Do the labour market returns to university degrees differ between high and low achieving youth? Evidence from Australia

Ma ks, Ga y N. A icle Do he labou ma ke e u ns o uni e si y deg ees di e be ween high and low achie ing you h? E idence om Aus alia Jou nal o Labou Ma ke Resea ch P o ided in Coope a ion wi h: Ins i u e o Employmen Resea ch (IAB) Sugges ed Ci a ion: Ma ks, Ga y N. (2018) : Do he labou ma ke e u ns o uni e si y deg ees di e be ween high and low achie ing you h? E idence om Aus alia, Jou nal o Labou Ma ke Resea ch, ISSN 2510-5027, Sp inge , Heidelbe g, Vol. 52, Iss. 5, pp. 1-14, h ps://doi.o g/10.1186/s12651-018-0241-0 This Ve sion is a ailable a : h ps://hdl.handle.ne /10419/216665 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/ Ma ks J Labou Ma ke Res (2018) 52:5 h ps://doi.o g/10.1186/s12651-018-0241-0 ARTICLE Do he labou ma ke e u ns o uni e si y deg ees di e be ween high and low achie ing you h? E idence om Aus alia Ga y N. Ma ks1,2* Abs ac In almos all de eloped coun ies he e has been subs an ial g ow h in uni e si y educa ion o e he las hal -cen u y. This g ow h has aised conce ns ha he bene i s o uni e si y educa ion a e declining and ha uni e si y educa- ion is no app op ia e o s uden s who, wi hou he expansion, would no ha e been admi ed. Fo such s uden s, oca ional educa ion o di ec en y o he labou ma ke may be mo e app op ia e. The pu pose o his s udy is o examine he e ec s o uni e si y and oca ional quali ica ions, ne o o he in luences on a a ie y o labou ma ke ou comes o Aus alian you hs up o age 25; and i he bene i s o uni e si y deg ees di e ac oss he achie emen con inuum. Achie emen is measu ed by es sco es in he OECD’s PISA assessmen s. The six labou ma ke ou comes in es iga ed a e: occupa ional s a us, hou ly and weekly ea nings, employmen , unemploymen and ull- ime wo k. The s udy inds ha uni e si y deg ees p o ide subs an ially supe io labou ma ke ou comes which a e no con ined o high and a e age achie e s, a leas o his coho in hei o ma i e yea s in he labou ma ke . Keywo ds: Uni e si y deg ee, Voca ional quali ica ions, You h, Occupa ional s a us, ea nings, unemploymen , Full- ime employmen JEL Classi ica ion: I26 Re u ns o Educa ion, J24 Human Capi al © The Au ho (s) 2018. This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0 In e na ional License (h p://c ea i eco mmons .o g/licen ses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons license, and indica e i changes we e made. 1 In oduc ion One o he mos d ama ic changes in he las hal -cen- u y in de eloped coun ies has been he expansion o uni e si y educa ion, om eli e o mass pa icipa ion. Be ween he 1960 and 2005, he p opo ion o US high school g adua es en olling a college inc eased om 45% o be ween 65 and 70% (Alon 2009, p. 788). Fo Canada, McIn osh (2010, p. 458) obse es ha pos -seconda y educa ion inc eased by a ac o o six be ween 1961 and 1997. A um e al. (2006, pp. 15, 16) show a doubling o highe educa ion pa icipa ion ac oss de eloped coun- ies om 20% in he 1960s o 40% in he 1990s. In mos coun ies g adua ion om uni e si y—wha he OECD (2009, p. 62) e e s o as e ia y ype A educa ion—has inc eased o a ound an a e age o 40% in OECD coun- ies wi h subs an ial inc eases in many coun ies o e he ela i ely b ie pe iod, 2000–2007. Acco ding o he Eu opean social su ey he p opo ion o g adua es in you h coho s in Eu opean coun ies inc eased, on a e age, om abou 12% in 1950–1960 o o e 40% in 2001–2010 (Koucký e al. 2010, p. 10). Pa icipa ion a es a e highes in I eland, Denma k, Spain and No way a a ound 60% o mo e. Fo Aus alia, less han 5% o coho s bo n be o e Wo ld Wa II a ended uni e si y. Fo he 1970–1981 coho , he igu e was close o 20% (Ma ks and McMil- lan 2007, p. 362). In 2012 o hose aged 25–35, 39% ha e a bachelo deg ee o highe uni e si y deg ee compa ed o 25% o hose aged 55–64 and 15% o hose aged 65 o mo e (Wilkins 2015, p. 70). Fo you h coho s bo n in Open Access Jou nal o Labou Ma ke Resea ch *Co espondence: [email p o ec ed].au 1 Di ec o a e o Go e nmen , Policy and S a egy, Aus alian Ca holic Uni e si y, Le el 4, 232 Vic o ia Pa ade, Eas Melbou ne, VIC 3002, Aus alia Full lis o au ho in o ma ion is a ailable a he end o he a icle Page 2 o 14 Ma ks J Labou Ma ke Res (2018) 52:5 he ea ly 1980s, 30–35% comple ed a uni e si y deg ee (Hé aul and Zaki o a 2015, p. 86). Acco ding o Je im and Vignoles (2015), 39% o he you hs en oll a uni e - si y by age 20, a a e simila o ha o o he Anglophone coun ies. The e is a b oad poli ical consensus o u he expand uni e si y educa ion by poli ical pa ies, uni e si ies and s akeholde s (Uni e si ies Aus alia 2013). Fo exam- ple, he Aus alian Labou Pa y’s pla o m has a a ge o 40% o 25–34yea -olds holding bachelo le el deg ee o highe by 2025. The jus i ica ions o u he expan- sion a e: uni e si ies’ ole in acili a ing inno a ion and he knowledge economy, he supe io labou ma - ke ou comes o uni e si y g adua es and sa is ying he inc eased demand o uni e si y educa ion. The e is also a, albei smalle , consensus ha you h unemploymen and o he labou ma ke p oblems in he school- o-wo k ansi ion a e bes add essed by u he in es men s in educa ion and aining a he han poli ically sensi i e e o ms o he labou ma ke . I is a gued ha oca ional educa ion and aining s ongly acili a es young people in he school- o-wo k ansi ion and expanding i would p o ide subs an ial economic bene i s in p oduc i i y, labou ma ke pa icipa ion and GDP g ow h (e.g. Rudd e al. 2007). Fo Aus alia, he OECD (2010) ad oca es bo h an expansion o oca ional educa ion and inc eased labou ma ke lexibili y. Du ing he las decade he e ha e been se e al subs an ial changes o he p o ision o oca ional educa ion (see Dempsey 2013) aiming o enhance he wo k skills o non-uni e si y bound you h. The e a e conce ns ha he expansion o he uni e - si y sec o has gone oo a . I is logical o suppose ha wi h expansion, uni e si y deg ees become de alued in he labou ma ke . The expansion has occu ed mainly by o me ly non-uni e si y ins i u ions—Cen es o Ad anced Educa ion, echnical, eache and nu sing colleges—being ans o med in o uni e si ies. O he 25–35-yea -old coho , only 23% ecei ed hei high- es uni e si y quali ica ion om one o he high s a- us ‘g oup o eigh ’ uni e si ies. Fo hose aged 65 and olde , he compa able igu e was 45% (Wilkins 2015, p. 70). Inc eased uni e si y pa icipa ion means he e a e mo e academically weake s uden s a uni e si y han in compa able olde coho s. The e is conce n ha he e is al eady an o e supply o uni e si y g adua es ha can- no ind app op ia e employmen especially in pa icula disciplines (Vogel 2013; Papadopoulos 2014). The labou ma ke ou comes o ecen g adua es ha e con inued o wo sen (ma ginally) ollowing hei de e io a ion in he a e ma h o Global Financial C isis (Gu h ie 2015). This conce n may be misplaced. Analyzing you h coho da a, Hé aul and Zaki o a (2015) ound p emi- ums o ea nings o be ween 8 and 11% o comple ed bachelo deg ees in he yea s ollowing g adua ion. They also iden i ied en olmen p emiums o a simila mag- ni ude which inc ease wi h ime since d opping ou o uni e si y. The e we e no comple ion p emiums o oca- ional educa ion al hough he e we e sizable en olmen e ec s, especially in he i s yea a e comple ing a spell o pos -school educa ion. In subsequen yea s, he labou o ce bene i s o en olmen in oca ional educa ion we e smalle and o en no s a is ically signi ican . A ecen analysis o Aus alian you h da a om se e al coho s no ed sligh ly lowe unemploymen a es o uni e si y deg ees o bo h males (2.0%) and emales (2.1%) com- pa ed o males (2.6%) and emales (2.2%) o e all (Hé aul e al. 2012). 1.1 Pu pose o his s udy The pu pose o his s udy is o compa e he bene i s (o o he wise) o uni e si y deg ees and oca ional quali ica- ions o a wide ange o labou ma ke ou comes and o examine i he labou ma ke bene i s associa ed wi h uni e si y deg ees also apply o low achie e s whom wi hou he expansion o uni e si y educa ion would no ha e had he oppo uni y o ob ain a uni e si y deg ee. The da a analyzed is om a younge Aus alian you h coho han analyzed in he s udies ci ed abo e. The labou ma ke ou comes in es iga ed a e: occupa- ional s a us, hou ly ea nings, weekly ea nings, employ- men , unemploymen , and ull- ime employmen . Hou ly ea nings allow compa ison o he e u ns o quali ica- ions pe hou wo ked. Among Aus alian young people he e is conside able pa - ime and casual employmen . Weekly ea nings a e a much be e indica o o he s and- a d o li ing. Simply ha ing a job is a posi i e labou ma ke ou come whe eas unemploymen is he mos undesi able ou come. A ull- ime job indica es a s ong a achmen o he labou ma ke and is o en he i s s ep in a ca ee . The e a e a a ie y o o he in luences on you h labou ma ke ou comes ha mus be conside ed in any analy- sis o he school- o-wo k ansi ion. These include gen- de , socioeconomic backg ound, achie emen o es sco es, school comple ion and labou o ce expe ience. Gende , quali ica ions and labou o ce expe ience a e s anda d a iables o include in labou o ce analyses. They a e also impo an in s udies o he you h labou ma ke (Ryan 2001). Typically, sepa a e analyses a e con- duc ed o young men and young women since he e a e impo an gende di e ences, no ably in ea nings, labou o ce pa icipa ion and ull- ime wo k. The OECD (2010) emphasizes he impo ance o a ‘good s a ’, gene ally de ined as app op ia e ull- ime employmen , o young people’s subsequen labou ma ke expe iences. Simi- la ly, he e is much e idence o he sca ing e ec s o Page 3 o 14 Ma ks J Labou Ma ke Res (2018) 52:5 unemploymen (A ulampalam e  al. 2000; Gangl 2004; G egg 2001; OECD 2010). Tes sco es aken a adoles- cence, a p oxy o abili y, ha e e ec s on (adul ) occupa- ional s a us, ea nings, unemploymen and o he labou ma ke ou comes (Ma ks 2014, pp. 91–112; Wa en e al. 2002). Analyses o labou ma ke ou comes in Aus alian you h coho s ypically ind e ec s o gende , socioeco- nomic backg ound, es sco es and labou o ce expe- ience on ea nings, unemploymen and occupa ional g oup, ne o educa ional quali ica ions (Huq 2014; Hé aul e al. 2012; Ma ks and Fleming 1998a, b; Buchle and Docke y 2015; Ma ks 2005). Inc easing school com- ple ion ( ha is comple ing yea 12) has long been a policy goal and ecen wo k sugges s ha school comple ion is bene icial (Ryan 2011). The ques ion o whe he he e ec s o educa ional qual- i ica ions on labou ma ke ou comes di e by s uden achie emen is examined using s uden s’ es sco es om he OECD’s P og am o In e na ional S uden Assess- men (PISA) es s conduc ed when hey we e 15yea s o age. 2 Ma e ials and me hods 2.1 Da a In 2003, a na ionally ep esen a i e sample o app oxi- ma ely 12,500 s uden s aged 15 yea s was selec ed o pa icipa e in PISA, conduc ed by he OECD. The PISA sample was cons uc ed by andomly selec ing 50 s u- den s aged 15yea s om each school om a sample o schools designed o ep esen all s a es and sec o s. Assessmen s in ma hema ical li e acy, eading li e acy, scien i ic li e acy and p oblem-sol ing we e adminis- e ed in hei schools o p o ide in o ma ion on s uden achie emen . S uden s also comple ed a backg ound ques ionnai e abou hei amilies, educa ional and oca- ional plans, a i udes o school and a ange o o he ma e s. In a ollow-up elephone in e iew, s uden s p o ided u he school and wo k in o ma ion. O he 12,500 s uden s in he PISA sample, 10,370 we e suc- cess ully con ac ed o pa icipa e in he subsequen 2003 Longi udinal S udies o Aus alian You h s udy. This was he basis o a subsequen longi udinal su ey o Aus- alian you h (LSAY-Y03) on school- o-wo k ansi ions. F om 2004 un il 2013, coho membe s we e in e iewed annually using compu e -assis ed elephone in e iews and in 2012 and 2013, esponden s also had he op ion o comple e hei in e iews online (NCVER 2010b). The o iginal PISA sample design o e -sampled small s a es, non-go e nmen schools and Indigenous s uden s. La ge schools had a g ea e chance o selec ion han smalle and e y small schools we e excluded. Sample a i ion be ween wa es compounded each yea , so ha he sample size in he inal 2013 wa e was only 3741, only 36% o he 10,370 esponden s in he o iginal LSAY Y03 sample. The e was di e en ial a i ion wi h a i ion mo e common among esponden s wi h pa en s wi h lowe s a us occupa ions and less educa ion, espond- en s om single pa en amilies and especially espond- en s wi h lowe achie emen sco es (Lim 2011, pp. 14, 15). The weigh s we e calcula ed o each wa e adjus ing o sampling p obabili ies and a i ion (Lim 2011). The es ima es p esen ed in his pape a e based on he inal weigh s and can be conside ed as he bes es ima es o popula ion pa ame e s. The majo dilemma in analysing ini ial you h labou ma ke ou comes wi h longi udinal da a is ha in he ea ly yea s mos esponden s a e in ull- ime educa ion so oo ew a e in he labou ma ke . In 2004, 90% we e in ull s udy and in 2008, he mid-poin o he s udy, 50% we e in ull- ime s udy. In la e yea s, a much highe p o- po ion we e in he labou ma ke bu he sample size is conside ably smalle due o a i ion. Analyzing da a om one o om a selec ion o calenda yea s will nec- essa ily disca d use ul da a. This p oblem is o e come by ocusing on obse a ions a he han esponden s, so ha all app op ia e esponden obse a ions a e u i- lized. The da a we e con e ed o a pe son-yea da a se so ha o each esponden he e a e be ween one and en obse a ions o hei educa ion and labou ma ke cha ac e is ics. 2.2 Measu es Mos o he measu es used in his s udy we e cons uc ed om he de i ed a iables al eady in he da a se . The de i ed a iables a e de ailed in a LSAY echnical pape (NCVER 2010a). 2.2.1 Labou ma ke ou comes (dependen a iables) The measu es o occupa ion s a us a e based on espond- en s’ occupa ions o each yea which we e coded o he ou -digi 2006 Aus alian and New Zealand s anda d classi ica ion o occupa ions (ANZSCO) by he Aus al- ian Bu eau o S a is ics. They we e hen con e ed o he Aus alian Socioeconomic Index known as AUSEI06 (McMillan e al. 2009). Concep ually, his measu e and simila socioeconomic indices a e based on he idea ha occupa ions con e human capi al (educa ion) in o ma e ial ewa ds (income). Ganzeboom e  al. (1992) de ail he concep ual basis and p ocedu e. AUSEI06 sco es we e calcula ed i e a i ely o each o he 358 ou - digi ANZSCO occupa ional codes o maximize he e ec o occupa ional s a us as an in e ening a iable be ween educa ion and income, ne o educa ion. AUSEI06 sco es we e escaled o ange be ween ze o and one hund ed. Indica i e sco es a e 100 o medical p ac i ione s and Page 4 o 14 Ma ks J Labou Ma ke Res (2018) 52:5 judges, 85 o school eache s, 63 o compu e echni- cians, 25 o mo o ehicle mechanics and 19 o bake s. Hou ly and weekly ea nings we e cons uc ed om he espec i e de i ed a iables. The a iables we e con- e ed in o 2013 dolla s adjus ed o in la ion by he consume p ice index (ABS 2015). Ea nings da a we e ob ained om bo h employees and he sel -employed. The measu es o employmen , unemploymen and ull- ime employmen we e cons uc ed om he de i ed a iables o labou o ce s a us and ull- ime o pa - ime employmen . Fo each yea , h ee dicho omous a i- ables we e cons uc ed dis inguishing employed (sco ed one) om no -being-employed (sco ed ze o); unem- ployed (sco ed one) e sus employed (sco ed ze o); and employed ull- ime (sco ed one) e sus no -employed- ull- ime, ha is employed pa - ime o employed unde some o he a angemen (sco ed ze o). 2.2.2 P edic o a iables Gende was measu ed by a dicho omous dummy a ia- ble, young men coded one and young women coded ze o. S uden achie emen is measu ed by s uden s’ com- bined sco es in he 2003 PISA es s in eading, ma h- ema ical and scien i ic li e acy. In he 2003 PISA ound all s uden s we e es ed in ma hema ics and due o he o a ed design abou hal we e es ed in eading, science o p oblem sol ing (OECD 2004, p. 336). S uden pe - o mance in each domain was measu ed by i e plausible alues. The plausible alues a e he esul o he o a ed design and I em Response Theo y me hodology. The measu es o s uden pe o mance ha e been ‘condi ioned’ o educe measu emen e o inc easing he co ela ions ac oss domains o be ween 0.8 and 0.9 (Bond and Fox 2001, p. 259; C omley 2009). The mean o he s uden s’ plausible alues in ma hema ics, eading, science and p oblem sol ing we e used o cons uc a single a iable measu e o es sco e, which was subsequen ly s anda d- ized o a mean o ze o and a s anda d de ia ion o one. Achie emen es s such as PISA a e cogni i ely demanding and a e o en unde s ood as measu es o abil- i y. In PISA, li e acy is de ined gene ally as “conce ned wi h he capaci y o s uden s o apply knowledge and skills in key subjec a eas and o analyze, eason and communica e e ec i ely as hey pose, sol e and in e p e p oblems in a a ie y o si ua ions” (OECD 2007, p. 16). Rinde mann (2008, p. 128) main ains he e is no impo - an heo e ical di e ence be ween s uden achie e- men and abili y es s since hey bo h assess “ hinking and knowledge”. Baume e al. (2009, pp. 3–5) poin s ou ha like in elligence es s, eading and ma hema i- cal assessmen s in ol e easoning and making logical in e ences. Rinde mann (2006, 2007, p. 687) concludes he e is a s ong ‘g’- ac o , ha is he gene al in elligence ac o , in in e na ional s uden assessmen s, including PISA. Bu han e al.’s (2017) measu e o adolescen s’ cog- ni i e abili y was hei PISA es sco es. Di ec e idence a he indi idual s uden le el is om he Ge man PISA 2000 s udy in which s uden s unde aking PISA also sa Ge man cogni i e abili y es s. B unne (2008, p. 153) epo ed co ela ions o a ound 0.8 o luid in elligence wi h e bal and ma hema ical es sco es. The measu e o socioeconomic backg ound used was he OECD’s cons uc ed measu e o economic, social and cul u al s a us (ESCS). I was cons uc ed om: he highes in e na ional socio-economic index o occupa- ional s a us o he a he o mo he ; he highes le el o educa ion o he a he o mo he con e ed in o yea s o schooling; he numbe o books in he home, and access o educa ional and cul u al esou ces, which we e ob ained by asking s uden s whe he hey had a hei home: a desk o s udy a , a oom o hei own, a quie place o s udy, a compu e hey can use o school wo k, educa ional so wa e, a link o he In e ne , hei own cal- cula o , classic li e a u e, books o poe y, wo ks o a books o help wi h hei school wo k, and a dic iona y (OECD 2004, p. 307). Fo hese analyses, he ESCS meas- u e was s anda dized o a mean o ze o and a s anda d de ia ion o one. Yea 12 comple ion s a us was based on he cons uc ed a iables on esponden s’ highes g ade o school edu- ca ion a he ime o in e iew. S uden s who had com- ple ed yea 12 we e assigned a sco e o one and s uden s ha had no , o had no ye , comple ed yea 12 we e assigned a sco e o ze o. Educa ional quali ica ions we e cons uc ed om he de i ed a iable which classi ied esponden s’ high- es educa ional le el ( o ha yea ) in o one o en ca - ego ies: VET ce i ica es I, II, III, IV, unknown, ad anced oca ional diploma, bachelo deg ee, g adua e diploma, highe deg ee and no pos -school quali ica ion. Respond- en s who had comple ed an app en iceship o aineeship we e ca ego ized unde he app op ia e ce i ica e le el. F om hese ca ego ies, h ee one-ze o dummy a iable measu es o educa ional quali ica ions we e cons uc ed: •VET ce i ica e III o IV, •Ad anced VET diploma, •Uni e si y deg ee (which includes bachelo deg ees and pos -g adua e quali ica ions). The e e ence ca ego y o hese dummy a iables was no holding any o hese quali ica ions. Measu es o he p opo ion o ime spen in ull- ime employmen and p opo ion o ime spen unem- ployed we e based on he de i ed a iables o ull- ime employmen and unemploymen . Fo each yea , i was Page 5 o 14 Ma ks J Labou Ma ke Res (2018) 52:5 asce ained i esponden s we e employed ull- ime a he ime o in e iew and, had hey expe ienced a leas one spell o unemploymen du ing ha yea (bu no nec- essa ily a he ime o in e iew). Fo each yea , da a om p e ious yea s we e used o cons uc wo measu es: he p opo ion o yea s esponden s we e employed ull- ime and he p opo ion o yea s hey expe ienced a leas one spell o unemploymen , independen o s udy s a us. Fo he analyses o employmen , unemploymen and ull- ime employmen he wo p opo ion measu es we e lagged by 1yea so ha he dependen and independen a iables we e no based on he same in o ma ion. The models include he a iable “yea s” he numbe o yea s since 2003, o index imp o emen s in labou ma - ke ou comes wi h ime and aging e ec s. Appendix Table7 p esen s he (weigh ed) numbe o cases and means o he analysis a iables om 2003 o 2013. In 2003 all esponden s we e in ull- ime educa ion so no labou ma ke ou comes could be analysed. Fo occupa ional s a us, ea nings and ull- ime employmen , he analyses we e es ic ed o esponden s employed and no s udying ull- ime. Fo employmen , he analyses we e es ic ed o esponden s no in ull- ime s udy. Fo unemploymen , he analyses we e es ic ed o espond- en s in he labou o ce. Appendix Table7 shows subs an ial o e - ime changes in he coho ’s educa ional quali ica ions and labou ma - ke ou comes. School comple ion inc eased d ama ically om 17% in 2004 o 71% in 2005 and hen pla eauing a 80%. Fo pos -school educa ion and aining he pe cen - age wi h a uni e si y deg ee inc eased om 3% in 2008 o 42% in 2013; om 12 o 27% o VET ce i ica es III o IV; and om 6 o 12% o oca ional diplomas. Unem- ploymen declined subs an ially om 18% in 2005 o < 5% in 2012 and 2013. A e age weekly ea nings (in 2013 dol- la s) among employed esponden s no s udying ull- ime inc eased om $554 in 2006 o o e $1000 in 2012 and 2013. Appendix Table8 p esen s he bi a ia e co ela- ions be ween a iables in he model. I sugges s posi i e e ec s o uni e si y deg ees and o a lesse ex en , yea 12 comple ion and combined PISA es sco e on he six labou ma ke ou comes examined. 2.3 Me hods Gene alized es ima ing equa ions (GEE) we e used o in es iga e he e ec s o he p edic o a iables on he dependen a iables o e he 10-yea pe iod (2004–2013) when esponden s we e aged be ween 16/17 and 25/26. The GEE app oach enables hese ela ionships o be analyzed o da a collec ed o e he en i e pe iod a he han o single yea s. The e o e, he es ima es a e based on a much la ge amoun o da a and obse a ions los h ough a i ion o he longi udinal s udy a e included in he analyses. The gene alized es ima ing equa ions (GEE) in oduced by Liang and Zege (1986a, b) is a me hod o analyzing clus e ed (o co ela ed) da a ex ending he gene alized linea model o non-no mal esponse a iables. GEEs ha e become an impo an s a egy in he analysis o co - ela ed da a. Less echnical discussions o GEEs a e a ail- able (Zo n 2001; Ghisle a and Spini 2004). One impo an ad an age o he GEE app oach is a subs an ial educ ion in he amoun o missing da a. Missing alues o which Yik a e missing whene e Yij is missing o all k > j a e called d opou s. O he wise, miss- ing alues ha occu in e mixed wi h non-missing al- ues a e in e mi en missing alues. The GEE app oach es ima es he wo king co ela ion om da a con aining bo h ypes o missing alues using he all a ailable pai s me hod, in which all non-missing pai s o da a a e used in he momen es ima o s o he wo king co ela ion ma ix pa ame e s. GEE analysis is a ype o andom e ec s model, bu he dependen a iable may no be no mally dis ibu ed. As in he case o gene alized linea models link unc ions a e speci ied o analyze non-no mally dis ibu ed ou - comes. GEE analysis can be unde s ood as a combina ion o andom e ec s models and gene alized linea models. Fo he analyses o occupa ional s a us, no link unc ion was speci ied so he es ima es can be in e p e ed in he same manne as coe icien s ob ained om o dina y leas squa es eg ession. Fo he analyses o he ea nings, he link unc ion is log because o he highly posi i e skew o he ea nings dis ibu ions. The es ima ed coe icien s a e in e p e ed as he pe cen age change in he depend- en a iable o a uni change in he independen a i- able. Fo he analyses o employmen and unemploymen he dependen a iables a e dicho omous and he link unc ion logi , so he coe icien s a e in e p e ed as odds a ios which a e he exponen s o he coe icien s. The co ela ions o ou come a iables among pe sons in longi udinal s udies mus be conside ed. The obse - a ions a e no s a is ically independen so he s anda d e o s equi e adjus men . The wi hin-pe son co ela- ions we e speci ied as i s o de au o eg essi e. Tables1, 2, 3, 4, 5 and 6 p esen he es ima es om he analyses. Fo each labou ma ke ou come h ee g oups o analyses we e conduc ed: o all esponden s and sepa- a ely o young men and young women. The es ima es o he wo s anda dized a iables— amily’s economic, social and cul u al s a us, and combined PISA es sco e—a e he di e ence on he espec i e ou come a i- able, ne o o he a iables in he model, o a one-s and- a d de ia ion on he p edic o a iable. The es ima es o he dicho omous dummy a iables a e in e p e ed Page 6 o 14 Ma ks J Labou Ma ke Res (2018) 52:5 as simply he di e ence on he labou ma ke ou come be ween a sco e o one (e.g. being male, ha ing comple ed yea 12, ha ing a uni e si y deg ee) con as ed o a sco e o ze o (e.g. being emale, ha ing no comple ed yea 12, no ha ing a uni e si y deg ee). Fo he p opo ion meas- u es he coe icien s a e he di e ence on he ou come measu e be ween 100% o he ime employed ull- ime (o unemployed) since 2003 and no ime spen employed ull- ime (o unemployed) since 2003. Few obse a ions sco e one and ze o, so i may be be e o di ide he es i- ma es by 10 which is he es ima e o a 10% inc ease in Table 1 E ec s o Quali ica ions and Co a ia es on Occupa ional S a us Reg ession coe icien s om epea ed design analyses. Full- ime s udy obse a ions excluded. Weigh ed o sample selec ion and a i ion ESCS economic, social and cul u al s a us, VET oca ional educa ional and aining, FT ull- ime, p op. y s. FT employed p opo ion o yea s since 2003 wi h ull ime employmen , p op. y s. unemp. spells p opo ion o yea s since 2003 wi h any spell o unemploymen * 0.05 > P > 0.01; ** 0.01 > P > 0.001; *** P < 0.001 All Young men Young women In e cep 33.59*** 33.46*** 31.01*** 30.84*** 32.90*** 32.77*** Male − 3.88*** − 3.90*** –––– Pa en s’ ESCS (s d.) 1.39*** 1.39*** 1.92*** 1.94*** 0.99** 0.97** PISA es sco e (s d.) 2.30*** 2.00*** 2.83*** 2.48*** 1.53*** 1.25*** Yea 12 comple ion 3.53*** 3.67*** 3.29*** 3.46*** 3.20*** 3.33*** VET ce i ica e III o IV 1.49** 1.49** 1.76* 1.79* 0.92 0.89 VET diploma 6.20*** 6.15*** 7.56*** 7.52*** 4.50*** 4.44*** Uni e si y deg ee 20.81*** 20.12*** 21.38*** 20.39*** 20.48*** 19.94*** P op. y s. FT employed 5.56*** 5.44*** 2.44 2.23 10.61*** 10.55*** P op. y s. unemp. spells − 1.50 − 1.50 − 1.52 − 1.41 − 2.45* − 2.53* Yea s since 2003 1.73*** 1.72*** 1.68*** 1.69*** 1.76*** 1.76*** Deg ee by es sco e – 1.74* – 2.09 – 1.58 N o obse a ions 24,351 24,351 11,913 11,913 12,438 12,438 N o pe sons (clus e s) 7415 7415 3602 3602 3813 3813 Clus e s wi h missing da a 2372 2372 1149 1149 1223 1223 Table 2 E ec s o quali ica ions and co a ia es on hou ly ea nings Logged eg ession coe icien s om epea ed design analyses. Full- ime s udy obse a ions excluded. Weigh ed o sample selec ion and a i ion. See Table 1 o abb e ia ions * 0.05 > P > 0.01; ** 0.01 > P > 0.001; *** P < 0.001 All Young men Young women In e cep 2.87*** 2.86*** 2.97*** 2.97*** 2.87*** 2.88*** Male 0.13*** 0.13*** – – – – Pa en s’ ESCS (s d.) 0.02* 0.02* 0.04** 0.04** 0.00 0.00 PISA es sco e (s d.) 0.03** 0.03* 0.01 0.01 0.05*** 0.07** Yea 12 comple ion − 0.04 − 0.04 − 0.05 − 0.05 − 0.01 − 0.01 VET ce i ica e III o IV 0.02 0.01 0.03 0.03 − 0.01 − 0.01 VET diploma 0.02 0.02 0.01 0.01 0.06* 0.04 Uni e si y deg ee 0.09*** 0.09*** 0.10*** 0.09** 0.09*** 0.09*** P op. y s. FT employed 0.17*** 0.17*** 0.25*** 0.24*** 0.02 0.03 P op. y s. unemp. spells − 0.18*** − 0.18*** − 0.22*** − 0.22*** − 0.08 − 0.08 Yea s since 2003 0.08*** 0.08*** 0.08*** 0.08*** 0.08*** 0.08*** Deg ee by es sco e – 0.01 – 0.03 – − 0.03 N o obse a ions 22,036 22,036 10,878 10,878 11,158 11,158 N o pe sons (clus e s) 7415 7415 3602 3602 3813 3813 Clus e s wi h missing da a 3585 3585 1694 1694 1891 1891 Page 7 o 14 Ma ks J Labou Ma ke Res (2018) 52:5 he p opo ion o yea s employed ull- ime employmen o he p opo ion o yea s wi h spells unemployed. The ables include addi ional in o ma ion: he numbe o obse a ions, he numbe o esponden s (clus e s) and he numbe o clus e s wi h missing da a. Fo dicho - omous ou comes, he ables also include he numbe o e en s, ha is he numbe o obse a ions whe e he esponden was employed, unemployed o employed ull- ime. To examine i he bene i s om uni e si y deg ees o labou ma ke ou comes di e ac oss he achie e- men con inuum, in e ac ion e ms we e included in he Table 3 E ec s o quali ica ions and co a ia es on weekly ea nings Logged eg ession coe icien s om epea ed design analyses. Full- ime s udy obse a ions excluded. Weigh ed o sample selec ion and a i ion. See Table 1 o abb e ia ions * 0.05 > P > 0.01; ** 0.01 > P > 0.001; *** P < 0.001 All Young men Young women In e cep 6.03*** 6.03*** 6.26*** 6.26*** 5.98*** 5.98*** Male 0.21*** 0.21*** – – – – Pa en s’ ESCS (s d.) 0.04*** 0.04*** 0.04* 0.04** 0.03*** 0.03*** PISA es sco e (s d.) 0.02 0.01 0.02 0.01 0.04*** 0.03 Yea 12 comple ion 0.06 0.06 0.02 0.03 0.13*** 0.14*** VET ce i ica e III o IV 0.03 0.03 0.03 0.03 0.01 0.01 VET diploma 0.03 0.03 0.00 0.00 0.10*** 0.10*** Uni e si y deg ee 0.27*** 0.25*** 0.24*** 0.22*** 0.30*** 0.29*** P op. y s. FT employed 1.07*** 1.06*** 1.06*** 1.06*** 1.01*** 1.01*** P op. y s. unemp. spells − 0.07 − 0.07 − 0.08 − 0.08 − 0.01 − 0.01 Yea s since 2003 0.06*** 0.06*** 0.06*** 0.06*** 0.05*** 0.05*** Deg ee by es sco e – 0.04* – 0.05 – 0.04* N o obse a ions 22,103 22,103 10,922 10,922 11,181 11,181 N o pe sons (clus e s) 7415 7415 3602 3602 3813 3813 Clus e s wi h missing da a 3585 3585 1694 1694 1891 1891 Table 4 E ec s o quali ica ions and co a ia es on employmen Logi eg ession coe icien s om epea ed design analyses. Con as employed s no -being-employed. Full- ime s udy obse a ions excluded. Weigh ed o sample selec ion and a i ion. See Table 1 o abb e ia ions * 0.05 > P > 0.01; ** 0.01 > P > 0.001; *** P < 0.001 All Young men Young women In e cep 1.05*** 1.04*** 1.08*** 1.09*** 1.17*** 1.15*** Male 0.14 0.14 – – – – Pa en s’ ESCS (s d.) 0.08* 0.08* 0.04 0.04 0.11* 0.11* PISA es sco e (s d.) 0.23*** 0.22*** 0.18*** 0.19*** 0.28*** 0.25*** Yea 12 comple ion 0.62*** 0.63*** 0.71*** 0.70*** 0.58*** 0.59*** VET ce i ica e III o IV 0.19 0.19 0.12 0.12 0.25 0.25 VET diploma 0.59*** 0.59*** 0.43* 0.43* 0.68*** 0.68*** Uni e si y deg ee 0.94*** 0.93*** 0.74*** 0.78*** 1.02*** 1.00*** P op. y s. FT employed 2.91*** 2.91*** 3.02*** 3.03*** 2.57*** 2.58*** P op. y s. unemp. spells − 0.92*** − 0.93*** − 0.87*** − 0.87*** − 0.96*** − 0.99*** Yea s since 2003 − 0.07*** − 0.07*** − 0.01 − 0.01 − 0.10*** − 0.11*** Deg ee by es sco e – 0.10 – − 0.14 – 0.28 N o obse a ions 26,964 26,964 13,058 13,058 13,906 13,906 N o e en s (y = 1) 23,935 23,935 11,733 11,733 12,202 12,202 N o pe sons (clus e s) 7415 7415 3602 3602 3813 3813 Clus e s wi h missing da a 1166 1166 549 549 617 617 Page 8 o 14 Ma ks J Labou Ma ke Res (2018) 52:5 second o each pai o analyses. Agg ega ion o es sco es o low and high achie ing g oups would be undesi able since i disca ds use ul in o ma ion (Jacca d and Tu isi 2003, p. 87). The coe icien s o he in e ac ion e ms a e in e p e ed as he di e ence in he e ec o ha ing a uni e si y deg ee on he espec i e labou ma ke ou - come o a uni di e ence (a one-s anda d de ia ion) in combined PISA es sco e ( om Jacca d and Tu isi 2003, p. 35). In he in e ac ion model, he in e p e a ion o he main e ec s o uni e si y deg ee and es sco e a e o Table 5 E ec s o quali ica ions and co a ia es on unemploymen Logi eg ession coe icien s om epea ed design analyses. Con as unemployed s employed. Full- ime s udy obse a ions excluded. Weigh ed o sample selec ion and a i ion. See Table 1 o abb e ia ions * 0.05 > P > 0.01; ** 0.01 > P > 0.001; *** P < 0.001 All Young men Young women In e cep − 2.22*** − 2.19*** − 1.85*** − 1.87*** − 2.35*** − 2.30*** Male 0.23* 0.23* – – – – Pa en s’ ESCS (s d.) − 0.10* − 0.10* − 0.11 − 0.11 − 0.10 − 0.10 PISA es sco e (s d.) − 0.35*** − 0.32*** − 0.32*** − 0.35*** − 0.37** − 0.28*** Yea 12 comple ion − 0.47*** − 0.48*** − 0.66*** − 0.65*** − 0.29* − 0.32* VET ce i ica e III o IV − 0.06 − 0.06 0.13 0.14 − 0.25 − 0.24 VET diploma − 0.25 − 0.25 − 0.13 − 0.13 − 0.36 − 0.35 Uni e si y deg ee − 0.56** − 0.57*** − 0.64** − 0.66** − 0.50 − 0.58** P op. y s. FT employed − 2.92*** − 2.91*** − 3.11*** − 3.12*** − 2.78*** − 2.76*** P op. y s. unemp. spells 1.51*** 1.51*** 1.47*** 1.48*** 1.56*** 1.64*** Yea s since 2003 − 0.01 − 0.01 0.01 0.01 − 0.02 − 0.02 Deg ee by es sco e – − 0.23 – 0.19 – − 0.62 N o obse a ions 25,492 25,492 12,541 12,541 12,951 12,951 N o e en s (y = 1) 1586 1586 816 816 770 770 N o pe sons (clus e s) 7264 7264 3533 3533 3731 3731 Clus e s wi h missing da a 1043 1043 501 501 542 542 Table 6 E ec s o quali ica ions and co a ia es on ull- ime employmen Logi eg ession coe icien s om epea ed design analyses. Con as : employed ull- ime s no employed ull- ime. Full- ime s udy obse a ions excluded. Weigh ed o sample selec ion and a i ion. See Table 1 o abb e ia ions * 0.05 > P > 0.01; ** 0.01 > P > 0.001; *** P < 0.001 All Young men Young women In e cep − 0.59*** − 0.62*** − 0.02 − 0.05 − 0.70*** − 0.73*** Male 0.42*** 0.42*** – – – – Pa en s’ ESCS (s d.) 0.02 0.02 − 0.02 − 0.02 0.05 0.05 PISA es sco e (s d.) − 0.01 − 0.06 0.00 − 0.06 0.01 − 0.04 Yea 12 comple ion 0.34*** 0.37*** 0.19* 0.22* 0.46*** 0.48*** VET ce i ica e III o IV 0.03 0.03 − 0.01 − 0.01 0.09 0.09 VET diploma 0.08 0.08 − 0.02 − 0.03 0.21 0.20 Uni e si y deg ee 0.66*** 0.56*** 0.56*** 0.45*** 0.70*** 0.62*** P op. y s. FT employed 3.76*** 3.75*** 4.06*** 4.03*** 3.38*** 3.38*** P op. y s. unemp. spells − 0.16 − 0.16 − 0.39* − 0.37* 0.05 0.04 Yea s since 2003 0.00 0.00 − 0.01 − 0.01 0.01 0.00 Deg ee by es sco e – 0.28*** – 0.27* – 0.27** N o obse a ions 23,935 23,935 11,733 11,733 12,202 12,202 N o e en s (y = 1) 16,649 16,649 8768 8768 7881 7881 N o pe sons (clus e s) 7005 7005 3414 3414 3591 3591 Clus e s wi h missing da a 840 840 402 402 438 438