A tale of two data sets: Comparing German administrative and survey data using wage inequality as an example

S übe , Heiko; G abka, Ma kus M.; Schni zlein, Daniel D. A icle A ale o wo da a se s: Compa ing Ge man adminis a i e and su ey da a using wage inequali y as an example 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: S übe , Heiko; G abka, Ma kus M.; Schni zlein, Daniel D. (2023) : A ale o wo da a se s: Compa ing Ge man adminis a i e and su ey da a using wage inequali y as an example, Jou nal o Labou Ma ke Resea ch, ISSN 2510-5027, Sp inge , Heidelbe g, Vol. 57, Iss. 1, pp. 1-18, h ps://doi.o g/10.1186/s12651-023-00336-9 This Ve sion is a ailable a : h ps://hdl.handle.ne /10419/278089 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/ S übe e al. J Labou Ma ke Res (2023) 57:8 h ps://doi.o g/10.1186/s12651-023-00336-9 ORIGINAL ARTICLE © The Au ho (s) 2023. Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License, which pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long as 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 licence, and indica e i changes we e made. The images o o he hi d pa y ma e ial in his a icle a e included in he a icle’s C ea i e Commons licence, unless indica ed o he wise in a c edi line o he ma e ial. I ma e ial is no included in he a icle’s C ea i e Commons licence and you in ended use is no pe mi ed by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om he copy igh holde . To iew a copy o his licence, isi h p:// c ea i eco mmons. o g/ licen ses/ by/4. 0/. Open Access Jou nal o Labou Ma ke Resea ch A ale o  wo da a se s: compa ing Ge man adminis a i e andsu ey da a using wage inequali y asanexample Heiko S übe 1,4* , Ma kus M. G abka2 and Daniel D. Schni zlein3,4,5 Abs ac The IAB’s Sample o In eg a ed Labou Ma ke Biog aphies (SIAB) and he Socio-Economic Panel (SOEP) a e he wo da a se s mos commonly used o analyze wage inequali y in Ge many. While he SIAB is based on adminis a i e epo s by employe s o he social secu i y sys em, he SOEP is a su ey da a se in which esponden s sel - epo hei wages. Bo h da a sou ces ha e hei speci ic ad an ages and disad an ages. The objec i e o his s udy is o desc ibe and compa e he e olu ion o wage inequali y o hese wo ypes o da a. Fo his pu pose, di e en sample es ic- ions a e applied. The compa ison wi hou any ha moniza ion o he da a shows di e en le els and ends. When he in o ma ion is la gely ha monized, compa able ends and simila le els eme ge. Keywo ds Wage inequali y, SIAB, SOEP, Adminis a i e da a, Su ey da a, Ge many JEL Classi ica ion D31, D33, J31 1 In oduc ion Fo yea s, he Socio-Economic Panel (SOEP) has been he s anda d da a se o analyzing income and wage1 inequali y in Ge many (e.g., S eine and Wagne 1998; Biewen 2000; Ge nand and P ei e 2007; Biewen and Juhasz 2012; Somme eld 2013). Howe e , esea ch based on adminis a i e da a has ecen ly gained impo ance, especially in labo economics (e.g., Ca d e  al. 2013; *Co espondence: Heiko S übe Heiko.S uebe @iab.de 1 Ins i u ü A bei sma k - und Be u s o schung (IAB), Regensbu ge S aße 10, 90478 Nü nbe g, Ge many 2 DIW Be lin/SOEP, Be lin, Ge many 3 Leibniz Uni e si ä Hanno e , Hanno e , Ge many 4 IZA, Bonn, Ge many 5 Innside S a is ics, Passau, Ge many 1 We use he e ms wage and sala y synonymously, since nei he he SOEP no he SIAB speci y whe he a pe son ecei es a ixed amoun pe pay pe iod (sala y) o is paid by he hou (wage). 8 Page 2 o 18 H.S übe e al. Fi zenbe ge and de Lazze 2022). La ge sample sizes and he (expec ed) accu acy o he in o ma ion included a e conside ed ad an ages o his ype o da a. Howe e , hese da a a e designed o adminis a i e pu poses and may no con ain all he in o ma ion needed o esea ch. A he same ime, mos la ge household su eys ha e begun o inc ease hei samples, in pa o include spe- ci ic subpopula ions.2 Compa ed o adminis a i e da a, su ey da a con ain much mo e in o ma ion and co e mo e opics, bu a he cos o much smalle sample sizes and pe haps less p ecision in ce ain quan i a i e a iables.3 Despi e hese de elopmen s, he e is a pauci y o wo k compa ing su ey and adminis a i e da a and examining whe he hese da a se s p oduce compa able esul s.4 In his a icle, we ill his gap o Ge many by compa ing ends and le els o wage inequali y based on he SOEP (e.g., Goebel e al. 2019) wi h esul s based on he Sample o In eg a ed Labou Ma ke Biog aphies (SIAB; e.g., F ode mann e al. 2021). Bo h da a se s a e widely used in empi ical labo esea ch and ha e been used ex ensi ely in pas esea ch on wage inequali y. Based on he SOEP, Biewen and Juhasz (2012), e.g., analyze de e minan s o he ise in income inequali y in he ea ly 2000s and show ha mos o he inc ease is due o ising labo income inequali y. Bu auel e  al. (2020) analyze he impac o he in oduc ion o he minimum wage on wage inequali y and ind ha his e o m leads o a educ ion in inequali y.5 In addi ion, he SOEP o ms he basis o many social epo ing s a is ics in Ge many. Se e al Ge man esea ch ins i u es publish inequali y s a is ics based on he SOEP a egula in e - als (e.g., S ockhausen and Calde ón 2020; G abka 2021), which a e inco po a ed in o policy and go e nmen al epo s (e.g., Bundes egie ung 2016, 2021; OECD 2018). Mos s udies on wage inequali y in Ge many based on adminis a i e da a ha e been conduc ed wi h da a se s om he Resea ch Da a Cen e (FDZ) o he Fede al Employ- men Agency (BA) a he Ins i u e o Employmen Resea ch (IAB). Dus mann e al. (2009), e.g., use he IAB Employmen Samples (IABS) o examine he Wes Ge man wage s uc- u e. They show ha wage inequali y in Wes Ge many has inc eased be ween 1975 and 2004. Klein e al. (2013) use he IAB’s Linked Employe -Employee Da a (LIAB) o ana- lyze he impac o expo ac i i y on wage inequali y wi hin and ac oss skill g oups. Ca d e al. (2013) use he In eg a ed Employmen Biog aphies (IEB) and ind ha inc easing he - e ogenei y a he es ablishmen le el and inc easing asse - i eness in assigning wo ke s6 o es ablishmen s mainly explain inc easing wage inequali y. Dus mann e al. (2014) use he SIAB and show ha eal wage g ow h is nega i e o he lowe end o he wage dis ibu ion. Fi zenbe ge and deLazze (2022) also use SIAB da a o examine whe he changes in he selec ion in o ull- ime employmen among Ge man men a e a cause o he ise in wage inequali y since he mid-1990s.7 Recen esul s based on he SOEP show ha (hou ly) wage inequali y—measu ed as he a io be ween he 90 h and 10 h pe cen iles—inc eased subs an ially in he ea ly 2000s, emained ela i ely s able be ween 2006 and 2014, hen declined un il 2016, and con inued o mo e ho izon al un il 2019, b inging he Ge man labo ma - ke back o he inequali y le el o he ea ly 2000s (G abka 2021). Using he SIAB da a h ough 2014, Fi zenbe ge and Seidli z (2020) con i m he same upwa d end in he ea ly 2000s o ull- ime wo ke wage inequali y, meas- u ed as he a io be ween he 80 h and 20 h pe cen- iles. Howe e , in hei sample, inequali y con inued o inc ease o men and dec ease o women h ough 2014. The dispa a e esul s based on di e en da a se s unde - sco e he need o a ho ough compa a i e analysis.8 The SOEP da a ha e he ad an age ha in o ma ion on indi iduals’ wages is embedded in he ull se o in o - ma ion a ailable in a la ge household su ey. This allows esea che s no only o calcula e indi idual ea nings o indi idual wages, bu also o include he household le el o o dis inguish be ween measu es o income be o e and a e ans e s and axes. Adminis a i e da a a e usu- ally p oduced by go e nmen agencies in he cou se o implemen ing ce ain ules, egula ions, and laws. S a is- ics a e a byp oduc o hese ac i i ies. On he one hand, adminis a i e da a can be conside ed e y accu a e in e ms o he in o ma ion equi ed o he adminis a i e p ocedu e om which he da a o igina e, e.g., as he e a e legal sanc ions in case o mis epo ing. On he o he hand, some addi ional in o ma ion collec ed as pa o he adminis a i e p ocedu e, such as educa ional s a us o occupa ion, may be conside ed less eliable. 2 Fo example, he SOEP included amilies wi h many child en in 2010, an o e sampling o high-income households in 2002, mig an s in 2013 and 2015, and e ugees since 2016 (B i zke and Schupp 2019). 3 See, e.g., he SOEP Wa e Repo 2018 (B i zke and Schupp, eds 2019), he echnical epo o he 11 h wa e o he UK Household Longi udinal S udy (Ca pen e 2021), o he PSID Main In e iew Use Manual o he 2021 elease (Beaule e al. 2021). 4 In AppendixA, we b ie ly e iew some a icles ha compa e adminis a- i e and su ey da a. 5 Fu he examples o pape s using he SOEP a e Ba olucci (2013), Selez- ne a and VanKe m (2016), G abka and Sch öde (2018), Ty owicz e al. (2018), Caliendo and Wi b od (2022), and Beckmannshagen and Sch öde (2022). 6 We use he e m wo ke o desc ibe bo h, blue colla (wo ke s) and whi e colla (employees). 7 The inding o an inc ease in wage inequali y since he ea ly 1990s is also con i med by Mölle (2016). Fu he examples o pape s using he SIAB, o i s p edecesso s— he IABS and he IEBS—a e, e.g., Kopczuk e al. (2010) and Felbe may e al. (2014). 8 Ano he da a sou ce o he analysis o wage inequali y is he Ge man S uc u e o Ea nings Su ey (VSE), see AppendixC. Page 3 o 18 8 A ale o  wo da a se s: compa ing Ge man adminis a i e andsu ey da a using wage inequali y… The e is no doub ha he se o po en ial con ol a i- ables in he SIAB is much smalle compa ed o he SOEP. Howe e , because su ey cos s a e subs an ial, he co e - age o su eys is usually limi ed (abou 30,000 indi idu- als in he las SOEP wa e), while he adminis a i e da a co e almos he en i e popula ion o indi iduals pa - icipa ing in he labo ma ke . The IEB co e s, in e alia, all wo ke s subjec o social insu ance con ibu ions in Ge many since 1975. The SIAB is a wo-pe cen andom sample o he IEB. This es ic ion is due o da a p o ec- ion pu poses. Ne e heless, he SIAB co e s labo ma - ke ela ed in o ma ion o a ound 1.78 million pe sons. Adminis a i e da a ha e he ad an age ha he cha - ac e is ics essen ial o he unde lying adminis a i e p o- cess a e usually o high quali y and he e o e ha e a low numbe o missings and a low measu emen e o . How- e e , adminis a i e da a can be a ec ed by, e.g., p o- cessing e o s (e.g., duplica e epo s, da a en y e o ) (Kap eyn and Ypma 2007; G oen 2012; Lindne and And easch 2014) and/o co e age e o s (e.g., adminis- a i e da a lack in o ma ion on he shadow economy). Whe eas, su ey da a can be subjec o a ious meas- u emen e o s (e.g., Bound e  al. 2001). Basically, a dis inc ion can be made be ween sampling e o s and non-sampling e o s. Sampling e o s occu when only a non- ep esen a i e subse o he popula ion is ac ually su eyed. Non-sampling e o s include co e age e o s, aming e o s, esponse/non- esponse e o s, measu e- men e o s, and p ocessing e o s (e.g., deLeeuw e al. 2008). In summa y, bo h su ey da a and adminis a i e da a a e subjec o di e en ypes o measu emen e o s and he e o e nei he da a sou ce should be conside ed he only ue one.9 Fo Ge many, his has been shown, e.g., by Obe ski e  al. (2017). They “ ound o o icial adminis a i e da a ob ained om he Ge man Fede al Employmen Agency ha he eliabili y o bo h su ey and adminis a i e da a was a om pe ec .” (Obe - ski e al. 2017,p. 1486). Any compa ison o su ey and adminis a i e da a addi ionally aces di e ences in he de ini ions o he uni o analysis, di e gen e e ence pe iods, o e en censo ing. As no ed abo e, bo h he SOEP and he SIAB ha e been used o analyze wage inequali y in he pas , using di e en analy ical me hods ha ake ad an age o he unique cha ac e is ics o he wo da a se s. To ou knowl- edge, howe e , no a emp has e e been made o de i e compa able inequali y es ima es om hese da a sou ces. In his a icle, we ill his gap by b inging in es ima es o wage inequali y ends o Ge many based on (1) samples ha exploi he s eng hs o each da a se , (2) samples ha a e as compa able in composi ion as possible, and compa ing ou indings om hese app oaches. 2 Da a In he ollowing, we b ie ly in oduce he wo da a se s we use—SOEP and SIAB. We hen de ine di e en sub- samples o he wo da a se s ha we use o he analyses. Mo e de ailed in o ma ion on he da a se s, da a p epa a- ion, and sub-sampling can be ound in AppendixB. 2.1 Ge man Socio‑Economic Panel (SOEP) The SOEP is a ep esen a i e household su ey. I has been conduc ed annually since 1984 and in he las wa e co e ed mo e han 20,000 households wi h mo e han 30,000 indi iduals (excluding child en). The SOEP co e s a wide ange o opics, including de ailed in o ma ion on ea nings and wages a bo h he indi idual and household le el (Goebel e al. 2019). Due o i s na u e as a house- hold su ey, he SOEP also co e s ci il se an s and sel -employed pe sons as well as ma ginally employed pe sons in addi ion o employees. Wages a e su eyed o he p e ious mon h, including any o e ime pay, and o he p e ious yea , b oken down in o one- ime paymen s and se e ance paymen s. Wages om seconda y employ- men a e que ied sepa a ely. I should be no ed ha he way in which in o ma ion o seconda y employmen is collec ed has changed undamen ally in 2017. Since hen, dependen employmen can be clea ly dis inguished om an hono a y and sel -employmen seconda y employ- men . P e iously, his dis inc ion was no possible. In case o i em non- esponse, he main impu a ion me hod is he so-called ow-and-column impu a ion de eloped by Li le and Su (1989). When longi udinal in o ma ion o he impu a ion p ocess is missing, OLS eg essions a e applied (see F ick and G abka 2005).10 We use he SOEP-Co e 37 o ou analyses. Fo mo e de ails and a b ie desc ip ion o he da a p epa a ion s eps pe o med see Appendix B1 and Sch öde e  al. (2020). 2.2 Sample o in eg a ed labou ma ke biog aphies (SIAB) The SIAB is a wo pe cen andom sample d awn om he IEB o he IAB. The IEB is an adminis a i e da a se wi h in o ma ion om a ious da a sou ces. I includes, 9 A b ie o e iew o a icles examining he e o s uc u e in ea nings da a is p o ided in AppendixA. 10 The SOEP also asks o bo h ag eed and ac ual wo king hou s, so ha hou ly wages can be calcula ed in addi ion o mon hly o annual wages. Since hou ly wages canno be calcula ed in he SIAB due o he lack o wo king ime da a, his wage concep is no p esen ed he e. 8 Page 4 o 18 H.S übe e al. among o he s, all wo ke s subjec o social insu ance con ibu ions and all ma ginally employed indi iduals in Ge many.11 The employmen in o ma ion in he IEB comes om he in eg a ed no i ica ion p ocedu e o heal h, pension and unemploymen insu ance (Bende e al. 1996). As a pa o his p ocedu e, employe s a e equi ed o submi no i ica ions on all hei employees subjec o social secu- i y insu ance o he ele an social insu ance ins i u ions a leas once a yea . Ci il se an s and sel -employed indi iduals a e no subjec o social secu i y insu ance and a e he e o e no included in he da a se . Wo ke s can be iden i ied by an a i icial indi idual ID and acked o e yea s. The da a is o ganized by employ- men spells. The maximum leng h o a spell is one calen- da yea . Fo each employmen spell, he beginning and end o employmen on a daily basis and he a e age g oss daily wage a e known, among o he hings. The wage da a in he SIAB is e y eliable. The in o - ma ion is used, e.g., o calcula e e i emen pensions and unemploymen insu ance bene i s. Howe e , he wage da a is only ele an up o he social secu i y con ibu- ion assessmen ceiling. Fo his eason, he wage in o - ma ion in he p ocess da a is op-coded, so ha we only obse e wages up o he con ibu ion assessmen ceiling. The e o e, ollowing S übe e al. (2023), we impu e op- coded wages using a 2-s ep impu a ion p ocedu e simila o Dus mann e al. (2009) and Ca d e al. (2013).12 We use he SIAB 7519 o ou analyses. Fo mo e de ails and a b ie desc ip ion o he da a p epa a ion s eps pe o med see Appendix B2 and F ode mann e al. (2021). 2.3 De ining sub‑samples o  heanalyses To highligh he s eng hs and weaknesses o he SOEP and SIAB da a and o compa e he wo da a sou ces, we de ine di e en sub-samples o he da a se s in Sec .2.3.1, which we hen use o he analyses. The a ionale behind he sub-samples he eby is he ollowing: he samples desc ibed in he nex sec ion play o he indi idual s eng hs o he wo unde lying da a se s. Resea che s s udying wage inequali y based solely on he SOEP o he SIAB will mos likely end up using one o hese sub- samples. These hus ep esen he s anda d use cases o he wo da a se s. In con as , he samples desc ibed in Sec .2.3.2 ollow he in en ion o making he wo da a se s as compa able as possible. The ollowing basic es ic ions apply o all samples we d aw o ou analyses in his a icle: • The analyses a e pe o med o he las wo decades co e ed by ou da a, i.e., 2000 o 2019. • We conside only wo ke s be ween he ages o 18 and 65. • Wages om sel -employmen o wages paid as pa o an in e nship o , e.g., a olun a y social yea a e no aken in o accoun .13 2.3.1 Exploi ing hes eng hs o  heSOEP o SIAB Fo each da a se , we c ea e wo sub-samples o ake ad an age o he espec i e s eng hs o he da a se . The basic es ic ions lis ed in Sec .2.3 apply he e as well, o cou se, bu a e no epea ed in he ske ches o he indi- idual da a se s. 2.3.1.1 SOEP‑Pu e 1 (mon hly wage) Wo ke s consid- e ed: All wo ke s subjec o social insu ance con ibu ions and ci il se an s. Sampling: All esponden s who epo ed ei he posi- i e wages in he las mon h o posi i e wages in a sec- ond job when wages om he main job we e ei he ze o o missing. Wage o he calcula ion o wage inequali y: Nominal g oss mon hly wage o pe son’s main job o , i no a ail- able, second job in he mon h p io o he in e iew. 2.3.1.2 SOEP‑Pu e 2 (annual wage) Wo ke s consid- e ed: All wo ke s subjec o social insu ance con ibu ions and ci il se an s. Sampling: All esponden s who epo ed posi i e indi- idual wages o he las calenda yea . Since he la es su ey yea a ailable in he SOEP 37 is 2020, he la es a ailable obse a ion o wages in he las calenda yea is o 2019. Wage o he calcula ion o wage inequali y: Nomi- nal g oss annual wages (sum o all labo wages exclud- ing income om sel -employmen ) o an indi idual in yea p eceding he su ey yea , including any one- ime paymen s. 11 In o ma ion on ma ginal pa - ime employed pe sons has been included in he IEB since 1999. 12 We impu e igh -censo ed wages mainly o display he mean wages in he desc ip i e s a is ics in Appendix B2. Impu a ion does no play a ole in he calcula ion o ou inequali y measu es. See also Foo no e 18. 13 A p ecise delinea ion o he g oups o people included in he analyses can be ound in he AppendixB1 o he SOEP and AppendixB2 o he SIAB. Page 5 o 18 8 A ale o  wo da a se s: compa ing Ge man adminis a i e andsu ey da a using wage inequali y… 2.3.1.3 SIAB‑Pu e 1 (a e age daily wage) Wo ke s con- side ed: All wo ke s subjec o social insu ance con ibu- ions. Sampling: Conside a ion o all employmen spells. Wage o he calcula ion o wage inequali y: A e age nominal g oss daily wage o all jobs held by a pe son du - ing he yea [weigh ed by he du a ion o he employmen spell; ollowing Dus mann e al. (2009)]. 2.3.1.4 SIAB‑Pu e 2 (daily wage aso June 30) Wo ke s conside ed: All wo ke s subjec o social insu ance con- ibu ions. Sampling: Conside a ion o all employmen spells as o June 30 o each yea . Wage o he calcula ion o wage inequali y: Sum o all nominal g oss daily wages o an indi idual as o June 30 o each yea .14 2.3.2 Gene a ing compa able sub‑samples The samples in he las sec ion co espond o wha esea che s migh use i hey we e analyzing only one o he wo da a sou ces. In con as , in his sec ion, we c ea e sub-samples ha a e as compa able as possible be ween he wo da a sou ces. Fi s , we add ano he basic es ic ion o all compa- able sub-samples: o ensu e ha he indi iduals con- side ed a e compa able, we exclude ci il se an s om he SOEP—o , mo e p ecisely, wages om ci il se an employmen ela ionships—in he compa able samples. The o he basic es ic ions lis ed in Sec .2.3 also apply he e, bu a e no epea ed. 2.3.3 SOEP‑SIAB‑compa able 1: G oss mon hly wage The SOEP asks o wage in o ma ion o he mon h p e- ceding he su ey mon h. In e iews a e conduc ed in almos all mon hs, bu he as majo i y o su eys ake place om Feb ua y o May.15 The e o e, we eplica e his su ey s uc u e in he SIAB. In each yea , we andomly assign an in e iew mon h o each pe son in he SIAB. We ensu e ha he andom assignmen o he mon h esul s in he same dis ibu ion as o indi iduals aged 18 o 65 in his yea ’s SOEP. I a pe son in he SIAB is no employed in he mon h p eceding he assigned in e iew mon h, his indi idual is no included in he analysis. Since he SIAB does no con ain he mon hly wage, we calcula e he g oss mon hly wage o he day as he g oss daily wage (hence o h daily wage) mul iplied by he numbe o days in he employmen spell in he espec i e mon h. The daily wage in he SIAB also includes bonus paymen s, e c. (e.g., Ch is mas bonus) ecei ed by he employee in he du a ion o he spell. This in o ma ion is no included in he mon hly wage in he SOEP. How- e e , he in o ma ion on bonus paymen s is collec ed e - ospec i ely o he las su ey yea . To mimic he wage measu e in he SIAB, we use his in o ma ion as a p oxy and add 1/12 o he one- ime paymen s collec ed e o- spec i ely in he SOEP o he mon hly wage. I wo ke s a e employed by mo e han one employe , only he in o ma ion o he main employmen , i.e., he job wi h he highes mon hly wage, is used. This leads us o he ollowing compa able samples: 2.3.3.1 SOEP‑compa able 1 Wo ke s conside ed: All wo ke s subjec o social insu ance con ibu ions. Sampling: All esponden s who epo ed ei he posi i e wages in he las mon h o posi i e wages in a second job when wages om main job we e ei he ze o o missing. Wage o he calcula ion o wage inequali y: Nominal g oss mon hly wage o pe son’s main job o , i no a ail- able, second job in he mon h p io o he in e iew; adjus ed p opo ionally o bonuses, e c. 2.3.3.2 SIAB‑compa able 1 Wo ke s conside ed: All wo ke s subjec o social insu ance con ibu ions. Sampling: Annual andom assignmen o a su ey mon h; dis ibu ion o su ey mon hs is p ede e mined by he ac ual dis ibu ion o su ey mon hs in SOEP. Wage o he calcula ion o wage inequali y: Nominal g oss mon hly wage (calcula ed o he day) o pe son’s main job in he mon h p io o he assigned in e iew. 2.3.4 SOEP‑SIAB‑compa able 2: G oss annual wage Since he SIAB does no include annual wages, we calcu- la e g oss annual wages by adding up all wages (employ- men spell du a ion mul iplied by he g oss daily wage) o each pe son in each yea .16 2.3.4.1 SOEP‑compa able 2 Wo ke s conside ed: All wo ke s subjec o social insu ance con ibu ions. Sampling: All esponden s who epo ed posi i e indi- idual wages o he las calenda yea . Wage o he calcula ion o wage inequali y: Nominal g oss annual wage o an indi idual in yea p eceding he su ey yea , including any one- ime paymen s. 14 Agg ega ed da a se s, such as he Es ablishmen His o y Panel (BHP), a e gene a ed a he FDZ as o June 30. The e o e, on his cu -o da e, indi idual- le el da a can be en iched wi h agg ega ed es ablishmen da a. In addi ion, June 30 is equen ly used as he cu -o da e in o icial s a is ics in Ge many. 15 A small ac ion o he SOEP su eys is conduc ed in Janua y and he e- o e con ains he mon hly wage o Decembe o he p e ious yea . We ha e chosen o epo all wages o a gi en su ey yea . 16 We apply he dis ibu ion o he in e iew mon hs o he SOEP o he SIAB, which is co ec o he case ha we conside mon hly wages. I we conside annual wages, we would ac ually ha e o use he dis ibu ion o yea +1. In o de o be able o conside he same pe iod h oughou he a icle, we e ain om doing so and use he dis ibu ion o yea . 8 Page 6 o 18 H.S übe e al. 2.3.4.2 SIAB‑compa able 2 Wo ke s conside ed: All wo ke s subjec o social insu ance con ibu ions. Sampling: Conside a ion o all employmen s. Wage o he calcula ion o wage inequali y: Sum o he nominal g oss pay olls o all employmen ela ionships o a pe son du ing he yea . 3 The de elopmen o wage inequali y inGe many 3.1 Inequali y measu es To analyze income inequali y and illus a e he di e - ence be ween su ey and adminis a i e da a, we ocus on wage pe cen iles and wage pe cen ile a ios. Pe cen- ile a ios a e widely used, e.g., by he OECD, and a e an in ui i e way o ep esen ing income inequali y.17 Pe cen ile a ios indica e he a io o he wages o wo indi iduals who a e in di e en posi ions in a gi en dis- ibu ion. Fo example, he P90/P10 a io compa es he wage a he 90 h pe cen ile wi h ha a he 10 h pe - cen ile. I he P90/P10 a io has inc eased o e a pe iod, his indica es ha inequali y be ween he op and he bo om ails o he wage dis ibu ion has inc eased. The disad an age o he P90/P10 a io is ha wage ends abo e he 90 h pe cen ile and below he 10 h pe cen ile a e no conside ed. Somewhe e abo e he 90 h pe cen- ile, howe e , is he wage h eshold abo e which employ- e s no longe epo de ailed wage in o ma ion, so ha inaccu acies can occu e en when wage in o ma ion is es ima ed. Thus, using he P90/P10 a io ci cum en s he unce ain y abo e he wage h eshold.18 Fo he pu poses o his a icle, we conside he ollowing h ee pe cen ile a ios: P90/P10, P90/P50, and P50/P10. 3.2 Using hes eng hs o su ey andadminis a i e da a Fi s , we measu e wage inequali y in bo h da a se s using he wo mos common wage measu es o he espec- i e da a se s: mon hly wage (SOEP-Pu e 1) and annual wage (SOEP-Pu e 2) o he SOEP and a e age daily wage (SIAB-Pu e 1) and daily wage on June 30 (SIAB- Pu e 2) o he SIAB. These a e he na i e measu es o he espec i e da a se , i.e., mon hly wages as well as annual wages a e collec ed di ec ly in he SOEP ques- ionnai e and wage in o ma ion a e p o ided as a e age daily wages in he SIAB. In he ollowing sec ion, we hen measu e wage inequali y using compa able sub-samples o he wo da a se s. The esul s based on mon hly and annual wages in he SOEP a e shown in Fig.1. Resul s based on a e age daily wage and daily wage a June 30 in he SIAB a e shown in Fig.2. Looking a he ajec o ies o he pe cen ile a ios in he wo da a se s, i is immedia ely appa en ha he P90/P10 alue is lowe in he SOEP han in he SIAB. The end o e he yea s also shows clea di e ences. In he SOEP, he mon hly P90/P10 pe cen ile a io is ela i ely s able un il 2016 and hen declines (see Fig. 1A).19 In con as , he annual P90/P10 pe cen ile a io shows a weak in e ed U-shape: i ises un il 2011 and hen alls again s a ing a ound 2013 (see Fig.1B). In he SIAB, he pa e n is much mo e igid. Howe e , a e y weak in e ed U-shape can be seen o he P90/ P10 pe cen ile a io o a e age daily wages (see Fig.2A). Compa ing he P50/P10 pe cen ile a ios, a simila pa - e n eme ges as o he P90/P10 a ios. In con as , he P90/P50 pe cen ile a ios a e qui e simila in all ou sam- ples: a nea ly cons an pa e n wi h a a io o abou wo. The ac ha he P90/P50 pe cen ile a ios a e qui e simila in all ou samples, bu he e a e majo di e ences in bo h he le el and end o he P90/P10 and P50/P10 pe cen ile a ios, sugges s ha he e a e di e ences in he da a se s a he lowe end o he wage dis ibu ion. The lowe P90/P10 and somewha lowe P50/P10 pe cen- ile a ios in he SOEP indica e ha he P10 alue in he SOEP mus be highe han ha in he SIAB. O cou se, his di e ence may be because he popula- ion co e ed by he wo da a se s is no ully compa able. Fo example, he SOEP includes wages o ci il se an s, which a e no included in he SIAB da a. In addi ion, he SOEP su eys do no ake place e enly h oughou he yea . The e o e, wages in he SOEP could be in luenced by seasonal e ec s. To examine whe he he di e ences desc ibed a e due o hese ac o s, we conside he compa able sub-samples 17 An al e na i e way o measu e inequali y would be o calcula e, e.g, Gini indices o he indi idual da a se s. Howe e , while he Gini index is only an agg ega e measu e, pe cen ile a ios p o ide in o ma ion on he shape o he unde lying dis ibu ion. 18 The 90 h pe cen iles a e no a ec ed by he impu a ion o wages. We impu e igh -censo ed wages so ha we can display he mean wages in he desc ip i e s a is ics in AppendixB2. The 90 h pe cen iles howe e would be a ec ed by impu ed wages i we we e o epo ou esul s di e en ia ed, eg., by gende and educa ion. As S übe e al. (2023,p. 6) no e: The “wage in o ma ion in he p ocess da a is op-coded, and hence we only obse e wages up o he social secu i y con ibu ion ceiling. While his ea u e only a ec s app oxima ely 5.2% o all employmen spells o wo ke s be ween 1975 and 2019, he p opo ion o censo ed obse a ions wi hin ce ain subg oups is subs an ial. Fo ins ance, nea ly 44% o he spells o egula ly employed male wo ke s wi h a deg ee om a uni e si y o uni e si y o applied science a e a ec ed by op-coding.” Howe e , looking a he SIAB- Pu e 2 sample, e.g., and no conside ing wages below he ma ginal ea nings h eshold, no yea has mo e han 90% o wages impu ed abo e he 90 h pe cen ile. 19 The end o a decline in mon hly wage inequali y e en began se e al yea s ea lie when using g oss hou ly wages (c . G abka 2022). Page 7 o 18 8 A ale o  wo da a se s: compa ing Ge man adminis a i e andsu ey da a using wage inequali y… in he ollowing sec ion. Howe e , ano he s a egy o unco e ing po en ial ac o s o explain he obse ed di - e ences be ween SOEP and SIAB may be o use o he da a sou ces wi h wage in o ma ion. The e o e, in u he analyses, we use he Fede al S a is ical O ice’s S uc u e o Ea nings Su ey (VSE, Ve diens s uk u e hebung) in AppendixC o gain u he insigh s. 3.3 Using compa able sub‑samples o su ey andadminis a i e da a In he ollowing, we conside he ou compa able sub- samples p esen ed in Sec .2.3.2. Figu es3 and4 show he compa able samples using mon hly and annual wages, espec i ely. The pe cen ile ajec o ies o mon hly wages (Fig.3A) look ela i ely simila in he wo da a se s. This is espe- cially ue o he 50 h pe cen ile. Howe e , he P10 o he SOEP is somewha highe han ha o he SIAB, especially om 2017 onwa d. On he o he hand, he P90 o he SIAB is consis en ly highe han ha o he SOEP. The pe cen ile ajec o ies o annual wages (Fig.4A) in he wo da a se s also look qui e simila . Howe e , he e he P10 o he SOEP is signi ican ly highe han he P10 o he SIAB, and one also sees clea di e ences in he wo P50 pe cen iles. The di e ences—in bo h mon hly and annual da a— ansla e in o s iking di e ences when pe cen ile a ios a e conside ed (see Figs.3B and4B). He e we can see ha he clea di e ences in pe cen ile a ios—wi h he excep- ion o P50/P10—s ill exis . These esul s sugges ha wages a he lowe (and uppe ) ends o he wage dis ibu ion a e cap u ed di e - en ly. I could be ha low wages a e unde ep esen ed in he SOEP. This could be due o sampling, bu i could also be due o esponden s no epo ing e y low wages o simply o ge ing abou hem. I could also be because in o ma ion on second jobs has been collec ed mo e p e- cisely in he SOEP since 2017. Ano he explana ion could be ha esponden s wi h se e al ma ginal pa - ime jobs a e e y likely o add hem oge he and hus epo highe wages as cap u ed in he SIAB. This las aspec migh be pa icula ly ele an , gi en ha he numbe o wo ke s wi h mo e han one job mo e han doubled be ween 1999 and 2019.20 In he pe iod unde conside a ion, low-income wages a e included in he adminis a i e da a (SIAB) because since 1999 jobs wi h wages below he ma ginal ea nings h eshold a e subjec o a lump-sum social secu i y con- ibu ion payable by he employe . Fo his eason, ma - ginal wages should be eco ded ela i ely eliably in he SIAB. He e, oo, howe e , i canno be uled ou ha , e.g., spells wi h low wages may occu due o subsequen wage decla a ions ha do no ep esen ac ual employ- men . Al hough hese a e adminis a i e da a, we obse e some employmen spells wi h un ealis ically low daily wages (e.g., ≤ 1 eu o)21. This could lead o an o e es ima- ion o ma ginal pa - ime employmen in he SIAB. To u he analyze his, we decided o add ano he es ic ion o ou compa able sub-samples below by lim- i ing he analyses o wages abo e he ma ginal ea nings h eshold.22 Fig. 1 De elopmen o wage pe cen iles a ios in he SOEP 2000–2019. Sou ce: own calcula ions, SOEP-Co e. 37.EU 20 Sou ce: h ps:// www. des a is. de/ DE/ Themen/ A bei / A bei sma k / Quali ae - A bei / Dimen sion-3/ zwei jobl. h m; las accessed: Oc . 23, 2022. 21 Less han 0.15% o wo ke s ha e a daily wage o ≤ 1 eu o (as o June 30); he 1% pe cen ile is 3.12 eu os and he 5% pe cen ile is 8.22 eu os. 22 To be e assess he di e ence be ween SIAB and SOEP highligh ed in his sec ion, we addi ionally ely on he VSE in AppendixC. 8 Page 8 o 18 H.S übe e al. 3.4 Using hema ginal ea nings h eshold asalowe wage loo Using he ma ginal ea nings h eshold23 as a wage loo , he esul s a e no longe ep esen a i e o he en i e Ge man labo o ce, bu only o wo ke s wi h wages abo e he ma ginal ea nings h eshold. Ne e heless, he pe cen ile a ios can be used o examine he de elopmen o wage inequali y in his g oup and o highligh di e - ences in he wo da a se s. The esul s o his exe cise a e s iking. Looking only a wages abo e he ma ginal ea nings h eshold, he end is now almos iden ical o all h ee pe cen iles conside ed (see Figs.5A and6A). Only in 2016 does he P90 in he SOEP dec ease sligh ly, bu he de elopmen in he ol- lowing yea s is again in line wi h ha o he SIAB. The ajec o ies o he pe cen ile a ios a e now also qui e simila (see Figs.5B and6B). Howe e , we s ill ind a le el di e ence in he P90/P10 a io o annual wages (see Fig.6B).24 4 Conclusion Two e y di e en da a se s a e o en used o desc ibe wage inequali y in Ge many. On he one hand, he adminis a i e da a o he SIAB and, on he o he hand, he in o ma ion ob ained om a popula ion su ey, he SOEP. Bo h da a sou ces ha e hei speci ic ad an ages and disad an ages. Using he ull in o ma ion om each da a sou ce yields compa able ends in wage inequali y om 2000 o abou 2015. Since hen, he esul s di e ge a i s glance. A e ha monizing he analysis popula ion, compa able ends and ela i ely simila le els eme ge again o bo h da a sou ces. Howe e , he compa ison o he wo da a sou ces also shows ha he e a e sys ema ic de ia ions be ween hem in he a ea o e y low wages. Va ious easons may be esponsible o his: On he su ey side, i migh be he case ha esponden s add up wages om se e al ac i i ies, so ha he SOEP shows highe wage le els in he lowe pa o he dis ibu ion. In addi ion, i canno be uled ou ha money lows om he employe o he employee o e and abo e he egula wage, e.g., o pay o sho - e m o e - ime in a mini-job, bu no o exceed he o icial mini-job h eshold. Acco dingly, he e a e signi ican ly mo e e y small wage epo s on he SIAB side. One o he easons o his migh be sho - e m egis a ions and cancella- ions o employmen ela ionships, which a e no consid- e ed ele an o esponden s and he e o e end no o be epo ed in su ey da a. In addi ion, he e is p esumably a p oblem on he pa o he SIAB, known om minimum wage esea ch: i he minimum wage legisla ion is no com- plied wi h o he mini-job limi is exceeded, elec onic pay- oll sys ems issue a wa ning message and wages o wo king hou s migh jus be adjus ed in he sys em in acco dance o he legal si ua ion (e.g., Bachmann e al. 2020,p. 26). This is likely o occu mo e equen ly in he lowe wage ange and in sho - e m employmen . O e all, howe e , he indings p esen ed he e show ha bo h da a sou ces a e well sui ed o adequa ely desc ibe wage inequali y in Ge many abo e he mini-job h eshold. Howe e , he obse ed di e ences a he bo om o he wage dis ibu ion equi e u he anal- ysis. The upcoming elease o SOEP da a linked o SIAB da a by social secu i y numbe (SOEP-ADIAB), planned o 2023, will p o ide a da abase helping o u he in es i- ga e in o hese di e ences. Fig. 2 De elopmen o wage pe cen iles a ios in he SIAB 2000–2019. Sou ce: own calcula ions, SIAB 7519 23 The daily, mon hly, and annual ma ginal ea nings h eshold a e p o ided a h p:// doku. iab. de/ dz/ Bemes sungs g enz en_ de_ en. xls; las accessed: Oc . 23, 2022. 24 In addi ion o he ma ginal ea nings h eshold, we also calcula ed he espec i e pe cen iles o al e na i e (lowe ) ea nings h esholds (e.g., 100 EUR and 200 EUR). Howe e , he e ec o esul s con e ging, was mos p onounced a he ma ginal ea nings h eshold. Page 15 o 18 8 A ale o  wo da a se s: compa ing Ge man adminis a i e andsu ey da a using wage inequali y… a mon hly basis is eco ded. Howe e , i is ques ionable whe he esponden s would epo one-day o e y sho employmen , especially i i is a second o i egula job. In he GMW esul s o he SOEP and he SIAB, only he job wi h he highes g oss mon hly wage is conside ed. In he VSE, he g oss nominal annual wages (GAW) o pa ial yea s we e ex apola ed o 12 mon hs. Only employmen ela ionships wi h 30 o mo e wo king weeks in 2014 o 2018 a e conside ed. In he GAW o he SOEP and he SIAB, on he o he hand, all jobs o a pe - son a e aken in o accoun . Compa ing he pe cen iles o he GMW o he VSE wi h he SOEP and he SIAB, i is no iceable ha he P10 and P50 in he SOEP a e highe han he alues in he VSE in bo h yea s. The P90, on he o he hand, is highe in bo h yea s in he VSE. In he SIAB, a di e - en pa e n eme ges. The P10 o he SIAB and he VSE a e ema kably close o each o he in bo h yea s. On he o he hand, he P50 and P90 o he SIAB a e always sig- ni ican ly highe han hose o he VSE. An explana ion o he highe P10 in he SOEP could be, ha esponden s wi h se e al ma ginal pa - ime jobs add hem oge he and hus epo highe wages han he o he wo sou ces. Compa ing he pe cen iles o he GAW o he VSE wi h hose o he SOEP, we ind a simila pa e n as o he GMW. Again, he P10 is highe in he SOEP han in he VSE, and he P90 is highe in he VSE. Bu he P50 is highe in 2018 only in he SOEP. Howe e , he e a e s onge de ia ions in he compa ison o SIAB and VSE. He e, all h ee pe cen iles o he SIAB a e signi ican ly lowe han hose o he VSE. The la ge de ia ions in he GAW can p obably be explained by he ac ha he SIAB includes all employmen , no ma - e how b ie . In o he wo ds, many people who a e pe haps only employed o a ew days a yea a e aken in o accoun he e and he e o e push he pe cen ile alues down. In he VSE, on he o he hand, hese sho jobs a e no aken in o accoun a all. I also canno be uled ou ha companies wi h mo e ma ginal pa - ime employees a e unde ep e- sen ed due o he composi ion o he companies su eyed. In addi ion, he VSE eco ds wages in Ap il and he e o e seasonal employmen is no eco ded in a ep esen a i e manne . Also in he SOEP low wages, especially hose due o sho - ime jobs, could be unde ep esen ed. This may be due o sampling, bu also because esponden s did no epo e y low wages o simply o ge abou e y sho - e m employmen . In he pe iod we conside , ma ginal pa - ime wages a e included in he SIAB, because since 1999 wages below he ma ginal ea nings h eshold jobs a e subjec o a lump-sum con ibu ion payable by he employe . Fo his eason, ma ginal pa - ime wages should be eco ded ela i ely eliably in he SIAB. He e, oo, howe e , i can- no be uled ou ha , e.g. due o subsequen epo ing o wages, spells wi h low wages occu ha do no ep esen ac ual employmen . Al hough hese a e adminis a i e da a, he e a e obse a ions wi h un ealis ically low daily wages. This could lead o an o e es ima ion o ma ginal pa - ime employmen —and hence an unde es ima ed P10—in he SIAB. To shed mo e ligh on he impac o he VSE con- s ain s jus men ioned, we show in Table10 wha hap- pens when we apply hese cons ain s o he SIAB. The g oss nominal mon hly wage in Ap il (GMW) o he VSE and he g oss nominal wage (GMW, SIAB-Compa able 1) co espond o he alues in Table9; hey a e only lis ed again o he sake o cla i y. GMW2 o GMW5 show wha happens when es ic ions o he VSE a e applied o he SIAB. GMW2 co esponds o GMW o he SIAB, excep o he ac ha he wo ke s mus be employed o he en i e mon h—as in he VSE—in o de o be aken in o accoun . As a esul , he P10, he P50 and he P90 inc ease. In addi ion, es ic ing he SIAB o he mon h Ap il—as in he VSE—causes all pe cen iles o all again (see GMW3). Howe e , wi h he excep ion o he P10 in 2018, hey a e abo e GMW’s alues. GMW4 and GMW5 co espond o GMW2 and GMW3, espec i ely, excep ha we now exclude employmen in sec o s T and U—as in he VSE. As can be seen om Table10, his addi ional cons ain has no e ec on P10, bu P50 and P90 inc ease sligh ly compa ed o GMW2 and GMW3. Howe e , he gene al pa e n emains: The P10 pe cen iles a e close, bu he P50 and P90 pe cen iles a e highe in he SIAB, and his di e ence has inc eased o e ime. I should be no ed ha i is e iden ha he VSE es ic- ions ha e an impac on he wage pe cen ile(s) ( a ios). Howe e , hey do no explain he la ge di e ence, o Table 5 Summa y s a is ics SIAB-Pu e 1 sample Sou ce:own calcula ions, SIAB 7519, yea s 2000–2019 Wage: nominal daily wage in eu o, mean o e (median o ) all obse a ions o he sample. Age: mean age in yea s Only wages abo e he ma ginal ea nings h eshold a e aken in o accoun No. o obse a ions Mean wage Median wage Mean age 11,492,264 90.44 76.15 41.77 Table 6 Summa y s a is ics SIAB-Pu e 2 sample Sou ce:own calcula ions, SIAB 7519, yea s 2000–2019 Wage: nominal daily wage in eu o, mean o e (median o ) all obse a ions o he sample. Age: mean age in yea s Only wages abo e he ma ginal ea nings h eshold a e aken in o accoun No. o obse a ions Mean wage Median wage Mean age 10,376,844 96.59 82.38 42.41 8 Page 16 o 18 H.S übe e al. abou 10%, in he P90 be ween VSE and SIAB. This is no ewo hy in ha bo h wage in o ma ion a e employe decla a ions whe e a high deg ee o accu acy should be expec ed. Wha could s ill be in es iga ed is whe he o o wha ex en d awing a company sample has an e ec on he wage inequali y ound. Since such an e alua ion is no possible wi h SIAB, we lea e his o u u e esea ch.30 Table 7 Summa y s a is ics SIAB: compa able sample 1 Sou ce: own calcula ions, SIAB 7519, yea s 2000–2019 Wage: nominal mon hly wage in eu o, mean o e (median o ) all obse a ions o he sample. Age: mean age in yea s. Wo king: - = ull- ime; p- = pa - ime. Educa ion: low = no oca ional aining; medium = oca ional aining; high = uni e si y (o applied sciences) Only wages abo e he ma ginal ea nings h eshold a e aken in o accoun Due o missings in wo king- ime and educa ion, pe cen ages do no sum up o 100% No. o Wage (in eu o) Mean Wo king (in %) Educa ion (in %) obse a ions mean median age ‑ p‑ low medium high 9,859,452 2905.86 2478.76 42.65 76.34 23.66 7.85 73.82 16.52 Table 8 Summa y s a is ics SIAB, compa able sample 2 Sou ce:wn calcula ions, SIAB 7519, yea s 2000–2019 Wage: nominal annual wage in eu o, mean o e (median o ) all obse a ions o he sample. Age: mean age in yea s Only wages abo e he ma ginal ea nings h eshold a e aken in o accoun No. o obse a ions Mean wage Median wage Mean age 10,808,103 33,353.38 28,406.06 42.13 Table 9 Selec ed 2014 and 2018 wage pe cen iles and wage pe cen ile a ios o VSE, SOEP, and SIAB VSE S uc u e o Ea nings Su ey. GMW: g oss nominal mon hly wage in Ap il. GAW: g oss nominal annual wage. To al wi hou app en ices. Economic sec ions A-S o he Classi ica ion o economic ac i i ies. Thus, no conside ed is employmen in sec o s T (P i a e households wi h employed pe sons) and U (Ex a e i o ial o ganiza ions and bodies). Sou ce: S a is isches Bundesam (2016, 2020) SOEP Socio-Economic Panel (SOEP-Co e. 37.EU, own calcula ions). GMW: g oss nominal mon hly wage (SOEP-Pu e 1). GAW: g oss nominal annual wage (SOEP-Pu e 2) SIAB Sample o In eg a ed Labou Ma ke Biog aphies (SIAB 7519, own calcula ions). GMW: g oss nominal wage (SIAB-Compa able 1); GAW: g oss nominal annual wage (SIAB-Compa able 2) All wages in eu o Da a se VSE SOEP SIAB Wage measu e GMW GAW GMW GAW GMW GAW Pe cen iles Yea 2014 P10 396 4800 450 5400 395 2787 P50 2269 29,697 2301 29,600 2355 24,636 P90 4707 62,939 4668 61,800 5109 58,597 Pe cen ile a ios Yea 2014 P90/P10 11.89 13.11 10.37 11.44 12.95 21.02 P50/P10 5.73 6.19 5.11 5.48 5.97 8.84 P90/P50 2.07 2.12 2.03 2.09 2.17 2.38 Pe cen iles Yea 2018 P10 438 5337 720 7800 448 3614 P50 2500 32,714 2600 33,000 2621 27,766 P90 5151 68,508 5000 66,097 5633 64,686 Pe cen ile a ios Yea 2018 P90/P10 11.76 12.84 6.94 8.47 12.57 17.90 P50/P10 5.71 6.13 3.61 4.23 5.85 7.68 P90/P50 2.06 2.09 1.92 2.00 2.15 2.33 30 I would be concei able, e.g., o use he SIEED—a 1.5% sample o all es ab- lishmen s in Ge many (see Schmid lein e al. 2021). Fo hese es ablishmen s, he en i e employmen his o ies o he pe sons employed in hese es ablish- men s is a ailable. Howe e , he SIEED would ha e he disad an age ha he da a a e only a ailable a mid-yea ( e e ence da e 30.6). In p inciple, how- e e , he SIEED would make i possible o show wha happens when di e - en es ablishmen s a e d awn and he a ious measu es o inequali y a e hen calcula ed on he basis o he associa ed employmen ela ionships. Page 17 o 18 8 A ale o  wo da a se s: compa ing Ge man adminis a i e andsu ey da a using wage inequali y… Acknowledgemen s We would like o hank Ma is Beckmannshagen and Joachim Mölle o e y use ul commen s and sugges ions. We would also like o hank Be nd Fi zen- be ge and pa icipan s o he 2016 IAB-IWH Wo kshop and he 2014 SOEP Use Con e ence o use ul commen s on an ea lie e sion o he p ojec . All e o s emain ou own. Au ho con ibu ions The au ho s ead and app o ed he inal manusc ip . Funding This a icle was w i en when Heiko S übe was an employee o he IAB, Ma kus M. G abka was an employee o he DIW Be lin/SOEP, and Daniel D. Schni zlein was an employee o he Leibniz Uni e si ä Hanno e . The au ho s did no ecei e any special unding o his publica ion. A ailabili y o da a and ma e ials The SIAB 7519 is no publicly accessible o da a p o ec ion easons (social secu i y da a). Howe e , he Resea ch Da a Cen e (FDZ) o he Fede al Employmen Agency (BA) a he Ins i u e o Employmen Resea ch (IAB) makes his da a se a ailable o he scien i ic communi y. The SOEP-Co e. 37. EU can be o de ed o scien i ic pu poses ia he esea ch da a cen e o he SOEP a DIW Be lin. De ails on he p ocedu e can be ound a h ps:// www. diw. de/ en/ diw_ 01.c. 601584. en/ da a_ access. h ml. Decla a ions Compe ing in e es s The au ho s decla e ha hey ha e no compe ing in e es s. Recei ed: 23 Decembe 2022 Accep ed: 19 Janua y 2023 Re e ences Bachmann, R., Boockmann, B., Gonscho , M., Kalwei , R., Klause , R., Laub, N., Rul , C., Vonnahme, C., Zimpelmann, C.: Auswi kungen des gese zlichen Mindes lohns au Löhne und A bei szei en, IZA Resea ch Repo 96 (2020) Bakke , B.F.: Es ima ing he alidi y o adminis a i e a iables. S a . Nee l. 66(1), 8–17 (2011) Ba olucci, C.: Gende wage gaps econside ed a s uc u al app oach using ma ched employe -employee da a. J. Hum. Resou . 48(4), 998–1034 (2013) Beaule, A., Campbell, F., Ma y, D., Insole a, N., Johnson, D., Juska, P., McGonagle, K., Simme , B., Wa a, J.: PSID main in e iew use manual: elease 2021, Ins i u e o Social Resea ch, Uni e si y o Michigan (2021) Becke , I., F ick, J.R., G abka, M.M., Hause , R., K ause, P., Wagne , G.G.: A compa i- son o he main household income su eys o Ge many: EVS and SOEP. In: Hause , R., Becke , I. (eds.) Repo ing on income dis ibu ion and po e y, pp. 55–90. Sp inge , Be lin (2003) Beckmannshagen, M., Sch öde , C.: Ea nings inequali y and wo king hou s misma ch. Labou Econ. 76, 102184 (2022) Bende , S., Hilzendegen, J., Rohwe , G., Rudolph, H.: Die IAB-Beschä ig ens ich- p obe 1975–1990, Bei äge zu A bei sma k - und Be u s o schung 197 (1996) Biewen, M.: Income inequali y in Ge many du ing he 1980s and 1990s. Re . Income Weal h 46(1), 1–19 (2000) Biewen, M., Juhasz, A.: Unde s anding ising income inequali y in Ge many, 1999/2000-2005/2006. Re . Income Weal h 58, 622–647 (2012) Blough, D.K., Ramsey, S, Sulli an, S.D., Yusen, R: The impac o using di e en impu a ion me hods o missing quali y o li e sco es on he es ima ion Table 10 Selec ed 2014 and 2018 wage pe cen iles and wage pe cen ile a ios o VSE and SIAB VSE: S uc u e o Ea nings Su ey. GMW: g oss nominal mon hly wage in Ap il. To al wi hou app en ices. Economic sec ions A-S o he Classi ica ion o economic ac i i ies. Thus, no conside ed is employmen in sec o s T (P i a e households wi h employed pe sons) and U (Ex a e i o ial o ganiza ions and bodies). Sou ce: S a is isches Bundesam (2016, 2020) SIAB: Sample o In eg a ed Labou Ma ke Biog aphies (SIAB 7519, own calcula ions). GMW: g oss nominal wage (SIAB-Compa able 1); GMW2: g oss nominal wage (SIAB-Compa able 1) bu wo ke s mus wo k en i e mon h (as in VSE); GMW3: g oss nominal wage in Ap il and wo ke s mus wo k en i e mon h (as in VSE); GMW4: g oss nominal wage (SIAB-Compa able 1) bu wo ke s mus wo k en i e mon h (as in VSE), no conside ing sec o s T and U (as in VSE); GMW5: g oss nominal wage in Ap il and wo ke s mus wo k en i e mon h (as in VSE), no conside ing sec o s T and U (as in VSE) All wages in eu o Da a se VSE SIAB Wage measu e GMW GMW GMW2 GMW3 GMW4 GMW5 Pe cen iles Yea 2014 P10 396 395 407 397 407 397 P50 2269 2355 2400 2388 2402 2390 P90 4707 5109 5153 5127 5157 5130 Pe cen ile a ios Yea 2014 P90/P10 11.89 12.95 12.66 12.92 12.67 12.92 P50/P10 5.73 5.97 5.90 6.02 5.90 6.02 P90/P50 2.07 2.17 2.15 2.15 2.15 2.15 Pe cen iles Yea 2018 P10 438 448 458 444 458 444 P50 2500 2621 2670 2659 2672 2661 P90 5151 5633 5680 5648 5683 5651 Pe cen ile a ios Yea 2018 P90/P10 11.76 12.57 12.39 12.73 12.40 12.74 P50/P10 5.71 5.85 5.82 5.99 5.83 6.00 P90/P50 2.06 2.15 2.24 2.12 2.13 2.12 8 Page 18 o 18 H.S übe e al. o he cos -e ec i eness o lung- olume- educ ion su ge y. Heal h Econ. 18(1), 91–101 (2009) Bollinge , C.R.: Measu emen e o in he cu en popula ion su ey: a nonpa a- me ic look. J. Labo Econ. 16(4), 576–594 (1998) Bound, J., B own, C., Duncan, G.J., Rodge s, W.L.: E idence on he alidi y o c oss-sec ional and longi udinal labo ma ke da a. J. Labo Econ. 12(3), 345–368 (1994) Bound, J., K uege , A. B.: The ex en o measu emen e o in longi udinal ea nings da a: do wo w ongs make a igh ? J. Labo Econ. 9(1), 1–24 (1991) Bound, J., B own, C., Ma hiowe z, N.: Chap e 59-measu emen e o in su ey da a. In: Heckman, J.J., Leame , E.E. (eds.) Handbook o econome ics, ol. 5, 1s edn., pp. 3705–3843. Else ie , Ams e dam (2001) B icke , J., Engelha d , G.V.: Measu emen e o in ea nings da a in he heal h and e i emen s udy. J. Econ. Soc. Meas. 33(1), 39–61 (2008) B i zke, J., Schupp, J.: SOEP wa e epo 2018. SOEP, Be lin (2019) Bundes egie ung: Be ich de Bundes egie ung zu Lebensquali ä in Deu schland. Bundes egie ung, Be lin (2016) Bundes egie ung: Lebenslagen in Deu schland - De Sechs e A mu s- und Reich- umsbe ich de Bundes egie ung. Bundes egie ung, Be lin (2021) Bu auel, P., Caliendo, M., G abka, M.M., Obs , C., P euss, M., Sch öde , C., Shupe, C.: The impac o he ge man minimum wage on indi idual wages and mon hly ea nings. Jah b Na l ökon S a 240(2–3), 201–231 (2020) Caliendo, M., Wi b od , L.: Did he minimum wage educe he gende wage gap in Ge many? Labou Econ. 78, 102228 (2022) Ca d, D., Heining, J., Kline, P.: Wo kplace he e ogenei y and he ise o Wes Ge man wage inequali y. Qua . J. Econ. 128(3), 967–1015 (2013) Ca pen e, H.: UK household longi udinal s udy—wa e 11 echnical epo . The Ins i- u e o Social and Economic Resea ch, Uni e si y o Essex, Colches e (2021) Code , J.: Using adminis a i e eco d in o ma ion o e alua e he quali y o he income da a collec ed in he su ey o income and p og am pa icipa ion. P oc. S a . Can. Symp. 92, 295–306 (1992) de Leeuw, E.D., Hox, J.J., Dillman, D.A.: Chap e 1- he co ne s ones o su ey esea ch. In: Hox, J., Leeuw, E.D., Dillman, D. (eds.) In e na ional Handbook o Su ey Me hodology, pp. 1–17. Eu opean Associa ion o Me hodology, F ank u (2008) Dus mann, C., Fi zenbe ge , B., Schönbe g, U a, Spi z-Oene , A.: F om sick man o Eu ope o economic supe s a : Ge many’s esu gen economy. J. Econ. Pe spec . 28(1), 167–188 (2014) Dus mann, C., Fi zenbe ge , B., Schönbe g, U., Spi z-Oene , A., Luds eck, J., Schönbe g, U.: Re isi ing he Ge man wage s uc u e. Qua . J. Econ. 124(2), 843–881 (2009) Felbe may , G., Baumga en, D., Lehwald, S.: Inc easing wage inequali y in Ge many — wha ole does global ade play? Be elsmann S i ung (2014) Fi zenbe ge , B., Seidli z, A.: The 2011 b eak in he pa - ime indica o and he e olu ion o wage inequali y in Ge many. J. Labou Ma k. Res. 54, 1 (2020) Fi zenbe ge , B., Seidli z, A., de Lazze , J.: Changing selec ion in o ull- ime wo k and i s e ec on wage inequali y in Ge many. Empi . Econ. 62, 247–277 (2022) F ick, J.R., G abka, M.M.: I em non esponse on income ques ions in panel su eys: incidence, impu a ion and he impac on inequali y and mobili y. Allg. S a . A ch. 89, 49–61 (2005) F ode mann, C., Schmucke , A., Se h, S., om Be ge, P.: Sample o In eg a ed Labou Ma ke Biog aphies (SIAB) 1975–2019. FDZ-Da en epo 01/2021 (en) (2021) Ganze , A., Lisa S., Jens S., Wol e S.: Es ablishmen His o y Panel 1975–2018. FDZ- Da en epo 01/2020 (en), e ised e sion ( 2) om Ap il 2021 (2020) Ge nand , J., P ei e , F.: Rising wage inequali y in Ge many. Jah b. Na l. ökon. S a . 227(4), 358–380 (2007) Goebel, J., G abka, M.M., Liebig, S., K oh, M., Rich e , D., Sch öde , C., Schupp, Jü gen.: The Ge man Socio-Economic Panel (SOEP). Jah b. Na l. ökon. S a . 239(2), 345–360 (2019) G abka, M.M.: Income inequali y in Ge many s agna ing o e he long e m, bu dec easing sligh ly du ing he co ona i us pandemic. DIW Wkly. Rep. 17(18), 125–133 (2021) G abka, M.M.: Löhne, Ren en und Haushal seinkommen sind in den e gangenen 25 Jah en eal ges iegen. DIW Wochenbe . 89(23), 329–337 (2022) G abka, M.M., Sch öde , C.: Inequali y in Ge many: dec ease in gap o g oss hou ly wages since 2014, bu mon hly and annual wages emain on pla eau. DIW Wkly. Rep. 8(9), 83–92 (2018) G oen, J.A.: Sou ces o e o in su ey and adminis a i e da a: he impo ance o epo ing p ocedu es. J. O . S a . 28(2), 173–198 (2012) Hause , R., Becke , I., G abka, M.M., Wes e heide, P.: In eg ie e Analyse de Einkom- mens- und Ve mögens e eilung: Abschlussbe ich zu S udie im Au ag des Bundesminis e iums ü A bei und Soziales, Bonn, Vol. A369 o Fo schungs- be ich . Bundesminis e ium ü A bei und Soziales, Be lin (2007) Johansson, F., Kle ma ken, A.: Compa ing egis e and su ey weal h da a. In . J. Mic osimula ion 15(1), 1 (2022) Kap eyn, A., Ypma, JY.: Measu emen e o and misclassi ica ion: a compa ison o su ey and adminis a i e da a. J. Labo Econ. 25(3), 513–551 (2007) Ka onius, I.K., Tö mäleh o, V.M.: Household income agg ega es in mic o and mac o s a is ics. S a . J. U. N. Econ. Comm. Eu . 20(1), 9–25 (2003) Klein, M.W., Mose , C., U ban, D.M.: Expo ing, skills and wage inequali y. Labou Econ. 25, 76–85 (2013) Kopczuk, W., Saez, E., Song, J.: Ea nings inequali y and mobili y in he Uni ed S a es: e idence om social secu i y da a since 1937. Qua . J. Econ. 125(1), 91–128 (2010) Lindne , P., And easch, M.: Mic o and mac o da a: a compa ison o he household inance and consump ion su ey wi h inancial accoun s in Aus ia. J. O . S a . (2014). h ps:// doi. o g/ 10. 1515/ jos- 2016- 0001 Li le, R., Su, H.L.: I em non- esponse in panel su ey. In: Kasp zyk, D., Duncan, G., Kal on, G., Singh, M.P. (eds.) Panel su eys, pp. 400–425. Wiley, New Yo k (1989) Lü hen, H., Sch öde , C., G abka, M.M., Goebel, J., Mika, J., B üggmann, D., Elle , S., Penz, H.: SOEP-RV: linking Ge man Socio-Economic Panel da a o pension eco ds. Jah b. Na l. ökon. S a . 242(2), 291–307 (2022) Mölle , J.: Lohnungleichhei —Gib es eine T endwende? Wi scha sdiens 96(13), 38–44 (2016) Obe ski, D. L., Ki chne , A., Eckman, S., K eu e , F.: E alua ing he quali y o su ey and adminis a i e da a wi h gene alized mul i ai -mul ime hod models. J. Am. S a . Assoc. 112(520), 1477–1489 (2017) OECD: A b oken social ele a o ? How o p omo e social mobili y. OECD Publishing, Pa is (2018) Pedace, R., Ba es, N.: Using adminis a i e eco ds o assess ea nings epo ing e o in he su ey o income and p og am pa icipa ion. J. Econ. Soc. Meas. 26(3–4), 173–192 (2000) Pischke, J.S.: Indi idual income, incomple e in o ma ion, and agg ega e consump- ion. Econome ica 63(4), 805–840 (1995) Roeme , M.: Using adminis a i e ea nings eco ds o assess wage da a quali y in he ma ch cu en popula ion su ey and he su ey o income and p og am pa icipa ion. Longi udinal employe -household dynamics echnical pape s, Cen e o Economic S udies. U.S, Census Bu eau (2002) Schmid lein, L., Se h, S., om Be ge, P.: Sample o In eg a ed Employe -Employee Da a 1975–2018. FDZ-Da en epo 14/2020 (en) (2021) Sch öde , C., König, J., Fedo e s, A., Goebel, J., G abka, M.M., Lü hen, H., Me zing, M., Schiko a, F., Liebig, S.: The economic esea ch po en ials o he Ge man Socio- Economic Panel s udy. Ge man Econ. Re . 21(3), 335–371 (2020) Selezne a, E., Ke m, P.V.: A dis ibu ion-sensi i e examina ion o he gende wage gap in Ge many. J. Econ. Inequal. 14(1), 21–40 (2016) Somme eld, K.: Highe and highe ? Pe o mance pay and wage inequali y in Ge many. Appl. Econ. 45(30), 4236–4247 (2013) S a is isches Bundesam . Ve diens s uk u e hebung 2014. Fachse ie 16, He 1 (published on 14.09.2016, co ec ed on 10.03.2017) (2016) S a is isches Bundesam . Ve diens s uk u e hebung 2018. Fachse ie 16, He 1 (published on 14.09.2020, upda ed on 06.10.2020) (2020) S eine , V., Wagne , K.: Has ea nings inequali y in Ge many changed in he 1980’s? Zei sch i ü Wi scha s- und Sozialwissenscha 118, 29–54 (1998) S ockhausen, M., Calde ón, M.: IW-Ve eilungs epo : S abile Ve häl nisse o z gewachsene gesellscha liche He aus o de ungen. IW-Repo 8 (2020) S übe H., Dau h W., Eppelsheime J.: A guide o p epa ing he Sample o In eg a ed Labou Ma ke Biog aphies (SIAB, e sion 7519 1) o scien- i ic analysis. J Labou Ma k Res. 57, 7 (2023). h ps:// doi. o g/ 10. 1186/ s12651- 023- 00335-w Tö mäleh o, V.M.: LIS and na ional accoun s compa ison. LIS Technical Wo king Pape Se ies 2 (2011) Ty owicz, J., an de Velde, L., an S a e en, I.: Does age exace ba e he gende - wage gap? New me hod and e idence om Ge many, 1984–2014. Feminis Econ. 24(4), 108–130 (2018) Publishe ’s No e Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in pub- lished maps and ins i u ional a ilia ions.