland and he UK, he wage a iance accoun ed o by occupa ions eaches alues o up o 60%.
Finally, occupa ions play a signi ican ole in he dis ibu ion o he wages o mos wo ke s, bu a
ma ginal one in he dis ibu ion o e y la ge wages.
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
3
1 In oduc ion
The aim o his pape is o analyse whe he occupa ional changes ha e d i en wage inequali y ends
o e ecen decades in Eu ope. Eu opean labou ma ke s, like hose o any o he pa o he wo ld,
a e con inuously unde going shi s in hei occupa ional s uc u es. Acco ding o many US s udies,
echnological changes and in e na ional ade ha e d i en a pola isa ion o employmen , o in o he
wo ds, a ela i e inc ease in he demand o low- and high-skilled wo ke s ela i e o hose in he
middle (W igh and Dwye , 2003; Au o , 2015; Au o e al., 2006). This e idence is mo e mixed o
Eu ope, whe e some s udies ind pe asi e pola isa ion (Goos e al., 2009), some mos ly upg ading
(Oesch and Picci o, 2019), and some a di e si y o pa e ns o occupa ional change (Eu o ound,
2017; To ejón-Pé ez e al., 2023). Ou side Eu ope and he U.S., he pa e n ha p edomina es is one
o occupa ional upg ading (To ejón-Pé ez e al., 2024). In pa allel, wage inequali y has g own in
many o hese economies, especially since he 1980s (OECD, 2011), al hough asymme ically in
Eu opean coun ies (Pe ei a and Galego, 2019). This aises he ques ion o whe he his s eady
inc ease o wage inequali y can be a leas pa ly d i en by occupa ional changes.
In he deba e on he ac o s ha explain wage inequali ies, some au ho s ha e highligh ed he key
ole ha occupa ions play abo e o he elemen s such as educa ion (Acemoglu and Au o , 2011;
Williams, 2013). None heless, he e is no consensus on whe he i is he di e ences be ween- o
wi hin-occupa ions, o he change in occupa ional sha es, ha d i e wage inequali y. While ce ain
au ho s assign a p edominan ole o wi hin-occupa ion di e en ials as a de e minan o ecen
ends in wage inequali y (Fe nández-Macías and A anz-Muñoz, 2020; Godecho , 2012; Kim and
Sakamo o, 2008; Mishel e al., 2013), o he s a gue ins ead ha i is p ecisely he di e ences be ween
occupa ions ha d i e his e olu ion (Acemoglu and Au o , 2011; Helland e al., 2017; Weeden e al.,
2007; Williams, 2013). A s iking ac is ha al hough some o hese s udies use he same da a, hey
o en ind di e en o e en opposi e esul s due o he di e en ope a ionalisa ion. Howe e ,
compa a i e s udies a e sca ce, pe haps due o he lack o homogeneous and ha monized da a o
apply he same ope a ionalisa ion ac oss coun ies.
In a ecen pape , Fe nández-Macías and A anz-Muñoz (2020) made a i s app oxima ion o
measu e he ole o occupa ional di e ences in wage inequali y e olu ion in i e Eu opean coun ies
o he pe iod 2005-2014, using da a om he Eu opean Su ey on Income and Li ing Condi ions
(EU-SILC). Howe e , hese au ho s emphasize he limi a ions o he da a used o hei analysis,
conce ning he measu e o wages (calcula ed as ull- ime equi alen a he han hou ly, and
es ima ed a he han measu ed di ec ly) as well as he sample size, le el o de ail in he analysis
and pe iod co e ed (2005-2014). In he p esen a icle, we y o add o his li e a u e by analysing
he ole o occupa ional change in wage inequali y ends in Eu ope in ecen yea s. We base ou
analysis on a wide ime ame (2002-2018) o a se o six EU coun ies wi h di e en economic
s uc u es and ins i u ional con ex s, using a di ec measu e o hou ly wages as dependen a iable
and bene i ing om he la ge sample size and le el o de ail o he Eu opean S uc u e o Ea nings
Su ey.
2 Li e a u e Re iew
Taking he de ini ion o income inequali y p oposed by he O ganisa ion o Economic Co-ope a ion
and De elopmen (OECD) as a e e ence (OECD, 2015), we de ine wage inequali y as how une enly
wages a e dis ibu ed wi hin a popula ion, in his pa icula case he wo k o ce. Wage inequali y is a
phenomenon ha has been widely s udied in he li e a u e, especially in de eloped economies, such
as he Uni ed S a es (Au o , 2014; Mouw and Kallebe g, 2010; Weeden e al., 2007) and Eu opean
coun ies (Godecho , 2012; Helland e al., 2017; Williams, 2013). In his espec , sec o s and
occupa ions (Ake man e al., 2013), globalisa ion (Helpman, 2018), wi hin- i ms he e ogenei y
(Schae e and Single on, 2020; Song e al., 2019), ade unions (Ca d, 2001; Pon usson, 2013), o
mo e ecen ly obo isa ion (Ba h e al., 2020), ha e been in es iga ed as some o he possible d i e s
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
4
o wage inequali y ends. Mo eo e , Acemoglu and Au o (2011) and Williams (2013) a gue ha
occupa ions may be mo e impo an han o he ac o s, such as educa ion, when explaining wage
inequali y ends in he U.S. and he UK espec i ely.
F om a heo ical poin o iew, we can iden i y ou mechanisms by which changes in occupa ions
ela e o a ia ions in o e all wage inequali y. The i s wo mechanisms would ela e changes in
wage inequali y o a ia ions in be ween-occupa ion di e ences, while he las wo ela e o changes
in he di e en ials wi hin-occupa ion. Fi s ly, wage inequali y may ise due o an inc ease in he
di e ences be ween he mean wages o occupa ions, e en i he occupa ional s uc u e i sel does
no change (Fe nández-Macías and A anz-Muñoz, 2020; Kim and Sakamo o, 2008). Secondly,
changes in he s uc u e o employmen by occupa ion may a ec wage inequali y i he e is ela i e
g ow h o employmen in low- and high-paid occupa ions (Goos and Manning, 2007). This way, he
p ocesses known as job pola isa ion o occupa ional upg ading could lead o an inc ease in wage
inequali y (by inc easing he sha e o employmen in he ails o he wage dis ibu ion). Thi dly, o al
wage inequali y may also expand i employmen inc eases ela i ely as e in hose occupa ions wi h
g ea e in e nal di e ences in e ms o wages (Fe nández-Macías and A anz-Muñoz, 2020). Fou hly,
wage inequali y can ise i in e nal di e ences wi hin-occupa ion inc ease, o in o he wo ds, i he
dis ibu ion o wages wi hin-g oup becomes mo e unequal (Mookhe jee and Sho ocks, 1982).
The e is abundan empi ical esea ch abou he ole o occupa ional wage di e en ials in explaining
ecen ends in wage inequali y, usually ocusing on one de eloped economy. Howe e , hei indings
a e o en con adic o y. On he one hand, Helland e al. (2017) o No way, Williams (2013) o he
UK, o Acemoglu and Au o (2011) and Weeden e al. (2007) o he U.S., amongs o he s, ind ha
inc easing wage inequali y in hese coun ies was mainly d i en by be ween-occupa ion di e ences.
These au ho s highligh ha job pola isa ion and widening occupa ional wage di e en ials we e
d i ing wage inequali y om he 70s o he i s decade o he 21s cen u y. In his line, Mouw and
Kallebe g (2010) epo simila indings o he U.S. om 1992 o 2008, whe e be ween-occupa ion
di e en ials explain wo- hi ds o he inc ease in wage inequali y. Howe e , hey also ound ha
wi hin-occupa ion di e ences we e sho ing he p e ious inc ease om 1983 o 1992.
On he o he hand, and in line wi h his la es inding o Mouw and Kallebe g (2010), o he au ho s
s a e ha he ends in wage inequali y mos ly e lec wi hin-occupa ion a ia ions. By con as o
he a o emen ioned s udies ocused on he U.S., Kim and Sakamo o (2008) emphasized he
inc easing ole o wi hin-occupa ion di e ences wi h espec o be ween-occupa ion ones om 1983
o 2002. This was pa ly suppo ed by he e idence o Mishel e al. (2013), who using he same da a
as Acemoglu and Au o (2011), bu wi h a dis inc ope a ionalisa ion, ound ha he weigh o
be ween-occupa ion di e ences aised un il 1994, ollowed by a signi ican decline a e wa ds
subs i u ed by wi hin-occupa ion di e en ials as he main d i e o wage inequali y ends. In Eu ope,
and pa icula ly in Finland, F ance, Ge many, Spain and he Ne he lands, Fe nández-Macías and
A anz-Muñoz (2020) ind ha changes wi hin-occupa ion we e p edominan ly d i ing wage
inequali y om 2005 o 2014. Fu he mo e, Godecho (2012) shows ha wage inequali y g ew in
F ance due o he inc ease pay o op inance manage s, which mos ly a ec s wi hin-occupa ion
di e en ials.
The e a e a numbe o easons ha may explain he con adic o y e idence on he ole o occupa ions
in explaining wage inequali y ends in ecen yea s, ela ed o he me hodological challenges o doing
his kind o analysis (Fe nández-Macías and A anz-Muñoz, 2020). In e na ional occupa ional
classi ica ions ha e unde gone se e al upda es since hei in oduc ion in he 1970s, no ably in he
la e 1990s and a ound 2010. These upda es, necessa y due o changes in he di ision o labou and
echnological ad ancemen s, aim o p oduce mo e consis en occupa ional ca ego ies. Howe e , hey
complica e he long- e m compa abili y o wage da a. Fo ins ance, Mouw and Kallebe g (2010)
a ibu e a hi d o he inc ease in wage a iance explained by occupa ions in he 1990s o
eclassi ica ion. In gene al, long- e m compa isons o occupa ional ends a e p oblema ic due o he
sensi i i y o esul s o mino me hodological choices in how occupa ional codes a e ea ed o he
analysis.
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
5
Ano he majo issue is he g owing in luence o la ge ou lie s in he wage dis ibu ion. S udies sugges
ha ecen inequali y spikes la gely s em om massi e income inc eases o he op 1% o e en
0.1% (OECD, 2011; Pike y, 2014). S anda d su eys, like he U.S. Cu en Popula ion Su ey, o en
ail o accu a ely cap u e his end due o unde epo ing, spa se da a, non-con ac o e usal o high
ea ne s. Inc easingly, esea che s ha e u ned o adminis a i e egis e s and ax e u n da a o be e
unde s and ends in ea nings inequali y. This is impo an because mos s udies on occupa ional
wage ends ely on su eys, po en ially o e looking a key pa o ecen wage inequali y g ow h.
These p oblems can be compounded by he common p ac ice o using log wages ins ead o mone a y
alues o he analysis, which can obscu e he in e p e a ion, pa icula ly i la ge ou lie s a e a
common ea u e o he wage dis ibu ion (Fe nández-Macías e al., 2017: 39–40). Fo ins ance, Kim
and Sakamo o (2008) ound less e idence o g owing be ween-occupa ional di e en ials when
analysing dolla wages ins ead o log wages.
These me hodological p oblems may pa ly explain he lack o compa a i e analyses ac oss coun ies,
al hough he mos os ensible eason is he lack o da abases ha include a la ge numbe o coun ies
and consis en in o ma ion o wages and occupa ions o e a long pe iod o ime. Ne e heless, he
ecen wa es o he EU S uc u e o Ea nings Su ey (SES) gi e us an oppo uni y o empi ically s udy
he link be ween ecen changes in he occupa ional s uc u e and wage inequali y ends in Eu ope,
aking in o accoun he me hodological challenges jus discussed. In pa icula , he S uc u e o
Ea nings Su ey p o ides e y p ecise in o ma ion on indi idual hou ly wages which is c ucial when
measu ing hei a iance, especially in compa ison o wages’ in o ma ion collec ed h ough su eys.
1
P obably, adminis a i e eco ds o employe s and employees o e , in some coun ies, e en mo e
de ailed in o ma ion on wages. Howe e , hese adminis a i e sou ces a e no e enly a ailable ac oss
Eu opean coun ies, and hey a e a om homogeneous. The e o e, he SES is he highes -quali y
s a is ical sou ce o compa a i e analysis on wage e olu ion in he EU.
3 Da a and me hodology app oach
3.1 Da a
The da a used comes om he Eu opean S uc u e o Ea nings Su ey, compiled by Eu os a o he
pe iod 2002-2018. I is a 4-yea ly c oss-sec ional da abase on labou ea nings, he indi idual
cha ac e is ics o employees and hose o hei employe . I con ains consis en in o ma ion o he
whole pe iod analysed on occupa ion, sec o and wages, he main a iables necessa y o ou
analysis. The sample is e y big in all coun ies, which allows o a de ailed b eakdown o wages by
occupa ions and sec o s. Addi ionally, hou ly wages a e measu ed accu a ely and wi h a high deg ee
o de ail, which is one o he main ad an ages o his da abase in compa ison o o he s p e iously
used o simila analyses, such as he EU-SILC. Ou sample includes all indi iduals in paid
employmen , co e ing six Eu opean coun ies cha ac e ised by ha ing di e en wel a e s a e models:
Finland, F ance, Poland, Romania, Spain and he Uni ed Kingdom.
2
In pa icula , he sample only
includes hose indi iduals ac i ely wo king, in o he wo ds, he unemployed and wo ke s on lea e a e
no analysed. This da abase´s main sho coming is he lack o in o ma ion on some sec o s
3
and
1
Al e na i ely, wages can be ga he ed om he EU-SILC, which is he da abase used in mos ecen s udies. Howe e , i s
in o ma ion o wages is p o ided as mon hly wages o p e ious yea and he numbe o hou s wo ked pe week is no
a ailable. The e o e, some assump ions mus be es ablished, such as wages do no di e om p e ious yea and indi iduals
do no change om one occupa ion o ano he ( hese changes canno be cap u ed). Also, hou ly wages o pa - ime
employe s in he EU-SILC a e calcula ed as an app oxima ion conside ing ha hey wo k he maximum numbe o hou s
allowed by egula ion. The e o e, we can a i m ha he S uc u e o Ea nings Su ey da a is much mo e p ecise.
2
Da a on he UK is only a ailable un il 2014, he e o e we ha e only analysed he pe iod 2002-2014 o his coun y.
3
Sec o s A/A-B and L- o-Q/O- o-U (NACE Re .1.1/NACE Re .2) o ag icul u e, ishing and Public Adminis a ion, Educa ion,
Heal h, Household ac i i ies and ex a e i o ial ac i i ies we e missing o some coun ies. The e o e, hey ha e no been
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
12
demons a e he di e si y o Eu opean labou ma ke s and he idiosync asies ha each coun y ha e
and ha should be conside ed when analysing wage inequali y.
Acco ding o To ejón-Pé ez e al. (2023), in gene al, employmen g ow h du ing he la es decades
in Eu ope has been mainly d i en by job c ea ion in he p i a e se ice sec o al hough in a pola ised
way. Howe e , employmen in public se ices inc eased conside ably in mid and high-paid jobs.
Finally, he p ima y sec o has been sh inking in he las decades by a educ ion o he size o low-
and mid-paid jobs om 1997 o 2010. The na owing o he p ima y sec o is o he o he ac o s
(apa om he dynamic o public se ices) ha p omo ed a p ocess o job upg ading be o e he
G ea Recession and job pola isa ion a e i .
Ou analysis o EU S uc u e o Ea nings Su ey da a o Eu ope be ween 2002 and 2018, he e o e,
e eals signi ican changes bo h in e ms o he occupa ional s uc u e and wage inequali y ac oss
di e en Eu opean coun ies, in line wi h p e ious li e a u e. Fo bo h he U.S. and Eu ope, some
au ho s ha e a gued ha echnology-d i en shi s in labou demand a e behind he obse ed
pa e ns o occupa ional change (Au o e al., 2003; Goos and Manning, 2007; Goos e al., 2014).
Howe e , i is no clea ha hese demand shi s ha e had an impac on wages o wage inequali y
(Böhm e al., 2024). On he one hand, occupa ional wage g ow h has been decoupled om
occupa ional employmen g ow h (Mishel e al., 2013; Roys and Tabe , 2019; Hsieh e al., 2019). On
he o he hand, Au o (2019), Ca d e al. (2013) and Fi po e al. (2011) ha e a gued ha changes in
labou demand ac oss occupa ions had a small signi icance when explaining he inc ease in wage
inequali y o ecen yea s.
In his espec , i is s iking ha he clea es cases o g ow h and educ ion o wage inequali y in ou
analysis, Finland and Romania espec i ely, show a simila end o job pola isa ion. This on i s own
sugges s ha he obse ed pa e ns o occupa ional change a e la gely un ela ed o wage inequali y
ends, since he same inpu (in his case job pola isa ion), can gene a e e y di e en ou pu s (bo h
an inc ease and a decline o wage inequali y). By con as , F ance and pa ially Poland expe imen ed
occupa ional upg ading and a decline in wage inequali y, which may ini ially sound mo e plausible
bu is also p oblema ic (because an expansion o high-paid jobs may educe wage inequali y
composi ionally, bu i may also expand he dis ance -o inequali y- be ween he a e age wages o
high-paid jobs and he es ). Hence, he compa ison o wage inequali y ends and occupa ional
changes al eady sugges s a weak o non-exis en ela ionship be ween hese wo phenomena.
Compa ing he e olu ion o wage inequali y indica o s and occupa ional s uc u e changes in he
Eu opean coun ies, some ques ions a ise abou he impo ance o be ween and wi hin-job wage
di e en ials, i hey ha e inc eased o dec eased o e ime, and i he e a e di e ences by coun y.
Table 3 summa ises he ends in he dispe sion o indi idual and job-a e age wages wi hin each
quin ile. F om an indi idual wo ke pe spec i e, Spain, Finland, Poland and Romania a e cha ac e ised
by a g ow h in he dispe sion o wages wi hin quin iles, ema kably in he las coun y. We migh
expec his endency o lead o a g ow h in wage inequali y due o a ise in wi hin-job di e ences. In
ac , a simila end is obse ed in he UK un il 2010, al hough he dispe sion wi hin each quin ile is
educed a e wa ds. Finally, F ance is a unique case since he dispe sion o indi idual wages has been
consis en ly educed in all quin iles du ing he pe iod analysed.
Rega ding be ween-job di e en ials, we de ec simila ends in all Eu opean coun ies s udied. In
his sense, he dispe sion o job-a e age wages wi hin each quin ile has con inuously inc eased om
2002 o 2018. Howe e , he e a e di e ences in he scale o his inc ease. In Spain and Finland,
be ween-job di e en ials inc eased pa icula ly in he medium and high paid quin iles. By con as ,
in F ance, Poland and he UK, his g ow h was concen a ed in middle-paid jobs. Las , Romania eco ds
he highes ise in hese di e ences, which a e mainly obse ed om 2002 o 2018 in low-paid jobs,
wi h an inc ease o 17.32 and 10.69 pe cen age poin s in he i s and second quin iles espec i ely.
These desc ip i e indings sugges ha bo h wi hin- and be ween-job wage di e en ials may ha e
con ibu ed o wage inequali y ends in Eu ope, al hough his con ibu ion was no s ong enough o
d i e a gene alised inc ease in wage inequali y, as p e iously discussed. In any case, o discuss mo e
p ecisely how occupa ional ends may ha e a ec ed wage inequali y in Eu ope, i is necessa y o

The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
13
ca y ou a a iance and Theil decomposi ion analysis, dis inguishing he pa icula ole ha be ween-
and wi hin-job di e en ials ha e had in his e olu ion.
Table 3: Indica o s o wi hin-quin ile wage inequali y (di e ences acco ding o indi idual and
be ween job a e age wages) by coun y, 2002-2018.
Yea
n
No. o
jobs
Be ween-indi idual di e ences
Change in
s anda d
de ia ion Q1
( e . 2002)
Change in
s anda d
de ia ion Q2
( e . 2002)
Change in
s anda d
de ia ion Q3
( e . 2002)
Change in
s anda d
de ia ion Q4
( e . 2002)
Change in
s anda d
de ia ion Q5
( e . 2002)
Spain
2002
127,614
187
-
-
-
-
-
2006
132,123
211
-0.04
0.21
0.20
0.00
-0.01
2010
134,403
453
0.08
0.31
0.76
0.35
0.48
2014
129,357
419
0.19
0.72
0.47
0.27
0.47
2018
132,744
450
0.13
0.47
0.69
0.60
0.96
F ance
2002
94,233
188
-
-
-
-
-
2006
67,012
195
-0.61
0.03
-0.62
-0.64
-0.24
2010
114,523
439
1.68
-0.28
-0.62
-0.23
-0.35
2014
141,169
426
-0.10
0.24
-0.45
-0.40
-0.07
2018
147,300
417
-0.45
-0.05
-0.59
-0.36
-0.11
Finland
2002
110,675
169
-
-
-
-
-
2006
154,212
193
0.06
0.19
0.25
0.25
0.44
2010
153,976
423
0.28
0.35
0.34
0.56
1.16
2014
146,897
410
0.04
0.40
0.30
0.64
1.07
2018
152,392
413
0.17
0.38
0.27
0.80
1.34
Poland
2002
355,732
205
-
-
-
-
-
2006
346,495
199
0.16
0.24
0.31
0.21
0.41
2010
347,044
463
0.50
0.44
0.56
0.76
0.78
2014
391,579
456
0.59
0.71
0.80
1.08
1.04
2018
426,895
461
0.69
0.88
0.96
1.22
1.00
Romania
2002
137,955
214
-
-
-
-
-
2006
141,042
212
1.24
0.98
0.99
0.86
1.10
2010
131,391
460
1.63
1.80
2.18
1.68
2.15
2014
138,816
443
2.08
2.42
2.70
2.29
2.30
2018
164,720
458
4.46
4.11
5.06
5.33
4.06
Uni ed
Kingdom
2002
71,748
212
-
-
-
-
-
2006
63,580
206
-0.15
0.08
0.09
0.14
0.02
2010
82,985
469
0.05
0.57
9.97
0.28
0.22
2014
80,196
463
-0.13
0.62
0.24
0.05
0.11
2018
-
-
-
-
-
-
-
(Con inues)
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
14
Table 3 (con inued)
Yea
n
No. o
jobs
Be ween-job di e ences
Change in
s anda d
de ia ion Q1
( e . 2002)
Change in
s anda d
de ia ion Q2
( e . 2002)
Change in
s anda d
de ia ion Q3
( e . 2002)
Change in
s anda d
de ia ion Q4
( e . 2002)
Change in
s anda d
de ia ion Q5
( e . 2002)
Spain
2002
127,614
187
-
-
-
-
-
2006
132,123
211
0.88
1.04
1.64
0.68
0.21
2010
134,403
453
-0.04
0.28
0.71
1.21
2.09
2014
129,357
419
3.70
1.46
2.81
2.77
2.06
2018
132,744
450
2.29
1.76
3.02
2.42
3.11
F ance
2002
94,233
188
-
-
-
-
-
2006
67,012
195
0.09
0.78
0.46
-0.32
-0.23
2010
114,523
439
-0.24
-0.26
0.15
-0.32
0.03
2014
141,169
426
0.36
1.46
1.15
0.08
0.25
2018
147,300
417
0.21
1.69
2.39
0.47
0.59
Finland
2002
110,675
169
-
-
-
-
-
2006
154,212
193
0.23
1.18
0.92
1.20
0.58
2010
153,976
423
0.45
0.98
0.73
1.06
3.15
2014
146,897
410
1.10
4.71
4.00
2.12
3.94
2018
152,392
413
1.71
3.29
3.31
7.72
4.63
Poland
2002
355,732
205
-
-
-
-
-
2006
346,495
199
2.46
1.61
1.40
0.59
0.48
2010
347,044
463
1.21
3.95
0.50
0.67
2.07
2014
391,579
456
1.87
2.59
2.38
1.81
2.56
2018
426,895
461
2.91
4.10
4.19
2.17
2.53
Romania
2002
137,955
214
-
-
-
-
-
2006
141,042
212
1.03
3.77
2.72
1.76
1.24
2010
131,391
460
2.32
2.23
1.85
1.78
0.83
2014
138,816
443
4.13
5.09
5.43
4.74
1.08
2018
164,720
458
17.32
10.69
6.83
8.32
2.06
Uni ed
Kingdom
2002
71,748
212
-
-
-
-
-
2006
63,580
206
0.63
1.65
1.89
0.63
-0.25
2010
82,985
469
0.08
0.52
2.93
0.14
1.21
2014
80,196
463
0.80
1.08
4.86
0.91
1.54
2018
-
-
-
-
-
-
-
Sou ce: Au ho ’s es ima ions based on SES da a.
4.2 Wage inequali y be ween- and wi hin-jobs
Table 4 p esen s he pe cen age o a iance in log wages explained by be ween-g oup di e en ials
and decomposes he Theil index. The o al a iance o wages can be spli in o wo ac o s when he
da a is g ouped by jobs: he a iance be ween- and wi hin-job. Acco ding o his app oach, be ween-
job di e ences accoun o a ound 35-60 pe cen o o al a iance in log wages in he Eu opean
con ex (Table 4). Simila ly, Fe nández-Macías and A anz-Muñoz (2020) and Eu o ound (2017) ound
ha be ween-job inequali ies explain a ound 25-40% and 40-50% o his a iance espec i ely, using
di e en da a sou ces. Howe e , signi ican di e ences can be obse ed by coun y. In Finland and
he UK, he a iance accoun ed o by jobs is compa a i ely la ge, eaching alues o up o 60%. By
con as , in Poland and Romania, jobs explain a ound 45-50 pe cen o he a iance o log wages,
while in Spain and F ance he in e ac ion o occupa ions and sec o s only explain a 35-40%. In all
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
15
coun ies he mos impo an componen o he dis ibu ion o log wages using a jobs’ app oach is
ac ually occupa ion as measu ed by he ISCO classi ica ion, which explains mos o he a iance
shown by he ull ange o jobs (up o 30%). None heless, sec o s also add signi ican ly o he
explana ion o log wages’ a iance, as well as i s combina ion wi h occupa ions, especially in Finland
and he UK (a ound 20-25%). Fo e e ence, able 4 also shows a column wi h he pe cen age o
a iance o log wages explained by a model wi h he usual socio-demog aphic a iables: we can see
ha he model using jobs (ISCO x NACE) can explain in all cases a signi ican ly la ge pe cen age o
a iance han his e e ence socio-demog aphic model.
Table 4: ANOVA decomposi ions (pe cen age o a iance in log wages explained by be ween-g oups
di e en ials) and inequali y indices decomposi ions (wages no logged) by coun y, 2002-2018.
Yea
n
No. o
jobs
% o a iance in log wages explained by be ween-g oups di e en ials
ISCO only
(39
ca ego ies)
NACE
only (18
ca ego ies)
ISCO +
NACE, no
in e ac ion
Jobs
(ISCO x
NACE)
Jobs,
wages no
logged
Jobs,
wages <
op 1%
Spain
2002
127,614
187
19.75
13.86
28.75
39.9
31.97
36.03
2006
132,123
211
16.32
14.12
26.82
38.37
32.6
35.67
2010
134,403
453
17.81
17.17
30.02
45.38
33.92
42.29
2014
129,357
419
16.83
15.4
28.1
42.15
33.48
38.9
2018
132,744
450
16.57
15.59
27.74
41.07
32.26
38.45
F ance
2002
94,233
188
31.12
3.73
34.72
46.56
10.7
43.39
2006
67,012
195
25.32
4.1
28.41
36.85
10.29
31.11
2010
114,523
439
25.01
6.17
28.12
39.04
12.16
36.36
2014
141,169
426
24.19
6.54
26.34
35.63
11.39
28.02
2018
147,300
417
23.11
4.56
24.68
33.41
13.28
25.03
Finland
2002
110,675
169
18.59
6.21
27.22
49.76
46.98
48.16
2006
154,212
193
21.36
8.21
32.3
53.39
47.67
51.87
2010
153,976
423
23.59
17.38
35.55
55.81
50.99
54.12
2014
146,897
410
25.53
21.04
39.63
60.48
55.21
55.86
2018
152,392
413
32.44
21.72
43.24
60.41
54.1
56.34
Poland
2002
355,732
205
18.67
11.65
29.23
45.24
29.91
40.52
2006
346,495
199
18.76
10.95
29.37
44.92
27.66
38.97
2010
347,044
463
21.4
19.12
35.36
49.51
36.77
43.52
2014
391,579
456
20.87
15.27
32.85
48.18
35.35
42.04
2018
426,895
461
22.09
15.8
32.18
46.91
36.95
41.75
Romania
2002
137,955
214
21.86
7.51
27.94
39.98
33.15
37.02
2006
141,042
212
26.43
7.93
31.55
43.77
32.41
41.6
2010
131,391
460
32.07
15.34
40.07
50.48
37.1
45.49
2014
138,816
443
31.55
14.82
39.38
50.29
37.12
45.44
2018
164,720
458
31.54
13.19
36.89
49.12
35.53
44.51
Uni ed
Kingdom
2002
71,748
212
23.74
13.55
35.81
52.39
29.28
45.24
2006
63,580
206
24.47
19.49
41
57.64
34.2
49.1
2010
82,985
469
25.28
25.88
41.06
56.49
8.48
47.4
2014
80,196
463
28.1
25.51
43.25
60.52
45.46
53.77
2018
-
-
-
-
-
-
-
-
(Con inues)
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
16
Table 4 (con inued)
Yea
n
No. o
jobs
% o a iance
explained by a
model wi h socio-
demog aphic
cha ac e is ics
Inequali y indices (wages no logged)
Gini
Theil
Theil
be ween
jobs
Theil
wi hin
jobs
Be ween
jobs/To al
Theil
Spain
2002
127,614
187
34.49
31.35
17.94
7.31
10.63
40.75
2006
132,123
211
34.05
28.29
14.55
5.77
8.78
39.66
2010
134,403
453
35.11
30.11
17.17
7.83
9.34
45.60
2014
129,357
419
32.18
29.57
16.16
6.94
9.22
42.95
2018
132,744
450
32.47
30.82
18.39
8.04
10.35
43.72
F ance
2002
94,233
188
34.79
35.34
28.04
9.85
18.19
35.13
2006
67,012
195
34.5
28.2
16.81
4.79
12.02
28.49
2010
114,523
439
30.39
27.15
16.29
5.62
10.67
34.50
2014
141,169
426
31.04
30.66
20.25
5.83
14.42
28.79
2018
147,300
417
30.43
31.41
21.58
5.93
15.65
27.48
Finland
2002
110,675
169
32.44
18.42
5.93
3
2.93
50.59
2006
154,212
193
35.45
20.26
7.42
3.99
3.42
53.77
2010
153,976
423
32.12
22.2
9.38
5.44
3.94
58.00
2014
146,897
410
39.71
21.54
8.45
5.17
3.28
61.18
2018
152,392
413
39.56
22.27
9.21
5.64
3.57
61.24
Poland
2002
355,732
205
29.13
32.62
20.65
9.23
11.42
44.70
2006
346,495
199
26.47
34.08
23.51
10.48
13.02
44.58
2010
347,044
463
24.34
33.09
22.16
11.52
10.65
51.99
2014
391,579
456
29.23
33.71
22.75
11.52
11.23
50.64
2018
426,895
461
29.54
31.51
19.79
9.86
9.93
49.82
Romania
2002
137,955
214
29.12
39.08
31.39
14.71
16.68
46.86
2006
141,042
212
31.14
39.65
31.51
15.44
16.07
49.00
2010
131,391
460
32.82
38.87
29.54
15.65
13.88
52.98
2014
138,816
443
35.43
38.21
28.74
15.07
13.67
52.44
2018
164,720
458
34.72
34.53
23.66
12.04
11.62
50.89
Uni ed
Kingdom
2002
71,748
212
28.88
34.59
23.53
11.33
12.19
48.15
2006
63,580
206
31.11
34.84
23.27
12.22
11.06
52.51
2010
82,985
469
24.69
35.56
26.76
12.87
13.9
48.09
2014
80,196
463
24.69
33.27
20.9
12.33
8.57
59.00
2018
-
-
-
-
-
-
-
-
Sou ce: Au ho ’s es ima ions based on SES da a.
I is c ucial o keep in mind ha wages a e known o p esen a log-no mal dis ibu ion, s ongly
asymme ic and igh -side skewed due o a high concen a ion o obse a ions below he mean, which
is usually swelled by la ge ou lie s. We sol ed his p oblem by ans o ming wages in o loga i hms;
howe e , his makes he dis ibu ion less unequal and can p oduce misleading esul s o ou
pu poses. To e alua e his e ec , we ha e also analysed he ole o jobs when explaining he a iance
o wages using nominal wages (i.e., no logged), as well as all nominal wages excep he op 1% o
hei dis ibu ion. In he i s scena io, jobs lose pa o hei explana o y powe , especially in F ance
whe e he pe cen age o wages’ a iance explained alls down conside ably. This means ha he e
a e e y la ge alues o wages whose occu ence canno be linked o occupa ional di e ences. In he
second si ua ion, when no including he op 1% o nominal wages’ dis ibu ion, he a iance
explained by jobs is simila o ha o logged wages. The e o e, jobs play a signi ican ole in he
dis ibu ion o mos wages, bu a ma ginal one in he dis ibu ion o e y la ge wages. In his sense,
using nominal wages o he analysis clea ly inc eases wi hin-job inequali y.
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
17
Ou analysis no only e eals di e ences in he ole o occupa ions when explaining he a iance o
wages by coun y, bu also di e en e olu ions ac oss Eu opean coun ies. Figu e 2 cap u es his
e olu ion o bo h occupa ions and jobs by coun y om 2002 o 2018. As a o emen ioned, he e a e
wo pe iods s ic ly compa able ha a e 2002-2006 and 2010-2018. Fo his eason, we ha e added
a e ical line o ma k he b eaks in ISCO and NACE classi ica ions. In igu e 2 we can dis inguish wo
clea ends among Eu opean coun ies in he ecen decades. On he one hand, in Spain, F ance and
Poland, he a iance explained by jobs clea ly and signi ican ly dec eased o e he pe iod. This end
is also obse ed in Romania a e he G ea Recession, al hough a mild inc ease is seen be o e i . On
he o he hand, Finland and he UK show a clea and consis en g ow h in he sha e o log wage’s
a iance explained by jobs.
Figu e 2: ANOVA decomposi ions by coun y on he pe cen age o a iance in log wages explained
by occupa ion and job, 2002-2018.
Sou ce: Au ho ’s es ima ions based on SES da a.
These indings sugges a dec easing ole o jobs in a bulk g oup o Eu opean coun ies om 2002 o
2018. Howe e , he e a e also some coun ies wi h an opposi e end. This con as s wi h he esul s
o Fe nández-Macías and A anz-Muñoz (2020), who, making use o he EU-SILC, ound an inc easing
o s able ole o jobs using he same app oach. I should be no ed ha hese au ho s ound he
inc easing end pa icula ly a ound he pe iod o he inancial economic c isis o 2008. Wi h ou da a,
we see his inc ease om 2006 o 2010, al hough we canno ensu e ha his g ow h is no a
consequence, o a leas in pa , o he changes in he ISCO and NACE classi ica ions. Bu mos
0.1
0.2
0.3
0.4
0.5
0.6
0.7
2002 2006 2010 2014 2018
Va iance's % explained
Spain
Ocuppa ion Job
0.1
0.2
0.3
0.4
0.5
0.6
0.7
2002 2006 2010 2014 2018
Va iance's % explained
F ance
Ocuppa ion Job
0.1
0.2
0.3
0.4
0.5
0.6
0.7
2002 2006 2010 2014 2018
Va iance's % explained
Finland
Ocuppa ion Job
0.1
0.2
0.3
0.4
0.5
0.6
0.7
2002 2006 2010 2014 2018
Va iance's % explained
Poland
Ocuppa ion Job
0.1
0.2
0.3
0.4
0.5
0.6
0.7
2002 2006 2010 2014 2018
Va iance's % explained
Romania
Ocuppa ion Job
0.1
0.2
0.3
0.4
0.5
0.6
0.7
2002 2006 2010 2014 2018
Va iance's % explained
Uni ed Kingdom
Ocuppa ion Job

The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
18
impo an ly, he longe pe iod co e ed, and he mo e p ecise da a used he e, e eal a mo e consis en
and s able e olu ion o he ole o jobs on wage inequali y in all EU coun ies analysed. And in any
case, i should be no ed ha he esul s o Finland and F ance a e consis en wi h hose o
Fe nández-Macías and A anz (2020): only in Spain we ind a di e en pa e n when using SES (as
we do he e) and SILC (as done by Fe nández-Macías and A anz, 2020), and gi en he cha ac e is ics
o bo h su eys we conside he esul s om SES o be mo e eliable. In e ms o he powe o
explana ion o occupa ion and job, we ound a small gap in F ance. Being job he in e sec ion o
occupa ion and sec o , his esul e eals ha sec o s play li le ole in explaining wage inequali ies
in F ance, and wage inequali y he e o e lies in he side o occupa ion and o he ac o s. This e idence
is also suppo ed by he esul s o he ANOVA analysis shown in Table 4. On he con a y, Finland has
a high gap in he powe o explana ion be ween occupa ion and jobs, and hence a highe ole o he
sec o s’ capaci y (see Figu e 2).
Figu e 3: Theil index decomposi ion by be ween and wi hin job di e ences and pe cen age o
o e all Theil explained by be ween job di e en ials by coun y, 2002-2018.
(1) Di e en scales ha e been applied o bo h axes in each coun y in o de o accu a ely ep esen he
e olu ion o Theil componen s.
Sou ce: Au ho ’s es ima ions based on SES da a.
In sho , we obse e wo main dynamics in Eu ope: on he one hand, he p edominan pa e n is one
o dec easing impo ance o occupa ions o explain o e all wage inequali y; on he o he hand, we
see an inc easing impo ance o occupa ions in Finland and he UK. Bu om his we canno conclude
0
10
20
30
40
50
60
70
0
10
20
30
40
2002 2006 2010 2014 2018
% Be ween/ o al
Theil index alue
Spain
Theil wi hin jobs Theil be ween jobs
Be ween jobs/To al Theil
0
10
20
30
40
50
60
70
0
10
20
30
40
2002 2006 2010 2014 2018
% Be ween/ o al
Theil index alue
F ance
Theil wi hin jobs Theil be ween jobs
Be ween jobs/To al Theil
0
10
20
30
40
50
60
70
0
10
20
30
40
2002 2006 2010 2014 2018
% Be ween/ o al
Theil index alue
Finland
Theil wi hin jobs Theil be ween jobs
Be ween jobs/To al Theil
0
10
20
30
40
50
60
70
0
10
20
30
40
2002 2006 2010 2014 2018
% Be ween/ o al
Theil index alue
Poland
Theil wi hin jobs Theil be ween jobs
Be ween jobs/To al Theil
0
10
20
30
40
50
60
70
0
10
20
30
40
2002 2006 2010 2014 2018
% Be ween/ o al
Theil index alue
Romania
Theil wi hin jobs Theil be ween jobs
Be ween jobs/To al Theil
0
10
20
30
40
50
60
70
0
10
20
30
40
2002 2006 2010 2014
% Be ween/ o al
Theil index alue
Uni ed Kingdom
Theil wi hin jobs Theil be ween jobs
Be ween jobs/To al Theil
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
19
ye whe he occupa ional change con ibu ed o no o he o e all ends in wage inequali y. In o de
o s udy his ela ionship, we ha e pe o med a Theil decomposi ion o be ween- and wi hin-job
di e en ials. Figu e 3 decomposes he e olu ion o Theil, as a measu e o he deg ee o inequali y in
he dis ibu ion o wages, in o he be ween and wi hin componen s o he analysed pe iod. The sum
o hese wo ac o s e lec s he o e all le el o inequali y in each coun y. We also p esen he sha e
o he o al Theil index ha co esponds o be ween-job di e ences, which allows us o in e p e i
jobs ha e gained mo e ele ance o no in explaining wage inequali y o e hese las decades ( his
measu e is equi alen o he a iance decomposi ion app oach discussed ea lie ).
In acco dance wi h he p e ious indings on he a iance analysis, bu wi h some pa icula i ies, he e
a e h ee pa e ns in he sha e o Theil explained by be ween-job di e en ials. Fi s , he e is again
an inc easing e olu ion om 2002 o 2018 in Finland, being he only coun y wi h a clea inc ease
o wage inequali y o he pe iod analysed which is plainly linked o an expansion o wage inequali y
be ween occupa ions. Wage di e en ials be ween occupa ions g ew s ongly du ing he las wo
decades and explain mos o he global Theil inc ease. In he Finnish case, i is he e o e e y plausible
ha he ecen ise in wage inequali y has been a leas pa ly d i en by occupa ional wage
di e en ials.
Secondly, al hough he wage a iance explained by jobs g ew signi ican ly in he UK and dec eased
in Romania and F ance, igu e 3 shows ha changes in Theil a e in ac mo e associa ed o a ia ions
in he wi hin componen . This is because in his g oup o coun ies, he e olu ion o he be ween
componen has been mo e s able. In Romania and he UK he e has been a educ ion o wage
inequali y, which has been mo e in ense in he wi hin componen han in he be ween one. In F ance,
he e has been a all in o e all inequali y as cap u ed by he Theil index, which is obse ed in bo h
componen s, ollowed by a sligh bu consis en g ow h ha has been mainly d i en by he wi hin
componen . Howe e , and despi e his g ow h, o e all wage inequali y did no each he alues ound
a he beginning o he pe iod. This pa ially alida es he esul s o Williams (2013) on he inc easing
ole o jobs when explaining wage inequali y in he UK. Howe e , he de elopmen o be ween and
wi hin componen s sugges ha i is he wi hin componen he one d i ing ecen changes in wage
inequali y, which is in line wi h he indings o F ance o Godecho (2012) and was no e lec ed in
he analysis o Williams (2013).
By con as , in Spain and Poland, i is unclea i wage inequali y changes a e d i en by he wi hin o
he be ween componen . Compa ing ou esul s wi h hose o Fe nández-Macías and A anz-Muñoz
(2020), we ind a much mo e s able e olu ion o bo h componen s.
10
I seems ha bo h wi hin and
be ween componen s ha e e ol ed simila ly o o e all wage inequali y, which has a ied li le.
Howe e , e en in hese coun ies he be ween componen has been mo e s able, while he wi hin
componen has changed mo e. In Spain, he global end o wage inequali y has been owa ds a small
inc ease, while in Poland he opposi e has been obse ed. Howe e , hese changes appea o be
inconsis en and non-signi ican .
Based on app oach by Mookhe jee and Sho ocks (1982), able 5 shows he con ibu ion o wage
inequali y ends o changes in wi hin-occupa ion inequali y, popula ion sha es be ween- and wi hin-
occupa ion and he mean wages o occupa ions. A i s look in o he esul s e eals ha he e a e
e y di e en ends ac oss coun ies and also be ween pe iods in each one o hem. Ne e heless,
in gene al e ms, his app oach suppo s mo e clea ly he p e ious obse a ion o a dec easing end
o wage inequali y in mos coun ies being mos ly d i en by di e en ials wi hin-occupa ion.
Addi ionally, changes in he mean wages o occupa ions also end o con ibu e nega i ely o wage
inequali y, educing i al hough in a so e way. The main excep ion is again Finland, whe e he
10
As we ha e al eady no ed, EU-SILC's in o ma ion o wages is p o ided as mon hly wages o p e ious yea and he
numbe o hou s wo ked pe week is no a ailable. The e o e, hou ly wages may no cap u e changes in he indi iduals’
occupa ion and employmen s a us a ec ing he income luc ua ion componen . Some adjus men s can be done; howe e ,
any adjus men inc eases he s a is ical noise, which is especially impo an when measu ing a iance, in his pa icula
case o wages.
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
20
opposi e in luence o changes in mean wages o occupa ions is ound du ing he pe iods 2002-2006
and 2010-2018.
On he con a y, he e is a posi i e e ec on wage inequali y ends o changes in he popula ion
sha es o bo h be ween- and wi hin-occupa ion componen s. Despi e he posi i e con ibu ion o
hese “composi ion e ec s”, hey a e mo e ola ile and smalle han he ones ela ed o he o he
wo componen s. Acco ding o he p e ious li e a u e, changes in he popula ion sha es be ween-
occupa ion componen a e he ones ha could e lec he impac o occupa ional changes on wage
inequali y ends. Bo h pola isa ion and upg ading p ocesses should lead o an inc ease in his
componen . The es ima ions ca ied ou con i m ha he “composi ion e ec ” ends o posi i ely
a ec (expand) wage inequali y, as we could expec acco ding o he pola isa ion and upg ading
ound in some o he Eu opean coun ies analysed. Howe e , in gene al, his e ec did no d i e wage
inequali y ends because he e ec s o he o he componen s (especially he wi hin-occupa ion and
mean wages o occupa ions) compensa e o i o e en clea ly domina e i . Tha said, we ind again
an excep ion in Finland, whe e wage inequali y inc eased mainly due o changes in he popula ion
sha es be ween-occupa ion componen , o , in o he wo ds, due o occupa ional changes, which
suppo s he indings o he Theil decomposi ion app oach.
Table 5: Decomposi ion o he end in agg ega e inequali y by coun y: Theil index Io (x103).
Yea
Con ibu ion o ∆𝑰 due o changes in
Change in
agg ega e
inequali y
Wi hin
occupa ion
inequali y
Popula ion sha es
Mean occupa ion
incomes
∆𝑰
∑𝒗𝒌∆𝑰𝟎
𝒌
∑𝑰𝟎
𝒌∆𝒗𝒌
∑(𝜸𝒌−𝐥𝐧𝜸𝒌)∆𝒗𝒌
∑(𝜽𝒌−𝒗𝒌)∆𝐥𝐧𝝁𝒌
Spain
2002-06
-30.7
-14.9
0.2
3.3
-19.3
2010-14
-6.0
-1.6
0.8
-0.3
-4.9
2014-18
11.5
5.1
0.4
-0.4
6.5
F ance
2002-06
-100.9
-54.9
1.3
-8.6
-38.7
2010-14
39.8
17.8
15.1
9.1
-2.1
2014-18
9.2
-3.9
11.7
-8.4
9.7
Finland
2002-06
10.5
2.2
1.3
5.1
1.9
2010-14
-6.8
-5.1
-0.9
-0.5
-0.2
2014-18
4.6
2.1
0.1
0.9
1.6
Poland
2002-06
27.3
1.1
6.2
17.8
2.2
2010-14
5.7
-1.4
6.1
2.7
-1.7
2014-18
-24.4
-10.9
1.0
0.6
-15.0
Romania
2002-06
6.9
-25.7
19.8
27.3
-14.5
2010-14
-7.3
-4.6
1.3
8.3
-12.3
2014-18
-44.5
-22.1
5.1
5.8
-33.3
Uni ed
Kingdom
2002-06
-0.7
-13.2
2.4
11.8
-1.7
2010-14
-41.4
-29.1
-4.3
-1.7
-6.3
2014-18
-
-
-
-
-
(1) equa ions’ no a ion has been de ined in he Appendix.
Sou ce: Au ho ’s es ima ions based on SES da a.
To summa ise, using a mo e p ecise da ase on wages and a wide ime ame han p e ious s udies,
we ound ha , in bo h pe iods be o e and a e he G ea Recession, Eu opean coun ies p esen
di e en pa e ns o associa ion be ween occupa ional change and ends in wage inequali y. The
mos common pa e n is one in which occupa ional change, which di e s conside ably ac oss
coun ies, had a posi i e bu smalle e ec han wi hin-job wage di e en ials on wage inequali y
endencies du ing his pe iod, pa icula ly in some coun ies such as F ance and he UK. Howe e , he
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
21
opposi e pa e n is clea ly ound in one o he coun ies analysed, Finland. Consequen ly, we canno
ejec he possibili y ha in ce ain cases, occupa ional change may ha e had an impo an e ec on
he e olu ion o wage inequali y. Wha we can ejec is ha , in gene al, occupa ional dynamics, due
o changes in he occupa ional s uc u e, explain ecen wage inequali y ends. Ins ead, ou indings
show ha ecen wage inequali y ends we e mainly d i en by changes in wi hin-occupa ion
inequali y and in he mean wages o occupa ions. Fe nández-Macías and A anz-Muñoz (2020)
sugges ha he wi hin componen may be mo e p one o d i e wage inequali y in he sho un
because i is mo e sensi i e o cyclical e ec s. Ou longe - e m analysis sugges s ha he e is no a
uni o m associa ion be ween occupa ional change and ends in wage inequali y in Eu ope, bu
ins ead he e a e some ( ew) coun ies in which occupa ional di e en ials became mo e impo an
and con ibu ed o g owing wage inequali y, and (mo e commonly) o he s in which hey became less
impo an and hei posi i e e ec on wage inequali y ends we e eclipsed by he nega i e end on
wi hin-job wage di e en ials.
5 Conclusions
The objec i e o his wo k was o analyse whe he occupa ional change has d i en wage inequali y
in Eu ope du ing ecen decades. To ha end, a a iance decomposi ion analysis, a Theil
decomposi ion in o be ween and wi hin componen s, and a Mookhe jee and Sho ocks (1982)
decomposi ion o he ends in wage inequali y app oach ha e been es ima ed ac oss six Eu opean
coun ies h ough he use o da a om he EU S uc u e o Ea nings Su ey. Conside ing he pe iod
2002-2018, he analysis conside ed he changes in he occupa ional s uc u e, he e olu ion o main
inequali y indices and he a ia ions in he ela i e weigh o bo h wi hin- and be ween-job
di e en ials in explaining wage inequali y. Based on his analysis, we p esen he ollowing
conclusions ega ding he ole o jobs, unde s ood as posi ions wi hin he di ision o labou along he
e ical and ho izon al dimensions (occupa ion and sec o ), in wage inequali y ends in Eu ope.
Fi s , he Gini and Theil indices show a dec ease o s agna ion o wage inequali y in all Eu opean
coun ies wi h he excep ion o Finland in he pe iod analysed. The Theil decomposi ion esul s e eal
ha he e a e di e en pa e ns in he sha e o wage inequali y explained by be ween-occupa ion
di e ences. While he ole o occupa ions g ew signi ican ly om 2002 o 2018 in Finland and he
UK, hei ela i e weigh dec eased in Spain, F ance, Poland and Romania. These indings con i m ha
he e is no unique end in e ms o he ole o occupa ions in he de elopmen o wage inequali y in
Eu ope, bu , on he con a y, hey ha e gained mo e ele ance in explaining wage inequali y o e he
las wo decades in he i s g oup o coun ies and diminished i in he second g oup.
Secondly, he e olu ion o bo h wi hin and be ween componen s sugges s ha occupa ional change,
which di e s conside ably ac oss EU coun ies, had a posi i e e ec on wage inequali y ends.
Howe e , he nega i e e ec o changes in wi hin-occupa ion di e en ials and he mean wages o
occupa ions we e mo e subs an ial, and mo e ele an o explaining ecen wage inequali y ends.
In pa icula , occupa ional shi s con ibu ed li le o wage inequali y ends in F ance, he UK and
Romania in he pe iod analysed, whe eas wi hin-occupa ion di e en ials played a much mo e
impo an ole. Howe e , in Finland, he e was a clea expansion o wage inequali y which was clea ly
linked o changes in he occupa ional s uc u e, in his case linked o a p ocess o job pola isa ion
acco ding o ou analysis. Mo eo e , in Spain and Poland, we canno a i m i wage inequali y changes
we e d i en by he wi hin o he be ween componen since bo h elemen s’ e olu ion was s able be o e
and a e he G ea Recession.
Thi dly, be ween-occupa ion di e en ials accoun o a ound 35-60 pe cen o o al wages a iance,
unde s anding occupa ions as he combina ion o e ical and ho izon al dimensions o labou
di ision. Howe e , signi ican di e ences a e ound by coun y. While in Spain and F ance occupa ions
explain abou a 35-40% o his a iance, in Poland and Romania, hey explain up o a 45-50%.
Mo eo e , in Finland and he UK, he wage a iance accoun ed o by occupa ions eaches alues o
up o 60%. In all coun ies he mos impo an componen o he dis ibu ion o wages using he
The ole o be ween- and wi hin-occupa ion di e ences in wage inequali y ends in Eu ope (2002-2018)
28
Appendix
Following he decomposi ion o he end in wage inequali y by g oups p oposed by Mookhe jee and
Sho ocks (1982), we can decompose he change in inequali y measu ed as he Theil coe icien 𝐼0
by occupa ions, as ollows,
∆𝐼0=∑𝑣𝑘
𝑘∆𝐼0
𝑘+∑𝐼0
𝑘
𝑘∆𝑣𝑘−∑ln𝛾𝑘






𝑘∆𝑣𝑘−∑𝑣𝑘∆ln𝛾𝑘
𝑘
wi h,
𝛾𝑘=𝜇𝑘𝜇
⁄
𝜇 =∑𝑣𝑘𝜇𝑘𝑘
𝑘 =1,…,𝑛
whe e, 𝑣𝑘 ep esen s he popula ion sha e and 𝜇𝑘 he mean wage, o each occupa ion k o a o al o
n occupa ions. The e o e, he change in inequali y measu ed wi h he Theil index can be decomposed
in o ou componen s: (i) he impac o in e empo al changes in wi hin-occupa ion inequali y, (ii) he
e ec o changes in he popula ion sha e o he wi hin-occupa ion componen , (iii) he e ec o
changes in he popula ion sha e o he be ween-occupa ion componen , and (i ) he con ibu ion o
inequali y changes a ibu able o ela i e changes in he mean wages o occupa ions. These ou
componen s can be calcula ed as:
(i): ∑𝑣𝑘𝑘 ∆𝐼0
𝑘
(ii): ∑𝐼0
𝑘
𝑘∆𝑣𝑘
(iii): ∑(𝛾𝑘−ln𝛾𝑘






)∆𝑣𝑘𝑘
(i ): ∑(𝜃𝑘−𝑣𝑘)∆ln𝜇𝑘𝑘
wi h,
𝜃𝑘= 𝑣𝑘𝛾𝑘
Acco dingly, ∆𝐼0 and he change in hese ou componen s can be calcula ed be ween any wo yea s
o each coun y. As well as wi h he be ween- and wi hin-componen s o he Theil index, we can
i mly compa e only wo sepa a e pe iods due o he changes in bo h ISCO and NACE classi ica ions
(2002-2006 and 2010-2018).

GETTING IN TOUCH WITH THE EU
In pe son
All o e he Eu opean Union he e a e hund eds o Eu ope Di ec cen es. You can ind he add ess o he cen e nea es you online
(eu opean-union.eu opa.eu/con ac -eu/mee -us_en).
On he phone o in w i ing
Eu ope Di ec is a se ice ha answe s you ques ions abou he Eu opean Union. You can con ac his se ice:
— by eephone: 00 800 6 7 8 9 10 11 (ce ain ope a o s may cha ge o hese calls),
— a he ollowing s anda d numbe : +32 22999696,
— ia he ollowing o m: eu opean-union.eu opa.eu/con ac -eu/w i e-us_en.
FINDING INFORMATION ABOUT THE EU
Online
In o ma ion abou he Eu opean Union in all he o icial languages o he EU is a ailable on he Eu opa websi e (eu opean-
union.eu opa.eu).
EU publica ions
You can iew o o de EU publica ions a op.eu opa.eu/en/publica ions. Mul iple copies o ee publica ions can be ob ained by
con ac ing Eu ope Di ec o you local documen a ion cen e (eu opean-union.eu opa.eu/con ac -eu/mee -us_en).
EU law and ela ed documen s
Fo access o legal in o ma ion om he EU, including all EU law since 1951 in all he o icial language e sions, go o EUR-Lex
(eu -lex.eu opa.eu).
Open da a om he EU
The po al da a.eu opa.eu p o ides access o open da ase s om he EU ins i u ions, bodies and agencies. These can be
downloaded and eused o ee, o bo h comme cial and non-comme cial pu poses. The po al also p o ides access o a weal h
o da ase s om Eu opean coun ies.
2