Public-Private Wage Differentials in Ireland, 1994 - 2001

Public-P i a e Wage Di e en ials in I eland, 1994-2001 By Ge y Boyle, Ro y McElligo and Jim O’Lea y * Na ional Uni e si y o I eland, Maynoo h Co esponding au ho : P o esso Ge y Boyle, [email p o ec ed] Abs ac A e public sec o wo ke s in I eland paid mo e han p i a e sec o employees, when such di e ences in p oduc i i y- ela ed pe sonal a ibu es and job cha ac e is ics a e con olled o ? We es ima e ha in 2001 he p emium enjoyed by public se an s was abou 13 pe cen . We ind ha he p emium, is signi ican ly bigge o hose nea he bo om o he ea nings dis ibu ion han o hose nea he op, was signi ican ly bigge o women han men in he mid-1990s bu no a he end o he 1990s, and does no a y signi ican ly ac oss di e en le els o educa ional a ainmen . We es ima e he p emium o 2001 o be no signi ican ly di e en om ha es ima ed o 1994 despi e his pe iod a pe iod o excep ionally apid ou pu and employmen g ow h, and co espondingly sha p igh ening o labou ma ke condi ions in he I ish economy. The mos ema kable di e ence be ween ou esul s and hose o o he esea che s o o he coun ies ela es o he absolu e size o he p emium. A numbe o possible explana ions o his di e ence a e discussed. * An ea lie e sion o his pape was p esen ed a he I ish Economic Associa ion Annual Con e ence in Bel as in Ap il 2004. The au ho s would like o acknowledge he help ul commen s and assis ance o hei colleagues Denis Conni e and Donal O’Neill and pa icipan s a a depa men al semina a NUI Maynoo h. Any e o s a e ou own esponsibili y. JEL codes: J31, J45 Key wo ds: public, p i a e, wage, di e en ials, I eland 2 Public-P i a e Wage Di e en ials in I eland, 1994-2001 1. In oduc ion Public sec o pay is a ma e o conside able policy in e es . A he mac oeconomic le el, he public sec o pay bill is a la ge elemen o go e nmen expendi u e, accoun ing o up o 70 pe cen o gene al go e nmen consump ion spending in some OECD coun ies (see Ju ges (2001)). As such, public sec o pay is an impo an d i e o he o e all ax bu den. A he mic oeconomic le el, a es o pay in he public sec o ha e an impo an bea ing on he ec ui men , e en ion and mo i a ion o public se an s and, by ex ension, on he quali y o public se ices. A he same ime, public sec o pay a es may in luence a es o pay in he p i a e sec o and in his way oo a ec he in e na ional compe i i eness o he economy. Re lec ing hese conce ns, he design o public sec o pay de e mina ion sys ems is an ongoing p eoccupa ion o policy- make s. The challenge is o iden i y and implemen mechanisms ha a e capable o deli e ing ou comes ha mee bo h e iciency and equi y c i e ia. Consequen ly, ecen de elopmen s in his ield, in I eland and elsewhe e, ha e sough o ins i u e a angemen s ha link a es o pay in he public sec o mo e di ec ly o p i a e sec o a es. In andem wi h hese de elopmen s, an in e na ional body o esea ch has eme ged which ocuses on he ope a ion o public sec o labou ma ke s. An impo an s and o ha esea ch has sough o quan i y public-p i a e sec o wage di e en ials, o explo e hei pa e n ac oss he ea nings dis ibu ion, educa ional a ainmen le els and occupa ions, and o analyse hei mo emen o e ime. 3 This pape is pa o ha s and o esea ch and, as a as me hodology is conce ned, d aws i s inspi a ion om i . To ha ex en , i is conce ned wi h simply applying a well- ied se o econome ic echniques o he analysis o public-p i a e sec o ea nings di e en ials in I eland, ocusing on he pe iod om 1994 o 2001. Howe e , he e a e aspec s o he I ish expe ience, speci ically i s economic pe o mance du ing he 1990s and ins i u ional ea u es o i s pay ba gaining sys em, ha make an ex ension o his esea ch o he I ish case a ma e o especial in e es . The pape is o ganised as ollows. In Sec ion 2 we desc ibe he economic and ins i u ional backg ound agains which public sec o pay has e ol ed in I eland o e he pas decade o so. In Sec ion 3 we discuss some o he con ibu ions o he ecen in e na ional li e a u e on public-p i a e sec o ea nings di e en ials. In Sec ion 4 we desc ibe he da ase we use o analyse public-p i a e sec o wage di e en ials in I eland. In Sec ion 5 we epo ou main esul s, and in Sec ion 6 we summa ise ou main indings and d aw some conclusions. 2. The I ish Con ex The I ish economy expe ienced apid g ow h o ou pu and employmen du ing he 1990s. Be ween 1993 and 2001 eal GDP doubled while GNP inc eased a an a e age annual a e o almos 8 pe cen in eal e ms. O e he same pe iod, o al employmen in he economy inc eased by a cumula i e 45 pe cen , o a an annual a e age a e o almos 5 pe cen (Table 1). The p incipal coun e pa o his igo ous expansion o employmen was an inc ease in he labou o ce which, hanks o a combina ion o s ong na u al inc ease, ising emale pa icipa ion a es and he eme gence o ne inwa d mig a ion, ose by 3 pe cen pe annum on a e age be ween 1993 and 2001 (Table 1). Unemploymen also declined sha ply, wi h he unemploymen a e alling om 15.7 pe cen in 1993 o 3.6 pe cen in ea ly 2001. Table 1: Labou o ce ends, I eland, 1993 and 2001 4 (000s) 1993 2001 To al Employmen 1,183 1,717 Unemploymen 220 65 Labou Fo ce 1,403 1,782 Unemploymen Ra e (%) 15.7 3.6 Sou ce: Budge a y and Economic S a is ics, Depa men o Finance, I eland. Some o he c edi o I eland’s economic pe o mance du ing he 1990s has been gi en o he so-called sys em o “social pa ne ship” ha has been an in eg al pa o he coun y’s ins i u ional a chi ec u e h oughou he pe iod. This sys em b ings oge he he go e nmen and o ganisa ions ep esen ing employees and employe s in a policy-o ien ed dialogue, he pu pose o which is o ind common g ound in ela ion o social and economic objec i es and he means o a aining hem. The main ou come o he sys em has been a se ies o h ee-yea p og ammes se ing ou ag eed policies ac oss a ange o issues1. A he co e o each o hese pa ne ship p og ammes has been a cen alised wage ag eemen , cha ac e ised by b oadly simila pay inc eases ac oss he public and p i a e sec o s. Go e nmen has pa icipa ed in hese ag eemen s, no only in i s capaci y as employe , bu also in i s ole as iscal au ho i y. This has allowed he wage ag eemen s o inco po a e commi men s by go e nmen in ela ion o income ax educ ions. In simple e ms, he logic behind he p oposi ion ha social pa ne ship played a key ole in he expansion o he I ish economy du ing he Cel ic Tige e a uns as ollows: pa ne ship pe mi ed ax educ ions which acili a ed wage mode a ion which in u n imp o ed cos compe i i eness and boos ed ou pu and employmen . O cou se, p oponen s o he social-pa ne ship model claim o he condui s o causa ion as well, such as he posi i e e ec s o pa ne ship on he indus ial ela ions clima e 1 These p og ammes and he pe iods o which hey applied we e as ollows. The P og amme o Na ional Reco e y (1988-1990); The P og amme o Economic and Social P og ess (1991-1993); The P og amme o Compe i i eness and Wo k (1994- 1996); Pa ne ship 2000 (1997-2000); and The P og amme o P ospe i y and Fai ness (2000-2002). The cu en p og amme, Sus aining P og ess, co e s he pe iod 2003- 2005. All o hese p og ammes a e published by Go e nmen Publica ions, S a ione y O ice, Dublin. 5 and i s con ibu ion o he p edic abili y o labou cos s. The poin is ha , wha e e he condui and wha e e he logic, he belie ha social pa ne ship is a aluable pe o mance-enhancing ins i u ion is a e y s ong one in policy-making ci cles2. Undoub edly, educ ions in income ax, h ough hei e ec on labou supply, played a ole in ensu ing low a es o wage in la ion in he I ish economy, a leas in he ea ly Cel ic Tige yea s. Howe e , by he la e 1990s, igh ening labou ma ke condi ions had s a ed o exe clea upwa d p essu e on wage se lemen s. Inc easingly, wage inc eases exceeded he no m se ou in he cen alised ag eemen . Fo example, Pa ne ship 2000 – he p og amme ha co e ed he pe iod om 1997 h ough 1999 – p o ided o a e age annual inc eases o 2.2 pe cen in basic pay ac oss he public and p i a e sec o s, bu by 1999 ac ual inc eases in ea nings ac oss many a eas o he economy we e unning a o e 6 pe cen . Amongs cons uc ion wo ke s, wage in la ion was unning a double-digi a es by 1999. Agains his backg ound a numbe o conce ns eme ged on he pa o public sec o employees and employe s. One was ha public sec o wo ke s we e being le behind in pay e ms by hei p i a e sec o coun e pa s. This pe cep ion appea s o ha e been based mo e on anecdo al e idence han on he a ailable s a is ical e idence3. None heless, i helped o c ea e a sense o g ie ance amongs public sec o wo ke s, and also aised conce ns abou he 2 A ai ly ypical iew is he ollowing om he Depu y P ime Minis e and Minis e o En e p ise, T ade and Employmen , Ms. Ma y Ha ney T.D.: “Social pa ne ship has been a majo ac o unde pinning I eland’s economic and social p og ess since he la e 1980s. I has played a i al ole in inc easing employmen … I has helped deli e sus ained and sus ainable inc eases in ou s anda ds o li ing. Abo e all, i has con ibu ed o ou dynamism and ou con idence ha we can success ully ackle common p oblems.” (Speech o Manpowe I eland G oup Eu opean Business Leade s’ B eak as , Dublin, Ap il 2002.) 3 I ish CSO da a show ha be ween 1996 and 2000, o example, a e age weekly ea nings inc eased a an annual a e age a e o 4.9 pe cen o public sec o wo ke s. The equi alen a es o inc ease o indus ial wo ke s and employees o banks, insu ance companies and building socie ies we e 5 pe cen and 4.1 pe cen espec i ely. All o hese g oups expe ienced a lowe a es o inc ease han cons uc ion wo ke s: be ween 1996 and 2000 skilled cons uc ion wo ke s enjoyed a e age weekly ea nings’ g ow h o o e 10 pe cen pe annum. 6 capaci y o o ganisa ions in he public se ice o ec ui , e ain and mo i a e s a . A he same ime he e was g owing disenchan men wi h he exis ing public sec o pay de e mina ion sys em. This was a sys em ha heo e ically p o ided o mos a es o pay o be benchma ked agains he p i a e sec o , wi h ela i i ies wi hin he public sec o supposedly playing a modes suppo ing ole. In p ac ice, howe e , he sys em had e ol ed o he poin whe e he e we e compa a i ely ew g ades whose pay was de e mined by di ec e e ence o p i a e sec o compa a o s, and a g ea numbe o g ades whe e in e nal ela i i ies we e he main de e minan s4. The esul ing sys em had become igid and in lexible, mili a ed agains he disc e e ea men o indi idual g ades and occupa ions, and had an inhe en endency o gene a e wage-wage spi als. Agains his backg ound, he I ish go e nmen , oge he wi h he social pa ne s, es ablished he Public Se ice Benchma king Body (PSBB) in July 2000. The pu pose o he PSBB was o examine pay and jobs in he public sec o o hose g ades no co e ed by he Buckley Re iew G oup5, and o make ecommenda ions in ela ion o a es o pay based, acco ding o he Body’s e ms o e e ence, on ‘in-dep h and comp ehensi e esea ch and analysis o pay le els ac oss he p i a e sec o ’. The backg ound o he es ablishmen o he PSBB is deal wi h mo e ully in O’Lea y (2002). Fo p esen pu poses, one aspec o i s manda e is especially dese ing o no e, namely he scope o he exe cise. While i was no asked o examine e e y single public sec o g ade, he numbe o g ades ha did come di ec ly wi hin i s emi co e ed 145,000 public sec o employees o o e 4 Amongs he g ades whe e emune a ion was explici ly benchma ked o he p i a e sec o we e hose a he e y op o he public-sec o hie a chy. Pay a es o hese g ades ha e been de e mined o many yea s now by he Re iew Body on Highe Remune a ion in he Public Sec o ( he “Buckley Re iew G oup”). The mos ecen epo o his g oup was published in 2000 (Go e nmen Publica ions, he S a ione y O ice, Dublin). 5 See p e ious oo no e. 7 60 pe cen o he o al. The e is no in e na ional p eceden ha we a e awa e o o a benchma king exe cise o his scale o scope. The PSBB epo ed in June 2002. I ecommended pay inc eases o e e y g ade i e iewed, anging om 3 pe cen in he case o labo a o y echnicians and hi d-le el lec u e s o 25 pe cen o ce ain g ades o pa amedic, and a e aging jus unde 9 pe cen o e all. The cos o implemen ing i s ecommenda ions is es ima ed a €1.1bn annually, o abou 3.5 pe cen o annual ne cu en go e nmen expendi u e. Implemen a ion is now in ain and, subjec o condi ions ela ing o changes in wo k p ac ices and he like being ul illed, he PSBB’s ecommenda ions will ha e been ully deli e ed by ea ly 2005. A cu ious ea u e o he PSBB’s epo is ha i u nished no speci ic jus i ica ion o any o he pay inc eases i p oposed. Ins ead, i p o ided a gene alised a ionale o i s co pus o ecommenda ions ha echoed i s e ms o e e ence and ci ed a numbe o b oad conside a ions. A no s age he e o e did he PSBB indica e ha i s pay ecommenda ions, ei he in gene al o in pa icula , a ose because o a pay gap be ween he public and p i a e sec o s and he pe cei ed need o b idge such a gap. In his espec he PSBB’s app oach di e ed ma kedly om ha o he Buckley Re iew G oup, whose 2000 epo con ained explici and unambiguous s a emen s o he e ec ha sala ies o op public se ice pos s we e subs an ially ou o line wi h hose o compa able posi ions in he p i a e sec o . S ill, he imp ession ha he PSBB’s esea ch showed ha he gene ali y o public sec o g ades we e unde paid ela i e o hei p i a e sec o coun e pa s, and ha such esea ch esul s o med pa o he a ionale o he pay ecommenda ions made by he PSBB, has been p opaga ed by in luen ial igu es, including public sec o ade-union leade s6. I is no possible o es he alidi y o 6 Fo example, Pe e McLoone, Gene al Sec e a y o he IMPACT ade union, esponded o he PSBB’s epo by saying: “The ou come has indica ed IMPACT’s iew ha public se ice pay ell behind du ing he economic boom” (see he IMPACT websi e a www.impac .ie) The Execu i e Commi ee o ano he public se ice union, he Associa ion o Highe Ci il and Public Se an s, esponded as ollows: “The independen Benchma king Body has clea ly es ablished ha public sec o 8 hese imp essions since he PSBB’s esea ch has no been published. One objec i e o his pape , he e o e, is o examine he e idence on he ex en o he pay gap be ween public and p i a e sec o wo ke s a he ime he PSBB’s esea ch was ca ied ou and how ha gap e ol ed o e he p eceding yea s. Ano he objec i e is o si ua e he I ish e idence in he con ex o he in e na ional li e a u e on his issue. To his end, we use a a ie y o econome ic echniques o in es iga e he size, na u e and dynamics o public-p i a e sec o pay di e en ials in I eland du ing he 1994-2001 pe iod, and o compa e and con as he I ish expe ience in his ega d wi h he in e na ional expe ience. 3. Some ecen in e na ional indings To ou knowledge, he kind o analysis we ha e unde aken has ne e be o e been ca ied ou on I ish da a. Howe e , as no ed abo e, a conside able olume o esea ch along hese lines has been unde aken o o he coun ies. Good e iews o he li e a u e and o he me hodological issues ha a ise a e con ained in Gunde son (1998) and G ego y and Bo land (1999). A b ie summa y o me hodology and p incipal esea ch indings ollows. Ul ima ely, he ques ion ha he bulk o his esea ch a emp s o answe is whe he employees who a e iden ical in all ma e ial espec s would ea n mo e by wo king in he public han he p i a e sec o . The gene ic me hod used o add ess his ques ion in ol es compa ing he ea nings o indi idual wo ke s, con olling o di e ences in p oduc i i y- ela ed cha ac e is ics and job a ibu es, ac oss he wo sec o s. This ypically in ol es es ima ing eg ession equa ions wi h some measu e o ea nings as he dependen a iable, and wi h a se o explana o y a iables designed o cap u e p oduc i i y- ela ed pe sonal a ibu es (such as age, expe ience and educa ion) and job cha ac e is ics (such as emune a ion is behind, and in some cases subs an ially behind, equi alen p i a e sec o emune a ion and he ai a e o he job” (see he AHCPS websi e a www.ahcps.ie) 9 g ea many employees, and ce ainly no o he majo i y o manage ial, p o essional and echnical wo ke s. Acco dingly, we use g oss mon hly ea nings in ou co e speci ica ion. Howe e , we also es he impac o employing he hou ly measu e. Since a e age weekly hou s wo ked a e highe in he p i a e sec o , he gap in hou ly ea nings be ween he wo sec o s is wide han he gap in mon hly ea nings. A sho coming o ou ea nings a iable, common o o he s udies in his ype, is ha i excludes impo an elemen s o emune a ion such as pension en i lemen s, bene i s-in-kind and pe o mance- ela ed pay. As a as he la e is conce ned, he e is a se ies in he ECHP da ase called ‘o he ea ned income’, bu i is a de icien measu e o pe o mance- ela ed pay in ha i includes lump-sum paymen s such as edundancy paymen s, back-da ed pay inc eases and he like. The e a e no da a in he ECHP ela ing o pension en i lemen s, so i is simply no possible o cons uc a measu e o emune a ion ha inco po a es he alue o pensions. The same applies o bene i s-in-kind. O cou se, he e a e o he dimensions o a job ha , while no pa o he emune a ion package, a e alued by employees. An ob ious one, especially ele an o an analysis o public-p i a e sec o pay di e en ials, is job secu i y. In gene al, public sec o employmen is mo e secu e han p i a e sec o employmen : he concep o ‘a job o li e’ is much mo e p e alen in he public han he p i a e sec o , and he isk o losing one’s job is much lowe . The e is no way o adjus ing ou dependen a iable o e lec his di e ence. 4.3 The explana o y a iables All he independen a iables ha a e ypically used in his kind o analysis a e eadily a ailable in some o m o o he om he ECHP da ase . Responden s a e asked hei age, gende and ma i al s a us. These a e s aigh o wa d and equi e no 16 elabo a ion. Responden s a e also asked whe he hey su e om any ch onic disabili y and, since his is an ob ious po en ial in luence on p oduc i i y, we ha e used i as an independen a iable. The educa ion a iable in Mince ean equa ions is usually ei he a ‘comple ed yea s o educa ion’ o a ‘highes le el o educa ion a ained’ a iable. The o me is p e e able on g ounds o g anula i y. Howe e , he ECHP da a se does no yield a sa is ac o y a iable o his ype, so we ha e been cons ained o use an a ainmen a iable ha dis inguishes be ween h ee le els o comple ed educa ion: lowe second-le el; uppe second-le el; and hi d-le el (deg ee o o he hi d-le el quali ica ion). In epo ing ou esul s below, hese a e designa ed as Le el 1, Le el 2 and Le el 3 espec i ely. Ou educa ion a iable he e o e, does no dis inguish be ween di e en hi d-le el quali ica ions and, since he span he e is e y wide ( om ce i ica es ea ned a e a wo-yea cou se a an Ins i u e o Technology o a Ph.D. ea ned a e up o en yea s s udy a a uni e si y), his is no en i ely sa is ac o y. A wo k-expe ience a iable is also ypical in Mince ean equa ions. Concep ually, he e a e se e al op ions he e, one being he o al yea s wo ked by an indi idual ac oss all employmen s (a measu e o o al wo k expe ience); ano he being yea s wo ked by an indi idual in his/he cu en employmen (a measu e o job- speci ic expe ience). The Li ing in I eland Su ey (LIS), om which he ECHP is de i ed, yields da a ha allow a measu e o he o me ype o be cons uc ed. This is no he case wi h he ECHP da a se . As a esul , we a e cons ained o using a measu e o he la e ype, which is a guably in e io . Howe e , gi en ha we also use age as an explana o y a iable, we do no hink his poses a se ious p oblem. The a iables discussed so a - age, gende , ma i al s a us, disabili y, educa ion and expe ience – e lec indi iduals’ p oduc i i y- ela ed a ibu es. We ha e also used a iables ha ela e o job cha ac e is ics: uni size; occupa ion; con ac ype and pa - ime s a us. As a as uni size is conce ned, he ECHP con ains in o ma ion on he numbe o people wo king in he 17 es ablishmen whe e he indi idual esponden is cu en ly employed. The a iable is s a i ied in o six size anges om less han h ee up o 500-plus. The con ac - ype a iable dis inguishes be ween pe manen , ixed- e m and casual con ac s. As o occupa ion, we dis inguish be ween he ollowing ca ego ies: ‘manage ial’ (including legisla o s and senio o icials); ‘p o essional’; ‘ echnical’; ‘cle ical’; ‘se ice’ (including sales wo ke s); ‘c a ’; ‘ope a i es’ (including assembly wo ke s), and ‘elemen a y’ (including unskilled labou e s). 4.4 Di e ences be ween he sec o s On a e age, he public se an s in ou sample a e paid much mo e han he p i a e-sec o wo ke s (Table 3). In e ms o ou chosen dependen a iable, g oss mon hly ea nings, public se an s ecei ed €2778 on a e age in 2001, compa ed wi h an equi alen igu e o €1905 o p i a e sec o wo ke s, a di e en ial o 46 pe cen . In e ms o he g oss hou ly ea nings a iable we ha e cons uc ed, he di e en ial is subs an ially la ge , a almos 60 pe cen . A e age hou ly ea nings o public-sec o wo ke s a e es ima ed a €18.23 in 2001, compa ed wi h €11.41 o p i a e- sec o wo ke s8. The co esponding di e en ials o ea lie yea s we e o a simila o de o magni ude. Fo he 1994-2001 pe iod, ou da a se sugges s ha public se an s achie ed an annual a e age inc ease in mon hly ea nings o 6 pe cen , sligh ly mo e han he 5.8 pe cen indica ed o p i a e sec o employees. The equi alen a es o inc ease in a e age hou ly ea nings o he public and p i a e sec o s a e 6.4 pe cen and 6.6 pe cen espec i ely. These igu es sugges ha o e he medium o long e m he a io o aw ea nings as be ween he wo sec o s in I eland does no change e y much, e en i he e a e app eciable sho - e m luc ua ions in ha a io. This echoes he indings o o he esea che s such as Fi zGe ald (2002) and Casey (2004). 8 These igu es imply ha he a e age hou ly ea nings o public sec o employees we e almos 39% highe han he a e age o all employees in 2001. This is somewha highe han he es ima e o 34% a i ed a by Casey (2004). 18 Table 3: Fea u es o he da a employed in he analysis o I ish public-p i a e wage di e en ials, 1994 and 2001 1994 1994 2001 2001 Va iable Public P i a e Public P i a e Obse a ions # 1,116 2,130 413 1,223 G oss mon hly ea nings € 1,847.52 1,281.19 2,777.55 1,905.25 G oss hou ly ea nings € 11.83 7.30 18.23 11.41 Age Yea s 39.95 33.05 41.73 35.13 Ma ied % 0.718 0.488 0.685 0.488 Male % 0.555 0.581 0.504 0.546 Pa - ime % 0.148 0.116 0.094 0.136 Disabili y % 0.080 0.075 0.056 0.067 Le el 3 Educa ion 0.342 0.177 0.433 0.196 Uppe Le el 2 Educa ion 0.420 0.491 0.387 0.491 Lowe Le el 2 Educa ion 0.237 0.333 0.179 0.313 Uni size 1-4 employees NA NA 0.061 0.121 Uni size 5-19 employees NA NA 0.194 0.262 Uni size 20-49 employees NA NA 0.206 0.218 Uni size 50-99 employees NA NA 0.133 0.117 Uni size 100-499 employees NA NA 0.247 0.196 Uni size 500+ employees NA NA 0.145 0.076 Tenu e <=3 yea s 0.242 0.454 0.288 0.602 Tenu e 4-8 yea s 0.134 0.270 0.153 0.179 Tenu e 9-13 yea s 0.143 0.098 0.114 0.074 Tenu e 14-18 yea s 0.002 0.000 0.087 0.051 Tenu e 19+ yea s 0.479 0.178 0.322 0.080 Con ac ype : Pe manen NA NA 0.891 0.866 Con ac ype: Fixed NA NA 0.061 0.037 Con ac ype: Casual NA NA 0.027 0.064 Con ac ype: O he NA NA 0.022 0.033 Occup. ype: Manage ial 0.037 0.066 0.068 0.093 Occup. ype: P o essional 0.323 0.083 0.327 0.069 Occup. ype: Technical 0.116 0.088 0.143 0.098 Occup. ype: Cle ical 0.186 0.143 0.186 0.153 Occup. ype: Se ice 0.118 0.174 0.126 0.196 Occup. ype: C a 0.058 0.173 0.031 0.146 Occup. ype: Ope a i e 0.060 0.163 0.051 0.169 Occup. ype: Elemen a y 0.101 0.110 0.068 0.076 Sou ce: ECHP, I eland. On a e age, public sec o wo ke s in I eland a e olde ; mo e highly educa ed and ha e spen longe in hei cu en job han 19 hei p i a e sec o coun e pa s. They a e also mo e likely o be ma ied and mo e likely o be emale. In hese espec s, he I ish pa e n mi o s ha o o he coun ies whe e simila esea ch has been ca ied ou (see, o example, Luci o a and Meu s (2004) o da a on F ance, I aly and he UK, and Melly (2002) o da a on Ge many). The e a e se e al o he ele an dimensions along which I ish public sec o and p i a e sec o wo ke s a e sys ema ically di e en . Public sec o wo ke s a e mo e likely o be employed on pe manen o ixed- e m con ac s. They end o wo k in la ge en i ies: 39 pe cen o public se an s wo k in en i ies (local uni s) ha employ 100 o mo e people; he co esponding p opo ion o he p i a e sec o is 27 pe cen . Impo an ly, he occupa ional p o iles o he wo sec o s a e e y di e en . Almos 33 pe cen o all public sec o wo ke s a e classi ied as ‘p o essional’, compa ed wi h less han 7 pe cen o he p i a e sec o . He e he high p opo ion egis e ed o he public sec o e lec s in e alia he classi ica ion o eache s. A he o he end o he occupa ional spec um, 32 pe cen o p i a e sec o wo ke s all in o ei he he ‘c a ’ o ‘ope a o ’ ca ego ies, compa ed wi h jus 8 pe cen o he public sec o . One o he di e ence be ween he wo sec o s is wo h no ing a his s age. I ela es o he dis ibu ion o ea nings wi hin each o he sec o s. A p io i, one would expec he dis ibu ion o be mo e comp essed in he public han he p i a e sec o . This is wha he in e na ional li e a u e shows. I is also wha ou da a se indica es o I eland. The coe icien o a ia ion in espec o a e age mon hly ea nings is consis en ly la ge o he p i a e han he public sec o o e he pe iod 1994 h ough 2001 (Table 4). Table 4: Coe icien o a ia ion in public and p i a e g oss- mon hly ea nings, 1994-2001 1994 1995 1996 1997 1998 1999 2000 2001 Public 0.5283 0.4512 0.4562 0.4769 0.4811 0.5072 0.5590 0.5389 P i a e 0.7126 0.7056 0.6477 0.6462 0.6339 0.6177 0.6164 0.5893 Sou ce: ECHP, I eland. 20 Howe e , in ou sample, he coe icien o a ia ion o he p i a e sec o declines subs an ially be ween 1994 and 2001, while i inc eases, albei ma ginally, o he public sec o o e his pe iod. Taking he wo sec o s oge he , ou da a se indica es ha he coe icien o a ia ion was conside ably lowe in 2001 han in 1994. Wha all o his sugges s is ha he dispe sion o employee ea nings declined in he non-ag icul u al economy du ing he Cel ic Tige pe iod and ha his was mo e han ully accoun ed o by he educed dispe sion o p i a e-sec o ea nings. O he esea che s ha e unco e ed simila e idence. Fo example, Ba e , Fi zGe ald and Nolan (2002) epo a small educ ion in ea nings dispe sion as measu ed by he a io o he op o he bo om decile o he pe iod 1994 o 1997, in ma ked con as o he la ge inc ease in dispe sion hey ound o he pe iod 1987 o 1994. They a ibu e he 1994-1997 pa e n o immig a ion, which hey a gue was concen a ed amongs skilled wo ke s and hus augmen ed he supply o skilled wo ke s ela i e o he supply o unskilled wo ke s in he I ish economy. I is likely ha a simila explana ion applies o he pe iod om 1997 h ough 2001 as well. 5. The cen al esul s 5.1 OLS eg ession esul s all employees We i s p esen OLS eg ession esul s o 2001 (Table 5). These a e ypical o all yea s o ou analysis. Two se s o esul s a e p esen ed; he i s based on g oss mon hly ea nings as he dependen a iable, he second on g oss hou ly ea nings. The coe icien s on all he explana o y a iables ha e he co ec sign and, wi h ew excep ions, a e s a is ically signi ican . Ea nings a e posi i ely co ela ed wi h age, educa ion a ainmen , job expe ience and es ablishmen size, and a y as expec ed ac oss he occupa ional spec um. Employees who a e male and ma ied, all o he hings equal, end o ea n mo e han o he s. Those on pe manen con ac s, all o he hings equal, ea n mo e han hose on ixed- e m o casual con ac s. 21 Table 5: OLS es ima es o he wage equa ion, 2001 Va iable Mon hly ea nings Hou ly ea nings Coe icien S anda d E o * Coe icien S anda d E o * Public sec o 0.1298 ( 0.0243 ) 0.1636 ( 0.0234 ) A g e 0.0209 ( 0.0061 ) 0.0210 ( 0.0059 ) A g e s q ua ed -0.0002 ( 0.0001 ) -0.0002 ( 0.0001 ) Male 0.2383 ( 0.0215 ) 0.1371 ( 0.0199 ) Ma ied 0.0783 ( 0.0251 ) 0.0813 ( 0.0251 ) Pa - ime -0.6518 ( 0.0337 ) -0.0823 ( 0.0305 ) Disabili y -0.0178 ( 0.0345 ) -0.0626 ( 0.0372 ) Educa ion le el 3 0.3279 ( 0.0319 ) 0.3219 ( 0.0305 ) Educa ion le el 2 0.1220 ( 0.0239 ) 0.1156 ( 0.0217 ) Uni : 5-19 0.0416 ( 0.0358 ) 0.0678 ( 0.0363 ) Uni : 20-49 0.0769 ( 0.0364 ) 0.1182 ( 0.0368 ) Uni : 50-99 0.1301 ( 0.0404 ) 0.1580 ( 0.0406 ) Uni : 100-499 0.1294 ( 0.0379 ) 0.1463 ( 0.0376 ) Un i: 500+ 0.1761 ( 0.0435 ) 0.1904 ( 0.0439 ) Tenu e: 4-8 0.0915 ( 0.0247 ) 0.0836 ( 0.0233 ) Tenu e: 9-13 0.0934 ( 0.0363 ) 0.0515 ( 0.0357 ) Tenu e: 14-18 0.1555 ( 0.0374 ) 0.1334 ( 0.0379 ) Tenu e: 19+ 0.2254 ( 0.0347 ) 0.1934 ( 0.0334 ) Mana g emen 0.3854 ( 0.0483 ) 0.3019 ( 0.0453 ) P o essional 0.3772 ( 0.0484 ) 0.4638 ( 0.0455 ) Technical 0.2361 ( 0.0441 ) 0.2258 ( 0.0401 ) Cle ical 0.1204 ( 0.0402 ) 0.1154 ( 0.0359 ) Se ice 0.0153 ( 0.0409 ) 0.0177 ( 0.0349 ) C a 0.2379 ( 0.0437 ) 0.2175 ( 0.0394 ) O p e a i e 0.1637 ( 0.0409 ) 0.1037 ( 0.0374 ) Pe manen 0.1597 ( 0.0380 ) 0.0997 ( 0.0326 ) Fixed- e m 0.1419 ( 0.0695 ) 0.0460 ( 0.0646 ) Cons an 6.3307 ( 0.1162 ) 1.2853 ( 0.1118 ) R 2 0.6493 0.5536 * He e oskedas ic obus s anda d e o s. A comple e se o esul s o he yea s 1994 o 2001 is a ailable by eques om he au ho s. The es ima e o he public-sec o p emium, using g oss mon hly wages as he dependen a iable, is 13 pe cen and highly signi ican . In o he wo ds, public sec o wo ke s in I eland in 2001 we e, on a e age, paid 13 pe cen mo e han p i a e sec o wo ke s, condi ioning on a wide a ay o p oduc i i y- ela ed job and pe sonal cha ac e is ics. The es ima e o he public sec o p emium using g oss hou ly ea nings is somewha highe , a o e 22 16 pe cen , and also highly signi ican . These es ima es a e conside ably highe han hose epo ed by Luci o a and Meu s o F ance, I aly and he UK o 1998, by Melly o 2000 o Ge many, and by Disney and Gosling o he UK o 1991-95 and 1990-99. This, despi e he ac ha hese au ho s use speci ica ions ha a e likely o explain a smalle ac ion o he aw public- p i a e ea nings di e en ial han ou s. Table 6 se s ou he es ima es o he p emium ob ained o each o he yea s 1994 h ough 2001 using di e en speci ica ions o he wage equa ion. Each speci ica ion uses g oss mon hly ea nings as he dependen a iable bu a di e en se o explana o y a iables. This exe cise illumina es wo aspec s o he es ima ed public sec o p emium: (i) i s sensi i i y o he inclusion o ce ain explana o y a iables, and (ii) i s beha iou o e ime. We i s explo e he ques ion o he sensi i i y o he esul s o changes in speci ica ion. The speci ica ion labelled ‘Basic’ in he able con ains he ollowing explana o y a iables: age, age squa ed, gende , ma i al s a us, educa ion and expe ience, as well as dummy a iables o pa - ime s a us and disabili y. This is he speci ica ion ha is mos ypical o he one ound in he in e na ional li e a u e. To his basic equa ion we add se ially he occupa ion a iable, he con ac - ype a iable, and he uni size a iable. Table 6: Es ima e o he public-sec o p emium (all employees) o a ious speci ica ions o he wage equa ion, 1994-2001 Equa ion ype 1994 1995 1996 1997 1998 1999 2000 2001 0.1526 0.1693 0.1605 0.1281 0.1535 0.0960 0.1335 0.1398 Basic (0.0180 ) (0.0188 ) (0.0187 ) (0.0198 ) (0.0205 ) (0.0219 ) (0.0236 ) (0.0245 ) Basic + Occupa io n 0.1407 (0.0174 ) 0.1509 (0.0184 ) 0.1471 (0.0186 ) 0.1215 (0.0194 ) 0.1407 (0.0205 ) 0.0905 (0.0211 ) 0.1200 (0.0231 ) 0.1398 (0.0250 ) 0.1599 0.1521 0.1300 0.1625 0.1027 0.1398 0.1442 Basic + Con ac Type NA (0.0187 ) (0.0188 ) (0.0195 ) (0.0204 ) (0.0218 ) (0.0235 ) (0.0245 ) NA NA NA 0.1049 0.1325 0.0769 0.1106 0.1257 Basic + Uni Size (0.0192 ) (0.0199 ) (0.0215 ) (0.0232 ) (0.0240 ) 23 Ex ended* 0.1407 0.1460 0.1418 0.0995 0.1336 0.0831 0.1080 0.1298 (0.0174 ) (0.0183 ) (0.0185 ) (0.0183 ) (0.0199 ) (0.0207 ) (0.0221 ) (0.0243 ) * The coe icien s o he ‘ex ended’ equa ion a e no compa able ac oss he ull pe iod – see ex o discussion. No e: s anda d e o s a e in pa en heses. Comple e eg ession esul s a e a ailable on eques om he au ho s. The inclusion o he occupa ion a iable has he e ec o educing he es ima ed public sec o p emium in each yea excep 2001, ypically by abou 1-1.5 pe cen poin s. The inclusion o con ac ype has he e ec o educing he es ima ed p emium o he yea s 1995 and 1996, bu has he opposi e e ec in each subsequen yea . The con ac - ype a iable could no be included o ei he sec o o 1994 because he ele an da a we e no collec ed o ha yea . The uni -size a iable could only be included om 1997, since he ele an da a we e no collec ed o he public sec o in ei he 1995 o 1996 and no da a on uni size we e collec ed o 1994. Once included, he uni -size a iable has he e ec o subs an ially educing he es ima ed p emium, ypically by mo e han 1.5 pe cen poin s. Compa ing he coe icien s lis ed in he i s and las ows o Table 6 om 1997 e eals ha he o e all e ec o including all h ee a iables, as we do in he ‘ex ended’ equa ion, is o educe he es ima ed p emium o each yea . Fo ins ance, in 2001, ou es ima e o he p emium would be 14 pe cen a he han 13 pe cen , had we omi ed all h ee a iables om ou equa ion. I is impo an o no e ha he coe icien s es ima ed o he ex ended equa ion a e no compa able ac oss he ull 1994-2001 pe iod. This is because bo h he uni -size and con ac - ype a iables a e excluded om his speci ica ion o 1994 and he uni -size a iable is excluded o 1995 and 1996. The coe icien s es ima ed o he ex ended model a e howe e compa able ac oss he yea s 1997 h ough 2001. Fo his sub-pe iod, he e is no disce nible end in he p emium, al hough i does luc ua e conside ably om yea o yea . O some in e es , in ligh o he igh ening o labou ma ke condi ions ha occu ed, is he ac ha he 2001 p emium is bigge han he 1997 p emium, hough no by a s a is ically signi ican ly ma gin. 24 To examine he beha iou o he public-sec o p emium o e a longe span we ha e o use one o he speci ica ions ha is co espondingly ime-consis en . Bo h he basic speci ica ion and he basic speci ica ion augmen ed by he occupa ional a iable pe mi us o compa e he p emium o e he ull 1994-2001 pe iod. In he o me case he e is some sugges ion o a decline in he p emium, bu he di e ence be ween he es ima es o 1994 and 2001 is no s a is ically signi ican . In he la e case, he p emium es ima ed o 1994 is i ually iden ical o ha es ima ed o 2001. Again, gi en he igh ening o labou ma ke condi ions ha occu ed o e his ime ame, and he widesp ead belie ha his had lowe ed ea nings in he public sec o ela i e o he p i a e sec o , his is a no ewo hy esul . 5.2 OLS eg ession esul s: in e ac ion e ec s A disad an age o he simple dummy- a iable app oach is ha es ima ed di e ences be ween he wo sec o s a e con ined o he in e cep e m; in o he wo ds ha e u ns o p oduc i i y- ela ed pe sonal a ibu es and job cha ac e is ics a e cons ained o be he same ac oss he wo sec o s. One way o mi iga ing his limi a ion is o un eg essions whe e he public sec o dummy is allowed in e ac wi h a ious explana o y a iables. This can be done wi h any o all o he explana o y a iables. He e, we epo on he esul s o wo such exe cises whe e he public sec o dummy a iable is in e ac ed wi h he gende and educa ion a iables espec i ely. O he wise, he speci ica ion o he equa ions is exac ly as abo e: in pa icula , hese equa ions a e based on he speci ica ion ha includes he ull se o explana o y a iables and uses g oss mon hly ea nings as he dependen a iable ( he ex ended speci ica ion). The esul s in espec o gende a e summa ised in Table 7. The second ow con ains he es ima ed public sec o p emium o women. We include he es ima ed p emium o all employees in he i s ow o acili a e he in e p e a ion o he esul s. The hi d ow shows he es ima ed di e ence be ween he p emium o women and he p emium o men. The p emium o women is posi i e and highly s a is ically signi ican h oughou he pe iod 25 Table 10: P obi es ima es o public-sec o s a us, I eland, 2000 Va iable Coe icien S anda d E o Age 0.0164 (0.0045) Male -0.0857 (0.0874) Ma ied 0.0047 (0.0967) Pa - ime -0.0104 (0.1296) Disabili y -0.0192 (0.1643) Educa ion le el 3 0.3824 (0.1330) Educa ion le el 2 0.1592 (0.1064) Tenu e: 4-8 0.4228 (0.1113) Tenu e: 9-13 0.5787 (0.1350) Tenu e: 14-18 1.0279 (0.1584) Tenu e: 19+ 1.1860 (0.1282) Uni : 5-19 0.0616 (0.1564) Uni : 20-49 0.2359 (0.1601) Uni : 50-99 0.2828 (0.1750) Uni : 100-499 0.3978 (0.1573) Un i: 500+ 0.6656 (0.1754) Managemen -0.6240 (0.2009) P o essional 0.5423 (0.1846) Technician -0.0793 (0.1811) Cle k 0.1435 (0.1686) Se ice -0.1012 (0.1677) C a -0.9353 (0.2047) Ope a o -0.6029 (0.1806) Pe manen -0.1508 (0.2316) Fixed 0.4844 (0.2786) Casual 0.0849 (0.2877) Cons an -1.8852 (0.3381) Log likelihood = -719.82443 Pseudo R2 = 0.2610 No e: s anda d e o s a e he e oskedas ic obus . Comple e eg ession esul s a e a ailable om he au ho s upon eques . 32 In he second s ep we ank he p opensi y sco es and di ide he obse a ions in o 10 equal s a a, wi h he lowes 10% o he p opensi y sco es in he i s s a a and he second 10% in he 2nd s a a and so on. Rosenbaum and Rubin (1983) demons a e ha he dis ibu ion o he obse able cha ac e is ics will be same wi hin each s a a. The e o e i one s a um is comp ised o all p i a e-sec o wo ke s hen he e a e no simila public-sec o wo ke s o compa e hem wi h so hese obse a ions need o be omi ed om he eg ession sample. As a wo king ule, we decided o omi a s a um o obse a ions i i con ained ewe han 20 indi iduals om ei he he public o p i a e sec o s. In he yea s 1997 o 2000, his ule esul ed in he i s h ee s a a (30% o ou obse a ions) being omi ed om ou eg ession sample; in 1996 we omi ed he i s wo s a a; and in 1994 and 1995 we excluded he i s s a a. Ou ‘ex ended’ wage speci ica ion was hen e- un o all yea s on his unca ed da ase and he esul ing p emium es ima es a e gi en in Table 11. As is appa en he esul s om he unca ed sample a e e y simila in all yea s and hey a e no signi ican ly di e en in any yea . This would imply ha sample selec ion bias, condi ional on obse able igh -hand-side a iables, is no an issue wi h ou da a. Table 11: Public-sec o p emium es ima es o he ‘comple e’ sample and he ‘ unca ed’ sample gene a ed by he p opensi y sco e me hod, I eland, 1994-2000 1994 1995 1996 1997 1998 1999 2000 0.1407 0.1460 0.1418 0.0995 0.1336 0.0831 0.1080 Comple e sample (0.0170) (0.0187) (0.0187) (0.0187) (0.0196) (0.0209) (0.0227) 0.1489 0.1479 0.1389 0.0999 0.1406 0.0840 0.1078 T unca ed sample (0.0171) (0.0186) (0.0184) (0.0186) (0.0202) (0.0217) (0.0228) No e: s anda d e o s a e in pa en heses. Comple e eg ession esul s a e a ailable om he au ho s upon eques . 6.2 Speci ica ion, measu emen e o and endogenei y bias 33 To he ex en ha wo ke s, due o unobse able alen s o as es, selec hemsel es in o pa icula sec o s hese ac o s will ha e an impac on he es ima ed public-sec o p emium. Thus i could be ha pa o he e ec we’ e cap u ing in he es ima ed p emium a ises om an unobse ed p oduc i i y o as e a iable. Hence he OLS es ima e o he p emium would be an o e es ima e o he ue p emium. I is impo an he e o e o assess he po en ial impo ance o his ac o . The panel ea u e o ou da a allows us o use indi idual-wo ke dummy a iables as a p oxy o possible unobse able ac o s. The applica ion o he ixed-e ec s es ima o o ou ex ended wage model implies ha he unobse able ac o s a e unchanged o e ime and ha he e u ns o hese unobse able ac o s a e simila o public and p i a e sec o wo ke s. Gi en hese assump ions, he applica ion o he ixed-e ec s es ima o will yield an es ima e o he p emium ha will be de e mined by hose employees ha mo e sec o o e he se en yea s o ou panel. (A non-ze o coe icien could also be gene a ed by he p esence o a di e en ial wage g ow h be ween he public and p i a e sec o s o e he se en-yea pe iod. This possibili y is es ed below.) This aises a numbe o issues ha impinge on he assessmen o he esul ing es ima e o he p emium. I seems easonable ha he po en ial p oblems o bo h measu emen e o and endogenei y bias will end o be exace ba ed wi h he ixed- e ec s es ima o . The p esence o bo h o hese biases will end o bias he ixed-es ima o owa ds ze o which will ope a e o o se he upwa d bias due o omi ed unobse able a iables ha a lic he c oss-sec ional eg essions. Bu , as Disney and Gosling (2003) pe suasi ely a gue, he e is no s ong eason o suppose, as does F eeman (1984), ha OLS will p o ide an upwa d bound and he ixed-e ec s es ima o a lowe bound es ima e o he p emium. In he p esence o bo h measu emen e o and endogenei y bias he app op ia e es ima o would be ins umen al a iables (IVs). As always ob aining IVs wi h he equisi e p ope ies is ne e easy. Fo una ely he ECHPS con ains some quali a i e a iables ha could se e as po en ial IVs. One in pa icula looks p omising, namely, wo ke s’ sa is ac ion le els in esponse o a 34 ques ion on job secu i y. Responses classi ied by public and p i a e-sec o employees o a numbe o yea s da a a e shown in Table 12. These da a show ha he e is a s ong co ela ion be ween “job secu i y” sa is ac ion le els and public-sec o s a us. Table 12: Sa is ac ion le els wi h job secu i y in he public and p i a e sec o s, I eland, 1994-2000 No e: ECHPS, I eland. 1994 1996 1998 2000 Sa is ac ion le el % % % % Public P i a e Public P i a e Public P i a e Public P i a e 1 8.46 10.91 3.5 5.82 3.83 3.76 1.55 2.18 2 4.53 8.62 2.43 6.49 2.19 5.28 1.81 4.05 3 6.2 15.27 4.71 11.73 2.19 9.85 3.1 6.34 4 9.45 17.67 7.45 18.89 9.49 16.41 6.98 17.36 5 20.18 21.92 22.95 30.12 19.34 29.94 20.16 32.95 6 51.18 25.63 58.97 26.96 62.96 34.75 66.41 37.11 Wi h his a iable as an ins umen we es ima ed he wage equa ion o he balanced panel using 2SLS. In he i s s age we eg essed he public-sec o s a us dummy on he “job secu i y” sa is ac ion a iable and all o he o he igh -hand-side a iables in he ex ended wage equa ion. We hen gene a ed he p edic ed alues o he public-sec o s a us dummy and used his a iable ins ead o he o iginal a iable in he es ima ion o he wage equa ion and he public-sec o p emium. Table 13 p esen s a ious es ima es o he p emium ha we e ob ained using he panel da ase . We i s compa e ou OLS es ima es o he p emium o he balanced panel wi h he a e age alue o he es ima e ob ained o he se en yea s om he unbalanced panel. The esul s a e b oadly simila which gi es us some con idence ha panel a i ion does no appea o i sel o be associa ed wi h any no iceable biases. The ixed-e ec s es ima o yields a subs an ially lowe p emium han he OLS es ima o . On he ace o i , his sugges s ha he 35 OLS es ima es a e subjec o a sizeable upwa d bias due o omi ed a iables. We es ima ed sepa a e models o men and women which p oduced e y simila es ima es o he p emium bu he women’s p emium was no s a is ically signi ican (Annex 8). We also allowed he p emium es ima e o a y o e ime which e ealed a p onounced downwa d end o e he pe iod om 1994 o 2000. Table 13: OLS, ixed-e ec s and ixed-e ec s and 2SLS es ima es o he public-sec o p emium, I eland, 1994-2000 Es ima o P emium es ima e OLS a e age annual (unbalanced panel)* 0.1218 (0.0195) OLS balanced panel 0.1472 (0.0120) Fixed-e ec s 0.0474 (0.0211) Fixed-e ec s and 2SLS 0.1296 (0.0540) * The simple a e age (1994-2000) o he p emium es ima es gene a ed by he ‘ex ended’ speci ica ion o he wage equa ion in Table 6. No e: s anda d e o s a e in pa en heses. Comple e eg ession esul s a e a ailable om he au ho s upon eques . The combina ion o he ixed-e ec s and he IV es ima o e eals an es ima e o he p emium ha is qui e close o he OLS es ima e. This sugges s ha he ixed-e ec s and OLS es ima o s a e bo h subjec o a subs an ial downwa d bias due o a possible combina ion o measu emen e o and endogenei y bias. Taking he ixed-e ec s and ixed-e ec s-2SLS es ima o s oge he sugges s ha he posi i e bias ha p obably a ises due o he omission o impo an a iables is mo e han o se by he nega i e biases ha a e po en ially gene a ed by measu emen e o and endogenei y. All in all he e o e we eel ha ou esul s p o ide easonable g ounds o a es ing o he obus ness o he OLS es ima es. 6. Summa y and conclusions 36 I ish public sec o employees a e on a e age much be e paid han hei p i a e sec o coun e pa s. The ECHP da ase ha we ha e used o he analysis p esen ed in his pape indica es ha he ma gin in he yea 2001 was 46 pe cen when measu ed in e ms o a e age g oss mon hly ea nings, he measu e we conside o be he mos app op ia e o hose a ailable. This di e en ial is e y la ge bu o limi ed ele ance. In pa icula , i does no necessa ily mean ha pay is highe in he public sec o on a like- o -like basis. The eason is ha public se an s a e ypically be e endowed han p i a e sec o employees wi h he kind o a ibu es ha a ac highe pay: hey end o be olde , mo e expe ienced and be e educa ed; hey also end o wo k in mo e highly-skilled jobs and in bigge es ablishmen s. The key ques ion add essed in ou pape , he e o e, is whe he public sec o wo ke s in I eland a e paid mo e han p i a e sec o employees, when such di e ences in p oduc i i y- ela ed pe sonal a ibu es and job cha ac e is ics a e con olled o , in o he wo ds whe he a public-sec o p emium exis s. The answe is clea and unambiguous. We es ima e ha in 2001, he la es yea o which we ha e comple ed ou analysis, he p emium enjoyed by public se an s was abou 13 pe cen , on he basis o g oss mon hly ea nings. We ha e analysed whe he and how his p emium a ies wi h gende , wi h educa ional a ainmen , and ac oss he ea nings dis ibu ion. In summa y, wha we ha e ound is ha he public sec o p emium is signi ican ly bigge o hose nea he bo om o he ea nings dis ibu ion han o hose nea he op, was signi ican ly bigge o women han men in he mid-1990s bu no a he end o he 1990s, and does no a y signi ican ly ac oss di e en le els o educa ional a ainmen . We ha e also analysed whe he and how he public sec o p emium changed be ween 1994 and 2001. This was a pe iod o excep ionally apid ou pu and employmen g ow h, and co espondingly sha p igh ening o labou ma ke condi ions in 37 he I ish economy. On he basis o in e na ional e idence, i migh be expec ed ha in such ci cums ances he in ense compe i ion amongs employe s o a declining pool o a ailable wo ke s would ha e pushed up ea nings in he p i a e sec o mo e quickly han in he public sec o , esul ing in a con ac ion o he public sec o p emium. Ou esea ch unco e s no compelling e idence ha his was in ac he case: we es ima e he p emium o 2001 o be no signi ican ly di e en om ha es ima ed o 1994. Some o ou esul s a e consis en wi h he esea ch indings o o he coun ies. In pa icula , ou inding ha he public sec o p emium declines wi h he le el o ea nings is in keeping wi h he in e na ional pa e n. In con as , ou indings ha emale public se an s no longe enjoy a signi ican ly bigge p emium han hei male coun e pa s, and ha he public sec o p emium o hose ha ha e comple ed hi d-le el educa ion is no signi ican ly di e en om he p emium enjoyed by ea ly school-lea e s, un coun e o he in e na ional e idence. Howe e , he mos ema kable di e ence be ween ou esul s and hose o o he esea che s ela es o he absolu e size o he p emium. The public sec o p emium we ha e es ima ed o I eland is high by in e na ional s anda ds. A 13 pe cen o 2001, i compa es wi h co esponding es ima es in he ange 4-6 pe cen o F ance, I aly and he UK, using 1998 da a (ou I ish es ima e o 1998 is also a ound 13 pe cen ). This con as is all he mo e no able gi en ha we use a speci ica ion ha would end o p oduce a lowe es ima e o he p emium han he speci ica ion used in mos o he published in e na ional li e a u e wi h which we a e amilia . I is also wo h poin ing ou ha ou es ima e o I eland p eda es he subs an ial boos o public-sec o wages and sala ies ha lowed om he so-called ‘benchma king’ p ocess. Ou pape lea es a numbe o impo an ques ions unanswe ed. P obably he mos impo an is: why is ou es ima ed public sec o p emium o I eland so la ge? One possibili y is ha ou es ima es a e biased upwa ds because o laws in ou me hodology. Fo example, we may ha e omi ed impo an explana o y a iables, o he sample we ha e used may be biased – an impo an possibili y o check o gi en he a i ion in he ECHP panel o e he 1994- 38 2001 pe iod. We ha e ca ied ou a ange o igo ous obus ness checks o es ablish whe he his migh be he case, and we a e easonably con iden ha ou es ima es a e no ma e ially a ec ed by such biases. Tha being he case, he ques ion emains: why, on a like- o -like basis, a e I ish public-sec o employees paid so much mo e han hei p i a e-sec o coun e pa s? One possible explana ion, a a ian o he omi ed a iables a gumen , has o do wi h he de iciencies in ou measu e o ea nings. Recall ha he a iable we use alls well sho o being a comp ehensi e measu e o emune a ion, excluding as i does bene i s-in-kind, pe o mance- ela ed pay and he alue o pension en i lemen s. O cou se, i also ails o ake accoun o di e en a es o PRSI con ibu ion (mos public se an s make lowe PRSI con ibu ions han hei p i a e sec o coun e pa s) and he alue o he g ea e secu i y ha ypically a aches o public sec o jobs. Howe e , he p oblem wi h an explana ion along hese lines is ha , i all o hese ac o s we e aken in o accoun , i is likely ha he es ima ed public sec o p emium would ise a he han all, excep pe haps in he uppe eaches o he ea nings dis ibu ion. Ano he possible explana ion is ha ce ain public sec o wo ke s equi e compensa ion o he haza dous, s ess ul and/o unpleasan na u e o he wo k hey do. This ype o a gumen is o en ad anced in he discussion o he pay o g oups like p ison o ice s, Ga dai, eache s and nu ses. Howe e , i is no en i ely pe suasi e o a leas wo easons. Fi s , haza dous, s ess ul o unpleasan wo k is no peculia o he public sec o . The e a e also a g ea many p i a e sec o employees who ha e o endu e such wo king condi ions. Second, unless he e is eason o belie e ha he incidence o haza dous, s ess ul and unpleasan wo king condi ions is g ea e in public se ice employmen in I eland han elsewhe e, he a gumen does no explain why he public sec o p emium is so much highe in I eland han in o he ju isdic ions. The no ion ha a la ge p emium o public sec o employees can pe sis is an uncom o able one o classically- ained economis s, since i sugges s ha ma ke s a e no unc ioning p ope ly. I ma ke s we e unc ioning p ope ly, p i a e sec o wo ke s would compe e o be e -paying public sec o jobs o he poin whe e he 39 public sec o p emium would disappea . He ein lies ano he possible explana ion. The ac ha a la ge p emium pe sis s may be indica i e o ma ke ailu es such as in o ma ion de iciencies, skill misma ches and ba ie s o en y. The a gumen he e is ha some combina ion o hese ac o s may e ec i ely p e en p i a e sec o wo ke s om compe ing o mo e luc a i e public sec o posi ions, excep ac oss a limi ed ange o ec ui men g ades. E en his explana ion, despi e i s in ellec ual appeal, is incomple e howe e . I may shed ligh on why a public sec o p emium pe sis s bu , in i sel , i does no explain why (i) he p emium in I eland is so much highe han elsewhe e, o (ii) why he I ish p emium was main ained a such a high le el h ough a pe iod cha ac e ised by he ype o labou ma ke condi ions ha migh ha e been expec ed o cause i o decline. An ob ious a ea o explo e o answe s o hese ques ions is he wage ba gaining amewo k ha was in place du ing his pe iod. A ull explo a ion o his a ea is ou side he scope o his pape . Tha said, an ob ious hypo hesis sugges s i sel . I is ha I eland’s social pa ne ship model, and in pa icula he eno mous alue ha poli icians and policy-make s a ached o i , con e ed on I ish public sec o ade unions g ea e ba gaining powe han hey would o he wise ha e had (and g ea e ba gaining powe han hei in e na ional coun e pa s ypically enjoyed). Acco ding o his hypo hesis, public sec o ade unions we e hus enabled o mo e e ec i ely p ese e ba ie s o en y, and mo e success ully pu sue pay claims ha ma ched he inc eases aking place ac oss he p i a e sec o du ing he Cel ic Tige pe iod. 40 41 Re e ences Ba e , A., Fi zGe ald, J., Nolan, B., 2002. “Ea nings Inequali y, Re u ns o Educa ion and Immig a ion in o I eland”, Labou Economics (No embe ) Blackaby, D., Mu phy, P., O’Lea y, N., 1996. “Public-P i a e Sec o Hou ly Ea nings Di e en ials in he UK: A Decile-Based Decomposi ion o he QLFS”. Economics DP 96-16, Uni e si y o Wales, Swansea. Casey, B., 2004. “An Economy-Wide Pe spec i e on Ea nings Da a in I eland”. ESRI Qua e ly Economic Commen a y (Sp ing). Conni e, D., Gash, V., O’Connell, P.J., 2000. “E alua ing S a e P og ammes: “Na u al Expe imen s” and P opensi y Sco es”. The Economic and Social Re iew, Vol. 31, No. 4, Oc obe , pp. 283- 308. Disney, R., Gosling, A., 2003. “A New Me hod o Es ima ing Public Sec o Pay P emia: E idence om B i ain in he 1990’s”. CEPR Discussion Pape No. 3787 (Feb.). Disney, R., Gosling, A., 1998. “Does i Pay o Wo k in he Public Sec o ?” Fiscal S udies, 19, 4, pp. 347-374. Fi zGe ald, J., 2002. “The Mac o-Economic Implica ions o Changes in Public Sec o Pay Ra es”, ESRI Qua e ly Economic Commen a y, pp. 43-58 (Win e ), Dublin. F eeman, R., 1984. “Longi udinal Analysis o he E ec s o T ade Unions”. Jou nal o Labou Economics, 2, 1, pp. 1-26. Gunde son, M., 1998. “Go e nmen Compensa ion: Issues and Op ions”. CPRN Discussion Pape No. W/03 (July). G ego y, R.G., Bo land, J., 1999 “Public Sec o Labou Ma ke s”. In Ashen el e , O., Ca d, D., eds., Handbook o Labou Economics, Vol. 3c, pp. 3573-3630. Ju ges, H., 2001, “The Dis ibu ion o he Ge man Public-P i a e Wage Gap”. Mimeo, Dep . o Economics, Uni e si y o Do mund, Ge many. Luci o a, C., Meu s, D., 2004, “The Public Sec o Pay Gap in F ance, G ea B i ain and I aly”. IZA Discussion Pape No. 1041. Melly, B., 2002, “Public-P i a e Sec o Wage Di e en ials in Ge many: E idence om Quan ile Reg ession”. Mimeo, SIAW, Uni e si y o S . Gallen, Swi ze land. O’Lea y, J., 2002. “Benchma king he Benchma ke s”. ESRI Qua e ly Economic Commen a y, pp. 77-91 (Win e ), Dublin.