NBERWORKINGPAPERSERIES

THEEFFECTSOFSOCIALINSURANCEBENEFITSONLEAVINGEMPLOYMENTATOLDERAGESINTHENETHERLANDS

AdriaanKalwij

ArieKapteyn

WorkingPaper31546

/papers/w31546

NATIONALBUREAUOFECONOMICRESEARCH

1050MassachusettsAvenue

Cambridge,MA02138

August2023

TheviewsexpressedhereinarethoseoftheauthorsanddonotnecessarilyreflecttheviewsoftheNationalBureauofEconomicResearch.

Atleastoneco-authorhasdisclosedadditionalrelationshipsofpotentialrelevanceforthisresearch.Furtherinformationisavailableonlineat/papers/w31546

NBERworkingpapersarecirculatedfordiscussionandcommentpurposes.Theyhavenotbeenpeer-reviewedorbeensubjecttothereviewbytheNBERBoardofDirectorsthataccompaniesofficialNBERpublications.

?2023byAdriaanKalwijandArieKapteyn.Allrightsreserved.Shortsectionsoftext,nottoexceedtwoparagraphs,maybequotedwithoutexplicitpermissionprovidedthatfullcredit,including?notice,isgiventothesource.

TheEffectsofSocialInsuranceBenefitsonLeavingEmploymentatOlderAgesintheNetherlandsAdriaanKalwijandArieKapteyn

NBERWorkingPaperNo.31546

August2023

JELNo.H0,J26

ABSTRACT

IntheNetherlands,from1989to2013,intheagegroup55-63theannualexitratefromemploymenttoreceivingsocialinsurancebenefitsinthefollowingyeardecreasedfromaround17percentto7percentformen,andfrom14percentto5percentforwomen.Wefoundthatlessgeneroussocialinsurancebenefitshavehadsmallbutsignificantnegativeeffectsontheseexitrates:Theannualexitratetosocialinsurancebenefitreceiptnextyear(atages56-64)wouldhavebeenabout14percenthigherforbothmenandwomenin2013shouldsocialinsurancebenefitsschemesof1989stillhavebeeninplace.Thisincreaseamountstostaying,onaverage,threemonthslongerinemploymentfromage55onwardsin2013thanin1989.Thesefindingsaredriventosomeextentbythereductioninthemaximumdurationofunemploymentinsurancebenefitsin2007,butpredominantlybymaking(early)retirementschemesactuariallyfairfrom2006onwards.Theincreaseindisabilityinsurance’sincomereplacementratein2006hasledtoaslightincreaseintheexitratefromemployment,conditionaloneligibility.Astheestimatedeffectsofchangesinthesocialinsurancebenefitsfrom1989to2013onworkingbeyondage55arerelativelysmall,theysuggesttheimportanceofotherfactorssuchaschangesinworkers’skills,improvedhealth(onwhichweprovidesomeevidence),andsocialinsurance’stightereligibilitycriteria.

AdriaanKalwij

UtrechtUniversitySchoolofEconomics

NetworkforStudiesonPensions,Aging

andRetirement

TheNetherlands

a.s.kalwij@uu.nl

ArieKapteyn

UniversityofSouthernCalifornia

CenterforEconomicandSocialResearch

635DowneyWaySuite312

LosAngeles,CA90089-3332

andNBER

kapteyn@

2

1.Introduction

Socialinsurancereformssincethemid-nineteen-ninetieshavereducedthefinancialincentivesforleavingemploymentatolderagesintheNetherlandsforthoseeligibleforbenefits.Thischapterinvestigatestheclaimmadeinourprevioustwochapters(Kalwij,Kapteyn&deVos2018;deVos,Kapteyn&Kalwij2018),thatthesereducedfinancialincentivesduringrecentdecadesmayhavedecreasedtheprobabilityofleavingemploymentatolderagesand,consequently,contributedtotherisingemploymentratesatolderagessincethemid-nineteennineties.Inparticular,wequantifythecontributionsofthechangesinthebenefitsschemesofvarioussocialinsuranceprogramsfrom1989to2013,includingearlyretirementschemes

1

,tothedeclineintheprobabilityofexitingemploymentandclaimingsocialinsurancebenefitsbeforethestatepensionageduringthisperiod.

Thereformsofsocialinsurancesincethemid-nineteen-ninetieshave,inpart,beentriggeredbyanageingDutchpopulation.Asteadyincreaseinlifeexpectancyforoveracenturyandsharplydecreasingfertilityduringthenineteen-sixtieswithabelowreplacementlevelfertilitysince,arethemaindriversofaforecasteddoublingoftheold-agedependencyratiointhecomingdecades(Figure1).Foreverypersonover65therewerefourpersonsofworkingage(20-64)in2010,whileitisforecastedthatin2040therewillbetwopersonsofworkingageforeverypersonover65.ThisageingoftheDutchpopulationburdens(future)publicfinancesasitincreasespublicexpenditureson,forinstance,healthcareandstatepensionswhicharepartlyfinancedthroughtaxationonearningsfromarelativelyshrinkingworkforce(OECD2011;vanEwijketal.2006).Thisburdenonpublicfinancescanbealleviatedby

1Earlyretirementandmandatoryprivate(occupational)pensionschemesareconsideredpartofsocialinsuranceasthesearefacilitatedthroughfavorabletaxationandregulatedbypublicpolicymakerswithpensionlaws.

3

keepingpeopleinworklonger,asitwillincreaseincometaxrevenues,andbyincreasingthestatepensionagetoreducepublicpensionexpenditures.

Overthelastdecades,therefore,socialinsuranceintheNetherlands,asinmanyotherOECDcountries,hasbeenredesignedtocreatestrongerincentivesforcontinuedworkatolderages(Gruber&Wise2004;Wise2012,2016A).Inparticular,asemploymentratesamong55-64yearsoldmenwerenotoriouslylowforseveraldecades(Figures2&3),withanall-timelowofabout20percentformenaged60-64inthemid-nineteen-nineties,afirstbatchofpensionreformsfrom2000onwardsaimedatkeepingworkersemployedbymakingearlyretirementlessattractive(deVos,Kapteyn&Kalwij2018).Previousstudiessuggestedthatthesereformsarelikelytohavecontributedtotheincreaseinlaborforceparticipationofthe55–64populationfromlessthan30percentinthemid-nineteen-ninetiesto45percentin2007(Euwals,deMooij&vanVuuren2009;VanOorschot2007).Asthereformsofearlyretirementschemesmostlyaffectedindividualsbornafter1949(deVos,Kapteyn&Kalwij2018),itsfullimpactcanonlybeassessedwiththedatacurrentlyavailable.Anassessmentoftheeffectsofearlyretirementbenefitschemesontheexitprobabilityfromemploymentisoneoftheobjectivesofthischapter.

Mandatoryprivateoccupationalpensionschemes(fromhereonwardsreferredtoasprivatepensions)intheNetherlandsusuallyincludeearlyretirementoptions,i.e.receivingprivatepensionsbeforethestatepensionage(SPA).

2

Apartfromtheconsequencesofpopulationagingforpublicfinances,decadesofgenerousearlyretirementschemes,thedotcomcrisisandtherecentgreatrecessionwithinitsaftermathhistoricallylowinterestrates,lefttheirmarksonthesolvabilityofpensionfundsandnecessitatedfurtherpensionreformsincludinga

2Thispaperdoesnotanalyzetheroleofvoluntaryprivatepensionsorotherfinancialproductsthatmightbeusedtofinance(early)retirement.

4

gradualincreaseinSPAfrom65to69from2013onwards.

3

AnincreasedSPAwill,byconstruction,reducepublicpensionexpendituresandalthoughitislikelytoalsodelaytheclaimingofprivatepensions,themostimportantimmediatemeasurestoimprovepensionfunds’solvabilityhavebeenthroughashiftfrombenefitsbasedonfinalearningstobenefitsbasedonaverageearningsandareductioninpensionentitlementsandbenefits(Kalwij,Alessie,Gardner&Ali2018).DeVos,Kapteyn&Kalwij(2018,Section4.2)haveshownthatanincreaseinSPAkeepsworkersemployedlonger.

Nexttopensionreformstherehavebeenreformsoftheunemploymentinsurance(UI)anddisabilityinsurance(DI)programs.WhiletheeffectivenessoftheUIreformsforolderworkersmighthavebeenmitigatedbytheintroductionin2007ofanincomesupportprogramforworkersaged60-64,theDIreforms,andinparticularstrictereligibilityrequirements,havesignificantlyreducedtheinflowintoDI(deJong,Lindeboom&VanderKlaauw2011;Kalwij,deVos&Kapteyn2016;Koning&Lindeboom2015).AnobjectiveofthischapteristoassesstowhatextentchangesovertimeinUIandDIbenefits,conditionalonbeingeligible,haveaffectedtheprobabilityofleavingemploymentandreceivingthesebenefits.

Theincreasedemploymentratesatolderagesfromthemid-nineteenninetiesonwardsaftermanyyearsofdecliningratesformenandstagnatingratesforwomenmaysuggestthatthesocialinsurancereformshavebeeneffective(Figures2and3).Figures2and3alsoshow,however,thatemploymentratesbeyondSPAhavemorethandoubledformensince1995andmorethanquadrupledforwomen,despitethefactthatatthoseagessocialinsurancereformsareunlikelytohavehadmuchimpact.AsarguedinKalwij,Kapteyn,anddeVos(2018),thismaysuggesttheimportanceofotherfactorsfortheriseinemploymentatolderages,suchas

3SPAis66andfourmonthsin2019andisexpectedtoincreaseto69andthreemonthsin2040(deVos,Kapteyn&Kalwij2018).

5

improvedhealth,increasedlevelsofeducation,andabettermatchofolderworkers’characteristicswithlabordemand.Oratleast,suchfactorsmayhavecontributedtothesuccessofthesocialinsurancereformsinunlockingtheunusedworkcapacityofolderworkersthatisdocumentedinKalwij,Kapteyn&deVos(2017).Whilewedonotobservethesevariables,wecontrolforhealthonanaggregatelevelusinglifetablesinformationasinparticularthehealthofmenhasimprovedsubstantiallyduringourobservationperiod(seeFigure4).Furthermore,wecontrolforindividualincomeasaproxyforsocioeconomicstatusandskills,whichmightbenegativelyrelatedtotheprobabilityofleavingemployment,asforinstancelowwageworkersfacehigherunemploymentriskthanhighwageworkers.Wecontrolforhouseholdincomeandhomeownershipasproxiesforhouseholdwealth,whichmightbepositivelyrelatedtotheprobabilityofleavingemployment(Bloemen2011).

Therestofthechapterisstructuredasfollows.Section2discussesthesocialinsurancereforms,whichhavebeendescribedinmoredetailinourpreviouschapter(deVos,Kapteyn&Kalwij2018).Inouranalysisweuseonlythemostsalientfeaturesofsocialinsuranceintermsofbenefitsanddurationsofbenefits.Thatis,wedonottakeintoaccountpossiblesector,firmorindividualspecificarrangementsaswedonotobservethese,norhowhouseholdcompositionmightaffectanindividual’sstatepension(SP)orsocialassistance(SA)benefitswhennolongereligibleforDIorUIbenefits.Anadvantageoftakingsuchacrudeapproachisthatourmeasureofsocialinsurancewealth(SIW),onceconditionedonindividualearningsandage,isunlikelytobeendogenoustoindividualandfirmcharacteristicsthatmayalsoinfluencetheemploymentdecision.Inotherwords,wemayassumeourmeasureofSIWisexogenousfortheworker’sdecisiontoleaveemployment,andwhetherthisisbyreceivingUI

6

orDIbenefitsorbyretiringearly.Italsomeansthatthereis,forgivenageandearnings

4

,novariationinSIWwithinacalendaryearsoweexploitmainlyvariationinSIWacrosstimeandage,orrathervariationintheimplicittaxrate(ITAX)calculatedbasedonSIW,toidentifytheeffectsofsocialinsurancebenefitsontheprobabilityofexitingemployment.Section3describesthedatatakenfromthe(Dutch)IncomePanelStudy.Inoursample,employedindividualsarefollowedfromtheagetheyenteredthepanel(55orover)untiltheyleftemploymentorreachedtheageof70(whichevercamefirst).

5

Thedatacovertheperiod1989-2014.Thus,westartobservingworkersbeforetheimplementationofmostofthereformsaimedatkeepingworkersemployedandstopobservingworkersjustafterSPAstartedtorisein2013.Section4presentsourempiricalresultsandsection5discussesthemainfindings.

4Wealsoconditiononhomeownership.ThisvariableisusedtoassesseligibilityforSA,whichismeanstested.Furthermore,therearenogenderdifferencesinthesocialinsurancerules.

5Aworkermayalsohavedied,emigrated,orleftthesampleforotherreasons(seeSection3).

7

Figure1Dependencyratios1970-2040.

Source:StatisticsNetherlands(dataextractedonApril5,2019;statline.cbs.nl).

Notes:Thetotaldependencyratioisdefinedasthenumberofpeopleunder20orover64asapercentageofthenumberofpeopleaged20-64,thechilddependencyratioasthenumberofpeopleunder20asapercentageofthenumberofpeopleaged20-64,andtheold-agedependencyratioasthenumberofpeopleover64asapercentageofthenumberofpeopleaged20-64.

Figure2Men’semploymentratebyage.

100

90

80

70

60

%50

40

30

20

10

0

19701975198019851990199520002005201020152020

YEAR

50-5455-5960-6465-69

Source:OECD(DataextractedonApril5,2019;).

Note:OnlyforthisFigure,employmentincludestheself-employed.

8

Figure3Women’semploymentratebyage.

100

90

80

70

60

%50

40

30

20

10

0

19701975198019851990199520002005201020152020

YEAR

50-5455-5960-6465-69

Source:OECD(DataextractedonApril5,2019;).

Note:OnlyforthisFigure,employmentincludestheself-employed.

Figure4Men’sandWomen’sannualmortalityratesatselectedagesbyyear.

2.5

2

1.5

%

1

0.5

0

198919941999200420092014

Age55,WomenAge55,MenAge60,Women

Age60,MenAge65,WomenAge65,Men

Source:HumanMortalityDatabase(DataextractedonMay2,2019;).

9

2.Socialinsurancereforms1989-2015:Themostsalientfeatures

Socialinsuranceprogramstypicallyconsistofeligibilitycriteria,incomereplacementratesandbenefitdurations.FortheNetherlands,thesearestipulatedinsocialinsurancelaws(UWV2013A,2013B,2013C,2014A,2014B)andpensionlaws(wetten.overheid.nl).AllclaimstounemploymentanddisabilityinsurancesceaseatSPAandearlyretirementisdefinedasretiringbeforeSPA.Fromthatageonwards,allworkersreceiveastatepension,mostoftensupplementedwithaprivate(occupational)pension.

Theimplementationoftheselaws,withinthelegalframework,maydependoncollectivebargainingbetweenunionsandfirms(orgroupsoffirmswithinasector),withoutcomesthatmaybeindividualspecific.Forinstance,privatepensionsmayhavesectorspecificarrangements.Municipalitiesandotherinstitutionsmayhavetheirownincomesupportprogramsforthoseinneed,ontopoftheprogramsoutlinedbelow.Furthermore,forinstance,incaseofaplantclosure,unions,firmsandlocalgovernmentsmaynegotiateredundancyschemes,suchasofferingearlyretirementtoolderworkersinsteadofUI.Finally,someprogramssuchasSocialAssistanceandStatePensiontakeintoaccountindividuals’householdsituation,whichweignoreinthischapter.Hence,individualsmayfacesocialinsuranceprovisionsthatarelessormoregenerousthandescribedbelow.Asdiscussedintheintroduction,weonlyconsiderthemostsalientfeaturesofsocialinsurancebenefitsschemesandbydoingsomanynuancesaremissinginourdiscussions.

Ourrathercrudeapproachmakesitpossible,givenourdata(Section3),tocalculatesocialinsurancewealthforallworkersinoursample;anecessaryinputforourempiricalanalysis(Section4).TableA1presentstheparametersforcomputingsocialinsurancewealthforeachworkerbasedonthediscussionsintheremainderofthissection.

10

2.1Socialassistance(SA)

Universalsocialassistance(SA)hasbeeninplacesincetheearlynineteen-sixties.Itisameanstestedprogramwithbenefitsequalto70%oftheminimumwage(TableA2).WhenclaimingSA,onedoesnotaccrueprivatepension(PP)entitlementsbutonedoesaccrue

(public)statepension(SP)entitlements,simplybybeingaresidentintheNetherlands(seeSection2.5).

UIandDIbecamelessgenerousaround2005(seeSections2.2and2.3).Tosupportolderworkerswitharguablypooremploymentopportunities,anincomesupportprogramwasintroducedin2007thatguaranteesindividualswhoenteredUIorDIattheageof60oroverandwhoseentitlementstoUIorDIbenefitshaverunout,non-meanstestedSAbenefitsuntil

SPA.

2.2Disabilityinsurance(DI)

Between1984and1992,DIbenefitswerebasedona70%replacementrateandlasteduntilSPA(ifoneremainedeligible).Before1984,thereplacementratewas80%.From1993onwards,thebenefitdurationwasshortenedtothreeyearswhenreceivingDIatages53-57,andtosixyearswhenreceivingDIatages58-64.Inresponsetothesereforms,policymakers,unionsandemployersagreedon(supplementary)privatedisabilityinsuranceforworkers,oftenatthesectororfirmlevelwithauto-enrollment,toguaranteework-disabledolderworkersareplacementincomeaftertheDIbenefithasrunoutanduntilSPA.In2006,majorreformstoDItookplacethatmainlyaffectedeligibility.Medicalscreeningbecamestricterandthethresholdforbenefiteligibilitywasincreasedfrom15to35percentreducedworkcapacity.Workerswereonlyentitledtofullbenefitsiftheywere80-100percentdisabled.Inaddition,thereplacementratewasraisedto75%.WhenDIbenefitsend,workersareentitledtoSA.If

11

theyareover60,benefitsarenotmeanstested.WhileonDIonekeepscontributingtotheprivatepensionschemeofthelastemployerbutataloweraccrualrate,asitisbasedonreplacementDIbenefitsandnotlastearnings.Nevertheless,aswasthecasewiththe1993DIreform,after2006supplementaryprivatedisabilityinsuranceseffectivelynullifiedtheeffectsofshorterDIbenefitdurationsonDIuptakefromtheperspectiveofworkers.Hence,conditionalonbeingeligibleforDI,thefinancialincentivestotakeupDIhavehardlychangedsincetheearlynineteen-eighties.

2.3Unemploymentinsurance(UI)

EligibilityforUIcanbetheresultofjobloss,forinstanceduetoafirmclosureorterminationofalaborcontract.TheUIreplacementrateis70%since1985(80%before1985).WhileonUI,onedoesnotaccrueprivate(occupational)pensionbenefitentitlements.

UIbenefitdurationdependsonworkexperience.Before2007,workerswithatleast35yearsofworkexperiencewereentitledtofouryearsofbenefitsandworkerswithatleast40yearsofworkexperiencewereentitledtofiveyearsofbenefits.Workexperiencehashoweveronlybeenregisteredsincetheendofthenineteen-nineties,soitwasassumedthatworkersstartedtheirfirstjobattheageof18.Therefore,before2007,forworkersjustbelowtheageof57.5,UIbenefitdurationequaledfouryearsandforworkersaged57.5orover,thedurationequaledfiveyears.AfterUIbenefitsranout,theunemployedwereentitledtocontinuationbenefits(SAlevel,notmeanstested).Ifaworkerwasbelow57.5whentakingupUI,thedurationofthesecontinuationbenefitswasoneyearbefore1995andtwoyearsfrom1995onwards.IfaworkerstartedreceivingUIattheageof57.5orlater,thisdurationwas3.5years.Summingup,before2004,aworkeraged57.5oroverwhowaseligibleforUI,couldhaveUIandcontinuationbenefitsuntilSPA.Aworkeryoungerthan57.5couldhaveUIbenefitsforfouryearsandcontinuationbenefitsforoneyear(ortwoyearsafter1994),followedbyafew

12

yearsofmeanstestedSAbeforeSPA.Hence,theageof57.5isapivotalageforthefinancialincentivesofclaimingUIatolderages.

6

From2004onwardstherearenomorecontinuationbenefits,butthesehavebeenreplacedin2007bynon-meanstestedSAbenefitsforworkersaged60orover(seesection2.1).Also,before2004theunemployedaged57.5oroverdidnotneedtoapplyforjobs,whilefrom2004onwardstheyhaveajob-searchrequirement.FromOctober2006onwardstheUIbenefitdurationislimited:onemonthforeveryyearworked,withamaximumof38months.

7

2.4Earlyretirement(ER)

8

AnimportanteligibilitycriteriumforERisthatonecanonlytakeupERwhenemployed.Workersenrolledinaprivatepensionschemehaveoftentheoptiontoretireearly,thatistostartreceivingpensionbenefitsbeforeSPA.Whilenowadaysearlyretirement(ER)schemesaredirectlylinkedtotheprivatepension(PP)schemes(seeSection2.6),mostgenerousERschemesthatdidnotaffectworkers’PPentitlementswereinplacesincethelate1970sandgraduallycoveringalmostallworkerstowardstheendofthenineteen-eighties.TheselatterERschemesweretargetedforreformfromthemid-nineteen-ninetiesonwardswhenitbecameapparentthatthesewouldnotbefinanciallysustainableastheywerebyand

6Forouranalysiswesetthisageequalto58asweobserveagesinfullyears(andatDecember31ofeachyear).

7WeassumedthisshorteningofUIbenefitsdurationwaseffectivefrom2007onwards(weonlyobservefullyears)andsetthemaximumdurationequaltothreeyearsforallindividualsaged56andover.From2005onwards,actualworkexperiencehasbeenusedbycaseworkers,butastherearenoofficialadministrativerecordsbefore1995,forolderworkerstheruleofhavingworkedfromtheageof18hasbeenusedbycaseworkers(henceamaximumbenefitdurationentitlementattheageof56).

8SeealsoEuwals,DeMooij,&VanVuuren2009;Euwals&Trevisan2011;Groot&Heyma2004;Lindeboom1996;Nelissen2001.

13

largeunfunded(Jansweijer1996).Until1996,ERschemeswereusuallycharacterizedbyveryhighnetreplacementratesandwithERbenefitsthatlasteduntilSPA.NoactuarialadjustmentsofERorPPbenefitstookplace.Inaddition,duringearlyretirementPPbenefitaccrualcontinuedandthePPreplacementrateremained70%ofgrossearningsafter40yearsofPPaccrual.Typically,before1996ERwasofferedtoworkersaroundtheages58to62.

9

WhenreachingSPA,workerswhohadbeenearlyretiredreceivedPPasiftheyhadcontinuedworkinguntilSPA.ThemainpoliticalargumentforhavingsuchgenerousERschemeswasthatworkerswhomadeuseofitoftenworkedindecliningindustries.WhileunhealthyworkerscouldmakeuseofDI,anequallygenerousoption,thehealthyonescouldmakeuseofER(Jansweijer1996;p.181/182).AnotherappealingargumentintimesofhighyouthunemploymentwasthatERschemeswouldcreatejobsfortheyoung;anargumentstilloftenmadebutwithlittlebasisinfact(Kalwij,Kapteyn,&deVos2010).Thegeneroussystemlasteduntil1996withonlyoneminorreform,whichreducedthereplacementratefortheperiod1993-1996to85%insteadofthe90%before1993.

Gradually,overtheperiod1997-2005,mostERschemeshavebeenmademoreactuariallyfair(beforeSPA)withatypicalgrossreplacementrateofaround70%atage62.ThischangewastriggeredbylegislationthatonlyallowedfavorabletaxtreatmentforactuariallyfairERschemes.ThenewERschemesdidnotallowforPPentitlementsaccrualduringearlyretirement.From2006onwards,workersbornbefore1950facedslightlylessgenerousERschemesbutthemajorERreformaffectedthosebornafter1949.TheselatteryoungercohortshaveERschemesthataredirectlylinkedtotheirPPschemesandwhenaworkerretiresearly,actuarialadjustmentsofERandPPtakeplace.

9Forouranalysisweassumedtheageof60.

14

2.5Statepension(SP)

Thestatepensionage(SPA)istheageatwhichallresidentsintheNetherlandsreceiveastatepension(SP).Since1956,SPAhasbeenequalto65forbothmenandwomen.Asof2013,SPAhasbeengraduallyincreasing(deVos,Kapteyn&Kalwij2018).

10

SPequals70percentoftheminimumwagefromSPAuntiltheendoflife.TheactualSPreceivedmaydeviatefromthisnormasitdependsonthenumberofyearsresidinginTheNetherland,whileforcouplesitdependsontheageofthespouse.

11

2.6Private(occupational)pensions(PP)

PPschemesaredesignedbyfirms,industriesandunionsandlegislatedby(national)pensionlaws.Suchschemesare,therefore,mostoftenindustryspecific.Almostallworkersaremandatorilyenrolledintheiremployers’PPschemesfromtheageof25onwards

12

untilSPA.Before2004andforatypicalPPscheme,PPamountedto70%offinalearningsatSPAifaworkerhadaccruedentitlementsforatleast40yearsinthatscheme.From2004onwards,PPisbasedonaveragelifetimeearnings(notimplementedretroactively),albeitoftenwith

10Thisincreaseisnottakenintoaccountinouranalysis.WesetSPAequalto65in2014(ourlastyearofobservation)astheincreasehadonlybeentwomonthsandweobserveagesinfullyears.

11WeassumedthatallworkerslivedintheNetherlandsfromtheageof15onwardsandareentitledtoafullSP.Wedidnotobservetheyearsofresidence,butshouldhavinglivedabroadcausedretirementincometodropbelowSA-level,thereisincomesupporttoguaranteeaminimumincomeatSA-level.Spousalcharacteristicsarenottakenintoaccountinouranalysisandwetreatindividualsassingles.SeeMastrogiacomo,Alessie,andLindeboom(2004)foradiscussionontheimportanceofthis.

12From2007onwardsitisfromage21.

15

higherannualaccrualratessothatformostolderworkersatthetimeofthisreformthechangemadelittledifference.

13

AsnotedinSection2.4,ERdidnotaffectPPuntil1997.FromthatyearonwardsERschemesgraduallychangedandby2005workersdidnolongeraccruePPentitlementswhileearlyretired.

14

Forworkersbornafter1949,actuariallyfairadjustmentsofPPtakeplacewhenreceivingPPbeforeSPA.

15

Finally,ithasbeenpossiblefrom2004onwardstostartclaimingPPafterSPA,butonlyforabout17%ofemployees(SER2006;SZW2005).Moreover,theseemployeescouldonlydeferpartoftheirPPbenefitsandmostoftenwithanage-limitof67.Hence,veryfewworkerspostponedPP.Inourempiricalwork,wethereforeassumethatallworkersreceivePPatSPA.

3.The