Does ICT Usage E ode Rou ine Occupa ions a he
Fi m Le el?
Pe i B€
ocke man
1
— Seppo Laaksonen
2
— Ja i Vainiom€
aki
3
Abs ac . We show ha disappea ing ou ine wo k can be linked o i m-le el ICT usage. Ou
esul s o he inc easing abs ac and declining ou ine occupa ion sha es o o al wage bill a e
consis en wi h job pola iza ion a he i m le el. The obse ed changes coincide wi h he usage o
ICT in i ms. These esul s a e based on decomposi ions and eg ession analyses ha e alua e he
hollowing-ou o ou ine occupa ions and job pola iza ion a he i m le el.
1. In oduc ion
Occupa ional s uc u es in Eu ope a e in u moil (Fe n
andez-Mac
ıas, 2012). This pape
examines he ou iniza ion hypo hesis and occupa ional pola iza ion using ich i m-le el
da a. We con ibu e o he ela i ely hin bu g owing empi ical li e a u e ha has used
i m-le el da a o explo e he sou ces o pola iza ion (Ake man e al., 2015; Ba el e al.,
2007; Co es and Sal a o i, 2015; C ino, 2010; Gaggl and W igh , 2017; Ha igan e al.,
2016). The use o i m-le el da a is a na u al ex ension o he li e a u e, which has ocused
on changes a he agg ega e, indus y, o local labou ma ke le els.
1
We analyse changes in he occupa ional s uc u e o labou demand a he i m le el.
This allows us o examine he e ogenei y in job pola iza ion and he echnology explana ion
o ou iniza ion a he mic o le el, whe e ac ual labou demand decisions a e made. The
li e a u e has documen ed a la ge amoun o he e ogenei y be ween i ms in e ms o p o-
duc i i y and wages (Ca d e al., 2013; Sy e son, 2011).
We e alua e an impo an aspec o ou ine-biased echnical change, i.e., employmen
pola iza ion. The esul s show ha he wage bill sha es o (non- ou ine) abs ac and se -
ice occupa ions ha e inc eased, whe eas he sha e o ou ine occupa ions ha e dec eased.
The agg ega e pola iza ion o he employmen dis ibu ion is decomposed in o be ween-
and wi hin- i m e ec s, in addi ion o he in luence o en y and exi . The changes o
abs ac and ou ine occupa ions a e d i en by wi hin- i m employmen ealloca ion, bu
he changes o manual/se ice occupa ions e lec mos ly ealloca ion o employmen
Decla a ions: The da a used in his s udy a e con iden ial bu o he esea che s can independen ly
ob ain access o he da a o eplica ion pu poses wi h he pe mission o S a is ics Finland. The
au ho s will p o ide guidance abou acqui ing he da a upon eques . The au ho s decla e ha hey
ha e no ele an o ma e ial inancial in e es s ha ela e o he esea ch desc ibed in his pape .
1
Labou Ins i u e o Economic Resea ch and IZA, Uni e si y o Jy €
askyl€
a, Jy €
askyl€
a, Finland.
2
Uni e si y o Helsinki, Helsinki, Finland.
3
School o Managemen , Uni e si y o Tampe e, 33014 Tampe e, Finland.
E-mail: [email p o ec ed]
LABOUR 33 (1) 26–47 (2019) DOI: 10.1111/lab .12137
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d JEL J23, J24, J31, O33
This is an open access a icle unde he e ms o he C ea i e Commons A ibu ion-NonComme cial-NoDe i s License, which pe mi s use
and dis ibu ion in any medium, p o ided he o iginal wo k is p ope ly ci ed, he use is non-comme cial and no modi ica ions o
adap a ions a e made.
be ween i ms. The pa e n is consis en wi h he Au o and Do n (2013) model’s gene al
equilib ium explana ion o he ising se ice employmen .
2
Second, we es ima e models ha u ilize speci ic measu es o he usage o in o ma ion
and communica ion echnologies (ICT) a he i m le el. The aim is o iden i y whe he
ICT a a i m leads o dec easing sha es o ou ine occupa ions and inc easing sha es o
non- ou ine abs ac occupa ions. The use o comp ehensi e i m-le el ICT indica o s
allows us o es ablish a mo e di ec link be ween obse ed occupa ional o educa ional
changes and compu e -based echnological change han agg ega e s udies based on ime
e ec s o indus y-le el agg ega es. We concen a e on i m-le el eg essions o changes in
employmen sha es because ou decomposi ions show ha he changes in wage bill sha es
a e p edominan ly wi hin- i m phenomena. Ou esul s show ha he usage o ICT coin-
cides wi h he g ow h in abs ac occupa ion wage bills and he decline in ou ine occupa-
ion wage bills a he i m le el. We p o ide a pa allel ocus on educa ional and
occupa ional sha es in o de o conside he possible di e ences in he e ec s o ou iniza-
ion on he demand o skills and asks.
Michaels e al. (2014) used coun y-indus y panel da a and showed ha wage bill sha es
(and ela i e wages) o bo h high- and low-educa ion le els a e posi i ely ela ed o indus-
y ICT capi al, whe eas hose o he middle educa ion le els a e nega i ely ela ed o
ICT. In he p e ious i m-le el li e a u e Ba el e al. (2007) ocus on a na ow manu ac-
u ing sec o ; Ake man e al. (2015) use a e y speci ic ICT echnology (b oadband In e -
ne ); Co es and Sal a o i (2015) use compu e sha e and an indica o o adop ion o new
echnology om a managemen su ey; Gaggl and W igh (2017) use ICT in es men s
a he han de ailed ICT echnologies; Ha igan e al. (2016) use he sha e o ‘ echies’
(ICT- ela ed pe sonnel) as a p oxy o ICT se ices; and Pekkala Ke e al. (2016) use one
indica o o he sha e o employees using ICT om 1 yea . In con as o hese ei he na -
ow o e y b oad ICT measu es used in he con empo a y li e a u e, we u ilize de ailed
in o ma ion abou a b oad se o speci ic ICT echnologies used by i ms in all sec o s o
he economy, which mo e comp ehensi ely cap u es he pene a ion o new echnology in o
he o ganiza ion o asks wi hin i ms.
T adi ionally, he d i ing o ce behind he inc ease in wage di e en ials, educa ion
p emiums, and skill-upg ading in indus ialized coun ies has been seen o be skill-biased
echnological change (SBTC) (see, e.g., Acemoglu, 2002; Bound and Johnson, 1992;
Machin, 2008; o e iews). To add ess he obse ed anomalies ha a e di icul o explain
wi h he s anda d SBTC model (Acemoglu and Au o , 2011; Ca d and DiNa do, 2002),
Au o e al. (ALM) (2003) no ed he di e ences be ween he asks ha wo ke s do in hei
jobs and he skills hey use o pe o m hese asks in di e en occupa ions. Compu e s
eplace ou ine asks bu complemen non- ou ine analy ical and in e ac i e asks. Reduc-
ion in ou ine asks is dubbed he ou iniza ion hypo hesis (o ou ine-biased echnological
change).
3
Au o e al. (2006) and Goos and Manning (2007) a gued ha he ou iniza ion p ocess
may lead o occupa ional pola iza ion, whe e employmen g ow h is concen a ed in low
and high skill (wage) occupa ions, whe eas jobs in he middle o he skill con inuum a e
diminished.
4
Low-wage (educa ion), non- ou ine manual, and se ice jobs a e unlikely o
be di ec ly a ec ed by compu e iza ion, bu non-homo he ic p e e ences in p oduc
demand may lead o inc easing demand o low-paid se ices and non- ou ine manual
jobs.
The a icle is s uc u ed as ollows. We i s desc ibe he da a and p esen he agg ega e
pa e ns and he esul s om decomposi ions. Fi m-le el eg essions o employmen
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
ICT Usage E ode Rou ine Occupa ions 27
s uc u e on ICT usage a e hen p esen ed and he es ima es a e epo ed. A summa y con-
cludes he a icle.
2. Da a
We use linked da a ha ma ch wage s uc u e s a is ics o i m-le el echnology indica-
o s. The Ha monized S uc u e o Ea nings Su ey (HSES) da a om S a is ics Finland
combine annual ea nings s uc u e s a is ics da a in o ha monized panel da a, whe e all
wage measu es and classi ica ions, such as indus y and occupa ion, a e consis en ac oss
yea s and sec o s. The co e o he annual ea nings s uc u e s a is ics is he se o i m-
and indi idual-le el pay oll eco d da a o employe ede a ions o hei membe i ms.
S a is ics Finland conduc s an augmen ing su ey o he non-membe i ms and sec o s
ha a e no co e ed by he Con ede a ion o Finnish Indus ies (EK). S a is ics Finland
has calib a ed weigh s o he pay oll eco d da a ha u he imp o e he ep esen a i e-
ness o he da a.
5
We use hese weigh s in all ou analyses. The ha monized da a a e a ail-
able o he p i a e sec o o e e y yea om 1995 onwa ds. In ou analysis, he ecession
yea s ollowing he inancial c isis a e 2008 a e excluded, because ou in e es is he s uc-
u al e ec s o new echnology.
Ha moniza ion o e ime is needed because o he di e ences and changes in bo h
collec i e wage con ac s and classi ica ions used o e ime and ac oss sec o s. The annual
ha moniza ion ac oss di e en sec o s akes in o accoun he di e ences in wage concep s
and compensa ion componen s used in di e en collec i e ag eemen s. Fo example, hou ly
and mon hly pay schedules a e made compa able.
In he panel da a, educa ion, occupa ion, and indus y a iables a e ha monized o e
ime o he la es e sions o s anda d classi ica ions o S a is ics Finland. Fo mal
educa ion is a ailable om a comp ehensi e egis e o comple ed deg ees. The indus y
classi ica ions o i ms a e a ailable a he i e-digi le el bu used in he analyses a
he wo-digi le el. Occupa ion codes in he p ima y da a a e con e ed in o
in e na ional ISCO 2001 codes a he 5-digi le el bu used in he analyses a he h ee-
digi le el. Un o una ely, i is no possible o comple ely ha monize some occupa ions
o whi e-colla manu ac u ing wo ke s o e he b eak poin o 2001–2002 due o a
classi ica ion change in he p ima y employe pay oll eco d da a. Hence, we ei he
pe o m all ou es ima ions using sepa a e da a be o e and a e his b eak poin using
he pe iods o 1995–2001 and 2002–2008 o ocus only on he la e pe iod.
This longi udinal HSES da a o he yea s om 1995–2008 con ain some 600,000–
750,000 employees pe yea . App oxima ely 28,000 i ms exis in he da a o a leas 1 yea
du ing he pe iod o 1995–2008. Using sampling weigh s, hese da a a e ep esen a i e o
he o al p i a e sec o (wi h he excep ion o he smalles i ms, which a e exemp om
he cons uc ion o pay oll eco d da a ga he ed by employe associa ions and S a is ics
Finland).
Ou wage concep , he ‘hou ly wage o egula wo king ime’includes basic pay and
a ious supplemen s o wo king condi ions and pe o mance-based pay ha is paid on
a egula basis. I does no include o e ime pay o one-o i ems, such as holiday and
annual pe o mance bonuses. In addi ion o wages, we obse e egula wo king hou s
pe mon h o each wo ke in hese wage s a is ics. Because he employe o each pe -
son is known, we a e able o calcula e he o al numbe o employed pe sons and hei
o al mon hly wage bill o each i m in hese s a is ics. Finally, we obse e he
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
28 Pe i B€
ocke man —Seppo Laaksonen —Ja i Vainiom€
aki
educa ion le el and he occupa ion o each pe son, so we can disagg ega e hese mea-
su es acco ding o educa ion and occupa ion in o de o examine he employmen
s uc u e a he i m le el.
The wage bill is di ided in o h ee educa ion g oups (low, in e media e, and high) and
h ee occupa ion g oups (abs ac , ou ine, and se ice occupa ions). We ha e also con-
s uc ed simila sha es o hou s wo ked and employed pe sons, bu he esul s o hese
a e simila o he wage bill, so we epo only he la e . The low-educa ion g oup consis s
o hose wi h basic compulso y educa ion only. The high-educa ion g oup consis s o hose
wi h a uni e si y le el bachelo ’s deg ee o highe . The in e media e g oup consis s o all
deg ees in be ween, i.e., om oca ional o non-uni e si y highe deg ees ha usually
in ol e 2–4 yea s o educa ion.
The occupa ional g ouping is an applica ion o he classi ica ion p esen ed in Acemoglu
and Au o (2011) o he Finnish ISCO occupa ions. The abs ac g oup includes manage s,
p o essionals, and echnicians; he ou ine g oup includes sales, cle ical, p oduc ion, and
ope a o ’s wo k; and se ices include occupa ions in p o ec ion, ood p epa a ion, building
and g ounds, cleaning, and pe sonal ca e and se ices. The idea he e is ha al hough each
occupa ion can be hough as a bundle o di e en asks, each occupa ion is domina ed by
a main ask, ha can be cha ac e ized as abs ac (non- ou ine cogni i e), ou ine, o se -
ice (non- ou ine manual) ask. Thus, he ou ine con en is measu ed a he occupa ional
le el ollowing Acemoglu and Au o (2011).
We augmen wage and employmen s uc u e da a (HSES) wi h comp ehensi e i m-le el
a iables quan i ying he use o ICT. Ou da a ega ding he use o in o ma ion echnology
and elec onic comme ce in i ms o igina e om he S a is ics Finland su ey ‘Use o
In o ma ion Technology in En e p ises’(ICT su ey). The su ey is a s a i ied andom
sample o i ms in he sampling ame o he Business Regis e o S a is ics Finland. I co -
e s all la ge i ms (100 employees o mo e) and a andom sample o smalle i ms wi h
mo e han i e employees. Ma ching hese wo da a sou ces educes he numbe o i ms in
he linked da a subs an ially due o he sampling in ICT su eys. To pa ially add ess his
issue, we use all panel yea s o c ea e indica o s o whe he a i m in any yea du ing he
pe iod o 2002–2008 epo ed ha ing ce ain ICT echnology. This e ains in he da a all
i ms su eyed e en once du ing he pe iod.
6
The a iables in his su ey desc ibe he
usage o ICT in i ms, including In e ne , in ane , b oadband, home pages, se ices
o e ed ia home pages, elec onic comme ce, and elec onic da a in e change (EDI). The
ull lis and explana ions o a iables a e p o ided in Appendix A.
The a iables ha desc ibe a ious aspec s o ICT a e highly co ela ed because hey mea-
su e he unde lying cha ac e is ics o i ms ha a ec he adop ion o new echnologies.
7
Fo
his eason, we use ac o analysis o comp ess his in o ma ion in o la en ac o s, which we
use as explana o y a iables in ou eg essions. This alle ia es mul i-collinea i y and a iance
in la ion in he es ima ed models because he ac o s a e o hogonal. We use he p incipal
ac o s me hod and based on he eigen alues, h ee ac o s a e adequa e o desc ibe he com-
mon a iance o he ICT indica o s. The cumula i e a iance explained is 71%. The ac o
loadings a e documen ed in Appendix B. We call Fac o 1 EDI because i loads on a iables
ela ed o he usage o EDI by he i m o a ious pu poses (sending and ecei ing in oices,
o de s, o sending anspo documen s). Fac o 2 loads on a la ge numbe o a iables
ela ed o b oadband o mobile access o he In e ne , he i m ha ing a websi e, and whe he
he i m o de s o sells h ough compu e ne wo ks. This ac o also loads on he i m ha ing
en e p ise esou ce planning, bu we call his he In e ne ac o o sho . The hi d ac o
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
ICT Usage E ode Rou ine Occupa ions 29
loads on wo a iables indica ing whe he he i m sha es supply chain managemen (SCM)
da a wi h supplie s o cus ome s, so we call i he SCM ac o .
3. Agg ega e pa e ns o job pola iza ion
Figu e 1 illus a es he agg ega e pa e n o job pola iza ion in he Finnish p i a e sec-
o .
8
I shows ha changes in employmen sha es by ini ial occupa ional wage deciles ha e
been U-shaped in bo h he 1995–2001 and 2002–2008 pe iods, simila ly o he pola iza ion
pa e n documen ed o he UK in Goos and Manning (2007). On he o he hand, we ind
no indica ion o wage pola iza ion in Finland in B€
ocke man e al. (2013), whe e i is
shown ha wage g ow h inc eases almos linea ly wi h ini ial wage le els.
Figu e 1. Employmen pola iza ion
No es: Deciles a e de ined by o de ing occupa ions by hei median wage and di iding occupa ions in o 10 g oups
wi h equal sha e o o al hou s. The line shows he quad a ic i .
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
30 Pe i B€
ocke man —Seppo Laaksonen —Ja i Vainiom€
aki

Figu e 2 documen s he de elopmen o he employmen (hou s) sha es o occupa ion
g oups o e he pe iod o 1995–2013. The igu e shows ha employmen in ou ine-in en-
si e occupa ions has dec eased s eadily o e he yea s. We also ind inc easing employmen
in bo h abs ac and se ice ask-in ensi e occupa ions. These pa e ns a e consis en wi h
he ou iniza ion hypo hesis o ALM (2003).
4. Changes in ou ine and non- ou ine occupa ion’s sha es
The main a iables o in e es in ou examina ions a e he sha es o he o al wage bill
by educa ion and occupa ion g oups a he i m le el and hei changes o e ime. To
ob ain in o ma ion abou he possible sou ces o changes in labou demand s uc u e, we
p esen i m-le el decomposi ions o he changes in wage bill sha es. This decomposi ion
augmen s he Be man e al. (1994) indus y-le el decomposi ion o an unbalanced panel o
i ms wi h en y and exi .
9
The agg ega e change in he wage bill sha e o a wo ke g oup
de ined by educa ion o occupa ion (indexed by g) can be decomposed as ollows:
DPg¼X
i
DSi

Pg
iþX
i
DPg
i

SiþwN
PN
PS

þwD
sPS
sPD
s

whe e Pg¼Eg
EA,Pg
i¼Eg
i
Ei,Si¼Ei
ES,wN
¼EN
EA
and wD
s¼ED
s
EA
s
.
00.2 0.4 0.6 0.8
Occupa ion sha es
1995 2000 2005 2010 2015
Yea
Abs ac Rou ine
Se ice
Employmen sha es by occupa ion g oups 1995-2013
Figu e 2. De elopmen o employmen sha es o Abs ac , Rou ine, and Se ice
occupa ions, 1995–2013 [Colou igu e can be iewed a wileyonlinelib a y.com]
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
ICT Usage E ode Rou ine Occupa ions 31
P
g
is he agg ega e sha e o he skill g oup gin he o al wage bill o all i ms (deno ed
by E
A
), P
i
is he co esponding sha e in con inuing i m i(i=1, ...,N),S
i
is he sha e o
i m iin he agg ega e wage bill, Dindica es change o e he pe iod ( s, ), and ba s
deno e a e ages o e he pe iod’s ini ial-yea ( s) and inal-yea ( ) alues. Supe sc ip s
indica e he sums o sha es o all i ms (A), su i ing i ms (S), en e ing i ms (N), and
exi ing i ms (D). I can be shown ha he en y and exi e ec s can also be w i en as
ENTRY ¼wN
ðPN
PS
Þ¼ðPA
PS
Þ;
EXIT ¼wD
sðPS
sPD
sÞ¼ðPS
sPA
sÞ:
These e ec s he e o e depend on he de ia ion o he en e ing and exi ing plan ’s a e -
age skill g oup sha es wi h ha o con inuing plan s. The mo e posi i e he en y e ec is,
he highe he g oup’s sha e in new plan s is compa ed wi h con inuing plan s (PN
PS
).
Simila ly, he mo e posi i e he exi e ec is, he lowe he g oup’s sha e is in exi ing plan s
compa ed wi h con inuing plan s (PS
sPD
s). Howe e , i is no ewo hy ha he en y
e ec is also gi en by he simple di e ence be ween he g oup’s agg ega e wage bill sha e
o all i ms and con inuing i ms in he inal yea o he pe iod. Simila ly, he exi sha e is
gi en by he simple di e ence in he sha es o con inuing i ms and exi ing i ms in he
ini ial yea o he pe iod.
The o he wo e ms a e s anda d in indus y-le el decomposi ions. The i s sum is he
be ween- i m e ec , which cap u es he shi s o employmen (wage bill) be ween i ms wi h
di e en a e age wage bill sha es. I is posi i e i he wage bill shi s owa ds i ms ha
ha e a high wage bill sha e o he skill g oup in ques ion. The second sum is he wi hin-
i m e ec , which cap u es changes in he wage bill sha e wi hin each i m, weigh ed by he
i m’s a e age sha e o he o al wage bill. The con en ional in e p e a ion in he li e a u e
is ha he wi hin componen cap u es echnological change wi hin i ms, he be ween com-
ponen cap u es p oduc demand changes ac oss i ms, and he en y/exi componen s
e lec demog aphic changes in he i m popula ion (Goos e al., 2014). I should be no ed
ha echnology could also lead o be ween- i m changes by o cing ce ain i ms o down-
size o close while enabling o he i ms o g ow as e . Simila ly, p oduc demand changes
could lead o wi hin- i m changes i i ms adjus hei p oduc ion owa ds di e en goods.
Howe e , in ou empi ical applica ion he wi hin componen domina es and as usual, ou
da a do no con ain in o ma ion abou he p oduc composi ion o i ms’ou pu o s udy
po en ial demand e ec s in p ac ice.
We use he wage bill sha es as indica o s o he ela i e demand o di e en skill g oups
de ined by educa ion o occupa ion. The main jus i ica ion o his is ha ela i e labou
demand equa ions wi h wage bill sha es as he dependen a iables, can be de i ed om a
anslog cos unc ion (see Be man e al., 1994), which is a lexible, second-o de app oxi-
ma ion o a gene al cos unc ion. Howe e , i is clea , ha an inc ease in he demand o
a wo ke g oup may lead o an inc ease in i s wage bill sha e ei he because he employ-
men sha e o he g oup ises, o because he ela i e wage o he g oup ises, o bo h. As
a gene al indica o o ising demand, he wage bill cap u es bo h e ec s, bu i is also o
in e es o know how he ising demand is di ided be ween wages and employmen .
To examine his issue mo e closely, we use a (no el)
10
e sion o he wage bill decompo-
si ion, in which bo h he wi hin- i m and he be ween- i m componen s o con inuing
i ms a e u he di ided in o e ms ha cap u e only changes in ela i e wages
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
32 Pe i B€
ocke man —Seppo Laaksonen —Ja i Vainiom€
aki
( espec i ely ela i e employmen ) o di e en skill g oups, as ollows (supe sc ip g
deno es he skill g oup o in e es , and udeno es all o he g oups; e.g., gis highly educa ed
and uindica es he sums/a e ages o medium- and low- educa ed):
DPg¼X
i
DSWi
Pg
iþX
i
DSEi
Pg
iþX
i
DPWg
i

SiþX
i
DPEg
i

Si
whe e DSWi¼Wi; Ei;
W E Wi; sEi;
W sE

,DSEi¼Wi; sEi;
W sE Wi; sEi; s
W sE s

,
DPWg
i¼Wg
i; Eg
i;
Wg
i; Eg
i; þWu
i; Eu
i;
Wg
i; sEg
i;
Wg
i; sEg
i; þWu
i; sEu
i;

, and
DPEg
i¼Wg
i; sEg
i;
Wg
i; sEg
i; þWu
i; sEu
i;
Wg
i; sEg
i; s
Wg
i; sEg
i; sþWu
i; sEu
i; s

:
The DSW
i
and DSE
i
e ms di ide he be ween- i m change in a plan ’s sha e o o al
wage bill DSiin o componen s e lec ing only wage o employmen changes. The DPWg
i
and DPEg
ido he same o he wi hin- i m change in he skill g oup’s sha e o he wage
bill DPg
i. A pa icula i m’sDPWg
iis posi i e only i he ela i e wage o he skill g oup g
compa ed wi h all o he g oups u,(Wg
i; =Wu
i; ), inc eases in he i m om pe iod -s o pe -
iod . The hi d e m he e o e e lec s only he e ec o wi hin- i m changes in ela i e
wage o he skill g oup on he agg ega e change in he skill g oups’wage bill sha e. Simi-
la ly, DPEg
iis posi i e only i skill g oup’s ela i e employmen (Eg
i; =Eu
i; ) inc eases in i m
i, so ha he las e m e lec s changes only in he ela i e employmen wi hin i ms. The
i s wo be ween e ec s, on he o he hand, e lec changes only in he wage s uc u e
ac oss i ms, and shi s only in employmen ac oss i ms owa ds i ms wi h highe wage
bill sha es o he skill g oups o in e es . This mo e de ailed decomposi ion is aluable o
iden i ying bo h he sou ces o inc easing skilled wage sha e (be ween and wi hin compo-
nen s) and whe he he e ec s o such changes a e channelled in o ela i e wages o
employmen .
Table 1 epo s he decomposi ion o changes in wage bill sha es by educa ion g oups
o he pe iod 2002–2008.
11
We ind ha he wi hin and o al changes o he low- and
in e media e-educa ion g oups a e nega i e o his pe iod. The espec i e changes o he
highes educa ed a e la ge and posi i e. The en y componen o he basic educa ion
g oup and he exi componen o he highes educa ed a e also posi i e. This means ha
exi ing i ms used ela i ely less high-educa ed wo ke s han con inuing i ms, because hei
exi inc eased he high-educa ion sha e in he emaining i ms in agg ega e. Howe e , he
posi i e en y e ec o he low educa ed means ha also new i ms we e disp opo ion-
a ely in ensi e in low-skilled wo ke s. Toge he hese esul s imply ha a leas o some
ex en he en e ing i ms ehi e hose low educa ed ha we e made edundan by he exi -
ing i ms. The be ween componen s o all educa ion g oups a e minimal. These esul s
imply a apid skill upg ading a he highes educa ional le el du ing he 2000s, which o e -
whelmingly occu s wi hin i ms. The pa e ns a e b oadly consis en wi h job pola iza ion
in he sense ha he in e media e educa ion g oup loses sha es, bu in gene al, hese esul s
show ha he de elopmen has been ‘linea ’wi h espec o educa ion. The la ges decline
in sha es occu s o he lowes educa ed and he la ges inc ease o he highes educa ed.
Wage bill sha e changes can u he be decomposed in o ela i e wage and employmen
changes (Table 1). We ind ha in he wi hin componen s o educa ion, he employmen
changes clea ly domina e he o e all change. In he be ween componen o educa ion, he
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
ICT Usage E ode Rou ine Occupa ions 33
wage and employmen changes a e mo e balanced, bu hei con ibu ion o he o e all
change is minuscule. These esul s show ha he main ma gin o adjus men is employmen
a he han wages. Fu he mo e, we obse e no majo decline in wages o g oups wi h
inc easing employmen , which would be equi ed o he pu e supply explana ion o be alid.
Table 2 epo s he decomposi ions o change in wage bill sha es by occupa ion g oups. In
con as o educa ion, bo h wi hin and be ween componen s a e impo an o occupa ional
changes, and hei e ec s ha e he same di ec ion, excep o he se ice occupa ions. We ind
ha in o al he ou ine occupa ion sha e declines and he abs ac and se ice occupa ion
sha es inc ease, such ha he o al change is clea ly consis en bo h wi h he ou iniza ion
hypo hesis and wi h job pola iza ion. The en y and exi e ec s a e small in gene al, bu he
en e ing i ms a e mildly se ice in ensi e and less in ensi e in abs ac occupa ions.
12
The
shi s in p oduc ion be ween di e en i ms ( he be ween componen ) seems o be mo e
impo an in explaining pola iza ion in he occupa ional sha es han in educa ional sha es.
The shi s in p oduc ion owa ds se ice-in ensi e i ms and away om ou ine-in ensi e
i ms clea ly con ibu e o he pola ized pa e n o o al change. This inding sugges s ha
changes in p oduc demand o ou sou cing may ha e a ole in explaining he inc ease in he
se ice occupa ions. Howe e , o he abs ac and ou ine occupa ions, he o e whelming
majo i y o change occu s wi hin exis ing i ms, which is consis en wi h echnological change
being impo an in explaining he declining sha es o ou ine occupa ions.
To gain u he insigh in o he ex en o which be ween- i m changes a e d i en by
indus y composi ion e sus be ween- i m, wi hin-indus y ealloca ions, we ha e pe -
o med also indus y-le el decomposi ions (no epo ed in de ail). Fo occupa ions, he
i m-le el be ween componen o he ou ine g oup is 0.018 and o he se ice g oup i
is 0.013. In compa ison, hey a e 0.012 and 0.016 a he indus y le el. These esul s
Table 1. Decomposi ions o wage bill sha e by educa ion g oup, 2002–2008 change
Educa ion g oup Wi hin Be ween En y Exi
To al
change
G oup’s sha e
in 2008
Basic 0.051 0.001 0.012 0.006 0.046 0.138
Wage/Employmen 0.002/0.048 0.003/0.004
In e media e 0.018 0.001 0.004 0.007 0.028 0.571
Wage/Employmen 0.003/0.021 0.001/0.002
High 0.068 0.000 0.007 0.013 0.074 0.291
Wage/Employmen 0.001/0.069 0.002/0.002
Table 2. Decomposi ions o wage bill sha e by occupa ion g oup, 2002–2008 change
Occupa ion g oup Wi hin Be ween En y Exi
To al
change
G oup’s
sha e
in 2008
Abs ac 0.046 0.005 0.004 0.000 0.047 0.464
Wage/Employmen 0.011/0.035 0.005/0.010
Rou ine 0.038 0.018 0.001 0.001 0.056 0.450
Wage/Employmen 0.010/0.028 0.001/0.019
Se ice 0.008 0.013 0.005 0.001 0.009 0.086
Wage/Employmen 0.001/0.007 0.004/0.009
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
34 Pe i B€
ocke man —Seppo Laaksonen —Ja i Vainiom€
aki
Ou i m-le el decomposi ions o he changes in wage bill sha es indica e likely easons
behind hese changes. The decomposi ions o educa ion g oups show ha changes in edu-
ca ion sha es a e owa ds mo e-educa ed g oups in a ‘linea ’ ashion wi h espec o educa-
ion le el. This pa e n is consis en wi h he s anda d SBTC model. The o al change
occu s o e whelmingly wi hin i ms, which is sugges i e o a echnological cause o hese
changes. Fo changes in occupa ional sha es, we ind ha he inc ease in abs ac and he
dec ease in ou ine occupa ions also occu subs an i ely wi hin exis ing i ms.
Ou decomposi ions also show ha o se ice occupa ions, he shi s in p oduc ion
owa ds se ice-in ensi e i ms and he en y o new se ice-in ensi e i ms is ela i ely mo e
impo an han o o he occupa ion g oups. This pa e n indica es ha changes in p oduc
demand o ou sou cing may ha e a ole in explaining he inc ease in se ice occupa ions,
which p oduces a pola ized pa e n o o al changes o occupa ion g oups, i.e., inc easing
abs ac and se ice sha es and dec easing ou ine sha e. Au o and Do n (2013) p opose a
heo y ha assumes ha lo e o a ie y a ionalizes he inc ease in demand o se ice jobs
by highly paid abs ac wo ke s. The e is scan empi ical e idence o his gene al equilib ium
e ec in he p e ious li e a u e. Ou decomposi ions a e consis en wi h his explana ion
because we ind ha he inc easing se ice sha e is d i en by be ween- i m ealloca ion.
We u he examine he con ibu ions o he changes in employmen and wages o wage
bill changes using a new decomposi ion me hod. We ind ha he main adjus men channel
o bo h educa ion and occupa ion g oups is he employmen change wi hin i ms a he
han ela i e wages.
Fu he mo e, we examined he echnology-based explana ions o ou iniza ion and job
pola iza ion a he i m le el by using i m-le el indica o s o ICT usage as explana o y
a iables in he i m-le el eg essions. We i s pe o m a ac o analysis on a la ge numbe
o indica o s o ICT adop ion a he i m le el o ob ain ac o sco es o h ee ICT ac-
o s. We hen use hese ac o sco es as explana o y a iables in eg essions o changes in
he wage bill sha es o di e en educa ion and occupa ion g oups.
The OLS eg essions show ha he ICT ac o s a e associa ed wi h inc eases in he demand
o highly educa ed wo ke s and educ ions in he demand o he low educa ed, whe eas he
in e media e educa ion g oup is independen o ICT. In eg essions o occupa ion g oups,
we ind ha ICT ac o s a e associa ed wi h inc eases in abs ac occupa ion sha es and
dec eases in ou ine occupa ion sha es. These occupa ional pa e ns suppo he ou iniza ion
hypo hesis a he i m le el. Since ou iniza ion is he main mechanism p oducing pola iza-
ion, hese esul s a e also consis en wi h job pola iza ion. The se ice occupa ion sha e is
independen o ICT a he i m le el, and he inc easing agg ega e sha e o se ices is ela ed
o demand e ec s, as no ed abo e.
We ind ha a subs an ial sha e o he medium-educa ed wo k in abs ac occupa ions.
As he e is no one- o-one mapping be ween educa ion g oups and asks pe o med wi hin
hem, he e ec o ou iniza ion on he in e media e educa ed as a g oup is mi iga ed. The
esul s highligh ha educa ional upg ading and occupa ional pola iza ion e lec di e en
aspec s o ou iniza ion. These esul s s and in con as wi h hose in Michaels e al.
(2014), who ound ha ICT capi al educed he demand o he middle-educa ed mos .
Acknowledgemen s
We a e g a e ul o Ma hias S i le and pa icipan s a he Compa a i e Analysis o
En e p ise Da a con e ence in Is anbul 2015, EALE con e ence in Ghen 2016, and ETLA
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
ICT Usage E ode Rou ine Occupa ions 41

esea ch semina 2016 o commen s. We a e also g a e ul o wo anonymous e iewe s
and he edi o o use ul commen s.
Funding
This wo k was suppo ed by a g an om he Palkansaaja Founda ion and by he
Academy o Finland S a egic Resea ch Council unding o he p ojec Wo k Inequali y
and Public Policy (g an numbe s 293120 and 314208).
Appendix A Va iables and de ini ions o e ms used in he ICT su ey
Va iables
PC Fi m uses compu e s
INTER Fi m has In e ne connec ion
WEB Fi m has a websi e
EPURCH Fi m o de s h ough compu e ne wo ks (websi es o EDI)
ESALES Fi m sells h ough compu e ne wo ks (websi es o EDI; no email o de s)
BROAD Fi m has b oadband (ADSL, SDSL, cable modem; as e han ISDN)
MOB Fi m has mobile access o In e ne (lap op, mobile phone; 3G, 4G, o slowe )
ERP Fi m has En e p ise Resou ce Planning sys em (ERP p og amme); 2006 onwa ds
SCMT Sha ing Supply Chain Managemen (SCM) da a wi h supplie s using compu e
ne wo ks (inc. In e ne ); egula exchange o in o ma ion on demand
o ecas s, in en o ies, p oduc ion plans, deli e y p og ess, p oduc planning;
a ailable 2006–2009
SCMA Sha ing SCM da a wi h cus ome s; see abo e; a ailable 2006–2009
CRMINF Managemen and sha ing o cus ome in o ma ion wi h o he business unc ions
wi hin i ms; 2006 onwa ds
CRMANA Fi m analyses cus ome in o ma ion o ma ke ing pu poses (p ice se ing,
p omo ing sales, choosing deli e y channels); 2006 onwa ds
AUTTIED Elec onic da a in e change used
AUTLASVA Recei ing e-in oices using elec onic da a in e change; a ailable 2007–2009
AUTLASLA Sending e-in oices using elec onic da a in e change; a ailable 2007–2009
AUTKULJ Using elec onic da a in e change in sending anspo documen s; a ailable 2007–
2009
De ini ions o some e ms
B oadband B oadband is a elecommunica ions connec ion wi h a capaci y o a leas 256
Kbps. In he s a is ics on he use o in o ma ion echnology in en e p ises,
b oadband has in p ac ice been de ined h ough he ype o echnology used in
he connec ion as ei he DSL (e.g. ADSL) o o he b oadband connec ion ( as e
han a adi ional elephone modem o ISDN)
E-in oice An e-in oice is an elec onic in oice cons uc ed acco ding o a gene ally used
message o ma , whose da a can be handled and in e p e ed au oma ically. E-
in oices a e ansmi ed ia a elecommunica ions se ice p o ide o a bank, e.g.,
Fin oice, eIn oice, TEAPPSXML, Pos iXML
E-mail in oice An e-mail in oice is an in oice sen as a pd - ile a ached o an e-mail
EDI EDI (Elec onic Da a In e change) is a p ocedu e by which in o ma ion loca ed in
an en e p ise’s da a sys em is used o p oduce a speci ied da a low ha is
ansmi ed elec onically o a ecei ing en e p ise, whe e i is di ec ly
inco po a ed in o he da a sys em (e.g. o de , paymen o de o in oice, p ice lis
o p oduc ca alogue)
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
42 Pe i B€
ocke man —Seppo Laaksonen —Ja i Vainiom€
aki
EDI comme ce EDI comme ce is elec onic comme ce ha occu s be ween en e p ises h ough he
medium o EDI
EDI in oice An EDI in oice is an elec onic in oice in machine code acco ding o he EDI
s uc u e s anda ds. EDI in oices a e o en sen ia a elecommunica ions se ice
p o ide
Elec onic in oice An elec onic in oice is an in oice ansmi ed in elec onic o m: an EDI in oice,
an e-in oice, an e-mail in oice o some o he elec onic in oice. Paymen s en e ed
by a cus ome in o an online banking sys em o di ec debi a e no elec onic
in oices
Homepage A homepage he e is de ined as an en e p ise’s own In e ne homepages o i s
sec ion in he homepages o a g oup. Homepages do no e e , o example, o
publica ion o an en e p ise’s con ac de ails on a ious company and add ess lis s
In e ne sales In e ne sales a e communica ions be ween a pe son and a da a sys em. Online
shopping, as de ined he e, is an o de placed by comple ing and sending a eady-
made elec onic o m on he In e ne and shopping in ac ual In e ne shops.
O de s placed wi h a s anda d email message a e no de ined as online shopping.
Pu chases made on an ex ane subjec o he same condi ions a e also coun ed
as In e ne sales
Online shopping Online shopping is he o de ing o goods and se ices ia a compu e ne wo k,
ega dless o paymen o deli e y me hod
Sou ce: S a is ics Finland.
Appendix B Fac o analysis o ICT a iables
Fac o analysis desc ibes a iabili y among obse ed a iables in e ms o a lowe num-
be o la en a iables ha a e called ac o s. The a iables in he ICT su ey a e (mos ly)
bina y indica o s. Fac o analysis assumes ha he obse ed a iables a e con inuous (o a
leas o dinal), because hey a e modelled as linea combina ions o con inuous la en ac-
o s. One could p oceed wi h a ac o analysis o bina y a iables, which speci ies a logis-
ic link unc ion be ween obse ed indica o s and la en ac o s. Al e na i ely, one can
con inue o use o dina y ac o analysis bu base i on a e acho ic co ela ion ma ix.
Te acho ic co ela ions a e es ima es o he co ela ion coe icien o la en bi a ia e no -
mal dis ibu ion based on obse ed bina y a iables. The esul s o ac o analysis using
e acho ic co ela ions a e usually simila o hose ob ained wi h he bina y ac o analy-
sis. The di e ence is mainly ha e acho ic co ela ions ea bina y a iables as incom-
ple ely obse ed unde lying a iables a he han obse ed i ems o la en ac o s.
We used e acho ic co ela ions o he 16 bina y indica o s o he usage o di e en
aspec s o ICT in he i ms in he ICT su ey. Using o hogonal o a ion and he p incipal
ac o s me hod yielded he ollowing o a ed ac o loadings o h ee ac o s wi h eigen al-
ues g ea e han one (Table A1).
Table A1 Ro a ed ac o loadings (pa e n ma ix) and unique a iances
Va iable Fac o 1 Fac o 2 Fac o 3 Uniqueness
Fi m has websi e 0.1322 0.8854 0.1564 0.1742
Fi m has b oadband 0.1354 0.7020 0.3243 0.3837
Fi m has mobile access o In e ne 0.2011 0.7382 0.2517 0.3513
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
ICT Usage E ode Rou ine Occupa ions 43
Table A1. Con inued
Va iable Fac o 1 Fac o 2 Fac o 3 Uniqueness
Fi m o de s h ough compu e ne wo ks 0.2590 0.5490 0.3133 0.5333
Fi m sells h ough compu e ne wo ks 0.4405 0.5793 0.1155 0.4570
Fi m has En e p ise Resou ce Planning 0.2516 0.6405 0.3845 0.3786
Sha es SCM da a wi h supplie s 0.3013 0.0714 0.8988 0.0963
Sha es SCM da a wi h cus ome s 0.2640 0.2855 0.8003 0.2083
Sha es cus ome in o ma ion (wi hin i ms) 0.1774 0.6967 0.3841 0.3356
Analyses cus ome in o ma ion o ma ke ing 0.1907 0.6689 0.4096 0.3484
Recei es e-in oices ia elec onic da a exchange 0.7500 0.3795 0.0792 0.2872
Sending e-in oices ia elec onic da a exchange 0.7880 0.4563 0.0414 0.1692
Elec onic da a in e change used 0.9605 0.1052 0.2008 0.0260
Recei ing o de s 0.7872 0.2495 0.2425 0.2593
To supplie s 0.7729 -0.1093 0.3617 0.2598
Sending anspo documen s 0.7375 0.1138 0.3758 0.3020
Sou ce: ICT panel 2001–2009 (in o ma ion e e s o he end o p e ious yea , i.e., 2000–2008).
Va iables a e indica o s ha he i m has o u ilizes he echnology indica ed by he a iable name.
Fac o 1: Loads on using au oma ed da a exchange o sending and ecei ing in oices and
o de s and sending anspo documen s. We call his ac o EDI ( o elec onic da a in e change).
Fac o 2: Loads on i m ha ing a websi e and access o In e ne , selling, and placing o de s ia
compu e ne wo ks (websi e o EDI). I also loads on he i m ha ing a special p og amme
(CRM) o sha ing and analysing cus ome in o ma ion wi hin he i m and on he i m ha ing
En e p ise Resou ce Planning (ERP). We call his a gene al In e ne ac o . Fac o 3: Loads on
sha ing supply chain managemen (SCM) da a wi h supplie s o cus ome s ia compu e s (de-
mand o ecas s, in en o y le els, p oduc ion plans, deli e ies, p oduc planning in o ma ion); we
call his ac o SCM.
Appendix C In o ma ion on educa ion and occupa ion g oups
Table A2 Occupa ional composi ion o educa ion g oups
Low Medium High
Abs ac 0.14 0.26 0.81
Rou ine 0.72 0.60 0.15
Se ice 0.15 0.13 0.04
111
No e: Based on he a e age dis ibu ion o he yea s 2002 and 2008.
Table A3 P edic ed and obse ed alues o he wi hin con ibu ions and ICT e ec s
Low Medium High
P ed. wi hin con . 0.022 0.012 0.030
Obse ed 0.051 0.018 0.068
P ed. EDI 0.025 0.013 0.038
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
44 Pe i B€
ocke man —Seppo Laaksonen —Ja i Vainiom€
aki
Table A3. Con inued
Low Medium High
Obs. EDI (OLS) 0.024 0.020 0.040
P ed. In e ne 0.044 0.025 0.055
Obs. In e ne (OLS) 0.031 0.029 0.050
P ed. SCM 0.010 0.005 0.014
Obs. SCM (OLS) 0.013 0.008 0.019
No es: P edic ed alues o he wi hin con ibu ions and he e ec o ICT o each educa ion g oup a e
ob ained by weigh ing he wi hin con ibu ions in Table 2 and he es ima ed OLS coe icien s o ICT
ac o s o each occupa ional g oup in Table 4 by he occupa ional employmen sha es in Table A2
o each educa ion g oup. The obse ed alues o educa ion g oups a e om Tables 1 and 3.
Re e ences
Acemoglu D. (2002) ‘‘ Technical Change, Inequali y, and he Labo Ma ke ’,Jou nal o Economic Li -
e a u e 40: 7–72.
Acemoglu D. and Au o D. (2011) ‘Skills, Tasks and Technologies: Implica ions o Employmen and
Ea nings’in Ashen el e O. and Ca d D. (eds.) Handbook o Labo Economics, Vol. 4B,
Ams e dam: No h-Holland: 1043–1171.
Ake man A., Gaa de I. and Mogs ad M. (2015) ‘The Skill Complemen a i y o B oadband In e -
ne ’,The Qua e ly Jou nal o Economics 130: 1781–1824.
Asplund R., Ba h E., Lundbo g P. and Nilsen K. M. (2011) ‘Pola iza ion o he No dic Labou
Ma ke s’,Finnish Economic Pape s 24: 87–110.
Au o D. and Do n D. (2013) ‘The G ow h o Low-skill Se ice Jobs and he Pola iza ion o he US
Labo Ma ke ’,The Ame ican Economic Re iew 103: 1553–1597.
Au o D., Le y F. and Mu nane R. J. (2003) ‘The Skill Con en o Recen Technological Change: An
Empi ical Explo a ion’,The Qua e ly Jou nal o Economics 118: 1279–1333.
Au o D., Ka z L. and Kea ney M. (2006) ‘The Pola iza ion o he U.S. Labo Ma ke ’, NBER
Wo king Pape No. 11986, Na ional Bu eau o Economic Resea ch (NBER).
Ba el A., Ichniowski C. and Shaw K. (2007) ‘How Does In o ma ion Technology A ec
P oduc i i y? Plan -le el Compa isons o P oduc Inno a ion, P ocess Imp o emen , and
Wo ke Skills’,The Qua e ly Jou nal o Economics 122: 1721–1758.
Be man E., Bound J. and G iliches Z. (1994) ‘Changes in he Demand o Skilled Labo wi hin US
Manu ac u ing: E idence om he Annual Su ey o Manu ac u es’,The Qua e ly Jou nal o
Economics 109: 367–397.
B€
ocke man P., Laaksonen S. and Vainiom€
aki J. (2013) ‘Is The e Job Pola iza ion a he Fi m Le el?’,
Tampe e Economic Wo king Pape s No. 91, Uni e si y o Tampe e.
B€
ocke man P., Laaksonen S. and Vainiom€
aki J. (2016) ‘A e Jobs Mo e Pola ized in ICT i ms?’, Dis-
cussion Pape Se ies No. 9851, Ins i u e o he S udy o Labo (IZA).
Bound J. and Johnson G. (1992) ‘Changes in he S uc u e o Wages in he 1980s: An E alua ion o
Al e na i e Explana ions’,The Ame ican Economic Re iew 83: 371–392.
Ca d D. and DiNa do J. E. (2002) ‘Skill-Biased Technological Change and Rising Wage Inequali y:
Some P oblems and Puzzles’,Jou nal o Labo Economics 20: 733–782.
Ca d D., Heining J. and Kline P. (2013) ‘Wo kplace He e ogenei y and he Rise o Wes Ge man
Wage Inequali y’,The Qua e ly Jou nal o Economics 128: 967–1015.
Co es G. M. and Sal a o i A. (2015) ‘Task Specializa ion Wi hin Es ablishmen s and he Decline o
Rou ine Employmen ’, Unpublished manusc ip , Uni e si y o Manches e and RCEA.
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
ICT Usage E ode Rou ine Occupa ions 45
Co es G. M., Jaimo ich N. and Siu H. E. (2017) ‘Disappea ing Rou ine Jobs: Who, How, and
Why?’,Jou nal o Mone a y Economics 91: 69–87.
C ino R. (2010) ‘Se ice O sho ing and Whi e-colla Employmen ’,The Re iew o Economic S udies
77: 595–632.
Fe n
andez-Mac
ıas E. (2012) ‘Job Pola iza ion in Eu ope? Changes in he Employmen S uc u e and
Job Quali y, 1995–2007’,Wo k and Occupa ions 39: 57–182.
Gaggl P. and W igh G. C. (2017) ‘A Sho Run View o Wha Compu e s Do: E idence om a UK
Tax Incen i e’,Ame ican Economic Jou nal: Applied Economics 9: 262–294.
Goos M. and Manning A. (2007) ‘Lousy and Lo ely Jobs: The Rising Pola iza ion o Wo k in B i-
ain’,Re iew o Economics and S a is ics 89: 118–133.
Goos M., Manning A. and Salomons A. (2014) ‘Explaining Job Pola iza ion: Rou ine-Biased Tech-
nological Change and O sho ing’,The Ame ican Economic Re iew 104: 2509–2526.
Ha igan J., Reshe A. and Toubal F. (2016) ‘The Ma ch o he Techies: Technology, T ade, and Job
Pola iza ion in F ance, 1994–2007’, NBER Wo king Pape No. 22110, Na ional Bu eau o
Economic Resea ch (NBER).
Ikenaga T. and Kambayashi R. (2016) ‘Task Pola iza ion in he Japanese Labo Ma ke : E idence o
a Long- e m T end’,Indus ial Rela ions 55: 267–293.
Jenkins S. P. (1995) ‘‘ Did he Middle Class Sh ink Du ing he 1980s? UK E idence om Ke nel
Densi y Es ima es’,Economics Le e s 49: 407–413.
Machin S. (2008) ‘An App aisal o Economic Resea ch on Changes in Wage Inequali y’,Labou :
Re iew o Labou Economics and Indus ial Rela ions, 22 (S1), 7–26.
Michaels G., Na aj A. and Van Reenen J. (2014) ‘Has ICT Pola ized Skill Demand? E idence om
Ele en Coun ies o e 25 Yea s’,Re iew o Economics and S a is ics 96: 60–77.
Mi unen M. (2013) ‘Ty€
oma kkinoiden pola isaa io Suomessa’, VATT Mimeo No. 33, Go e nmen
Ins i u e o Economic Resea ch.
Pekkala Ke S., Maczuskij T. and Mali an a M. (2016) ‘Wi hin and Be ween Fi m T ends in Job
Pola iza ion: Role o Globaliza ion and Technology’, Wo king Pape No. 41, The Resea ch
Ins i u e o he Finnish Economy (ETLA).
Sy e son C. (2011) ‘Wha De e mines P oduc i i y?’,Jou nal o Economic Li e a u e 49: 326–365.
Vainiom€
aki J. (1999a) ‘Technology and Skill Upg ading: Resul s om Linked Wo ke -plan Da a o
Finnish Manu ac u ing’in Hal iwange J. C., Lane J. I., Sple ze J. R., Theeuwes J. M. and
T oske K. R. (eds.) The C ea ion and Analysis o Employe -Employee Ma ched Da a, Ams e -
dam: Else ie /No h-Holland: pp. 115–145.
Vainiom€
aki J. (1999b) ‘Technology, Demand o Skills, and Wages: Resul s om Linked Wo ke and
Plan Cha ac e is ics Da a o Finnish Manu ac u ing’, Tampe e Economic Wo king Pape s
No. 2/1999, Uni e si y o Tampe e.
No es
1
This pape is a subs an ially e ised e sion o ou ea lie wo king pape B€
ocke man e al.
(2013), in which we i s p esen ed he i m-le el decomposi ions and eg essions on echnology
explana ions o pola iza ion, which ha e also been used in o he pape s concu en ly o la e .
2
Co es e al. (2017) shows ha he changes in he demog aphic composi ion o he economy do
no explain he job pola iza ion in he US con ex .
3
ALM (2003) p o ided indus y-le el e idence using he ask measu es de i ed om he Dic-
iona y o Occupa ion Ti les and showed ha analy ical and in e ac i e non- ou ine ask inpu s
inc ease mo e and ou ine ask inpu s dec ease mo e in indus ies ha in es mo e hea ily in com-
pu e capi al. Acemoglu and Au o (2011) p esen ed a comple e ask-based model o echnological
change o explain he anomalies ela ed o SBTC model.
4
Jenkins (1995) p ecedes he cu en li e a u e on job pola iza ion.
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
46 Pe i B€
ocke man —Seppo Laaksonen —Ja i Vainiom€
aki

5
The e is a size h eshold o a leas i e pe sons o he i m o be included in he S uc u e o
Ea nings Su ey da a. The size limi o i e pe sons applies also o he S a is ics Finland’s su ey o
non-membe i ms included in he da a, which a e o en small. The da a also excludes he ollowing
indus ies: ag icul u e, o es y and ishe ies, household employe s, as well as in e na ional o ganiza-
ions employmen . Fu he mo e, he company’s op manage s and owne s and hei amily membe s
and he employmen spells beginning o ending du ing he e e ence mon h a e also excluded. A e
hese exclusions he da a co e 70% o he o al p i a e sec o employmen . These da a a e he e o e
ep esen a i e o he popula ion o i ms wi h mo e han i e pe sons, o he han he excluded indus-
ies and op manage s and owne s, when he su ey weigh s a e used.
6
The non-panel na u e o ICT su eys p e en us om es ima ing ully di e enced models wi h
ICT a iables also measu ed as changes. Because ICT i ems ela e o ai ly new echnologies, hei
usage may in p ac ice e lec changes.
7
Fo he 16 ICT indica o s only wo pai wise co ela ions a e no s a is ically signi ican a he 5%
le el. The co ela ions a e usually in he in e al o 0.30–0.50 (max 0.64) and s a is ically signi ican
a he 1% le el.
8
Ea lie Finnish e idence ega ding job pola iza ion a he agg ega e le el is p o ided in Asplund
e al. (2011), Mi unen (2013), and B€
ocke man e al. (2013). Pola iza ion is documen ed o a la ge
se o indus ialized coun ies (Goos e al., 2014; Ikenaga and Kambayashi, 2016). The agg ega e pa -
e n o employmen changes o h ee occupa ion g oups epo ed in Goos e al. (2014) is consis en
wi h job pola iza ion in Finland, so we main ain ha he omission o smalles i ms in ou da a does
no c ucially a ec ou esul s.
9
See Vainiom€
aki (1999a,b) o a de ailed de i a ion o his decomposi ion augmen ed o include
en y and exi e ec s.
10
This decomposi ion was i s p esen ed in Vainiom€
aki (1999a), bu i has gone unno iced in he
li e a u e. Vainiom€
aki (1999b) includes a mo e de ailed de i a ion and jus i ica ion o he decompo-
si ion.
11
We p esen hese decomposi ions he e as backg ound o he eg essions below only o he
2002–2008 pe iod, because he ICT a iables in eg essions a e a ailable only o he 2000s.
12
The e a e 5,327 con inuing i ms o e he pe iod o 2002–2008; 5,716 en e and 5,746 exi du ing
his pe iod. Thus, he small en y/exi e ec s imply ha he employmen s uc u es o en e ing and
exi ing i ms do no di e much om hose o con inuing i ms, a he han small amoun s o en y
and exi .
13
The (weigh ed) co ela ions o indi idual ICT indica o s wi h log i m size a e all s a is ically sig-
ni ican a he 1% le el and in he in e al om 0.10 o 0.44.
14
The e ec s o bo h ICT ac o s o o he o iginal ICT indica o s (no epo ed) a e essen ially
simila , bu he coe icien s a e smalle and less signi ican i i m size is no included in he equa-
ion as a con ol a iable.
15
Ou es ima ing equa ion can be in e p e ed as a i s -di e ences e sion o a le el equa ion o
wage bill sha es. Then, i m ixed e ec s in he le els o hese sha es a e con olled by di e encing.
16
The IV esul s a e epo ed in he wo king pape e sion (B€
ocke man e al., 2016).
©2018 The Au ho s. LABOUR published by Fondazione Giacomo B odolini and John Wiley & Sons L d
ICT Usage E ode Rou ine Occupa ions 47