Forschungsinstitut zur Zukunft der Arbeit Institute for the Study of Labor Does the Welfare State Make Older Workers Unemployable IZA DP No 4440 September 2009 Gilles SaintPaul Does the Welfare ID: 204760
Download Pdf The PPT/PDF document "DISCUSSION PAPER SERIES" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
DISCUSSION PAPER SERIES Forschungsinstitut zur Zukunft der Arbeit Institute for the Study of Labor Does the Welfare State Make Older Workers Unemployable? IZA DP No. 4440 September 2009 Gilles Saint-Paul Does the Welfare State Make Gilles Saint-Paul Toulouse School of Economics, Birkbeck College and IZA Discussion Paper No. 4440 September 2009IZA P.O. Box 7240 53072 Bonn Germany Phone: +49-228-3894-0 Fax: +49-228-3894-180 E-mail: iza@iza.org Anyopinions expressed here are those of the author(s) and not those of IZA. Research published in this series may include views on policy, but the institute itself takes no institutional policy positions. The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center and a place of communication between science, politics and business. IZA is an independent nonprofit organization supported by Deutsche Post Foundation. The center is associated with the University of Bonn and offers a stimulating research environment through its international network, workshops and conferences, data service, project support, research visits and doctoral program. IZA engages in (i) original and internationally competitive research in all fields of labor economics, (ii) development of policy concepts, and (iii) dissemination of research results and concepts to the interested public. IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be available directly from the author. IZA Discussion Paper No. 4440 September 2009 Does the Welfare State Make Older Workers Unemployable? This paper discusses the specificities of the labor market for older workers. It discusses the implications of those specificities for the effect of labor market institutions on the employability of those workers. It shows that while unemployment benefits indexed backwards and hiring costs are likely to harm these workers more than the average worker, the converse is true for employment protection, provided it is uniform across workers and not specifically higher for older workers. It provides some evidence on the impact of labor market institutions on older workers by comparing their outcome in the United States and France. It discusses how the welfare state can be reformed in order to improve outcomes for older workers. JEL Classification: J23, J24, J26, J31 Keywords: older workers, labor market institutions, employment, employment protection, welfare state, pensions, retirement Corresponding author: Gilles Saint-Paul GREMAQ-IDEI Université des Sciences Sociales Manufacture des Tabacs Allée de Brienne 31000 Toulouse E-mail: gilles.saint-paul@univ-tlse1.fr olderworkerslessthanyoungerones.Sections2and3discusstheseissuesfromatheoreticalpointofview.Section4compareslabormarketoutcomesforolderworkersintwocountries:arigidone(France),andaexibleone(theUS).Itthendiscussestheroleofrigiditiesbyrelatingtheempirical ndingstotheprecedinganalyticaldiscussion.Section5drawsthepolicylessonsfromourexercise.2Whyisthelabormarketforolderworkersspeci c?Therearethreecharacteristicsofolderworkersthatpresumablyhaveanimpactontheirlabormarket.Themostimportantoneisthattheirremainingcareertimeissmall:theyareexpectedtoleavetheirjobandretirefairlysoon.Thisisincontrasttoayoungerworkerwhocanbeexpectedtoremainwiththe rmformorethanadecade.Aswewillsee,thisobviousfacthasprofoundimplicationsfortheemploymentofolderworkers.Thesecondcharacteristicisthattheirproductivityis(likely)falling.Whilethisisplausibleitisnotobvious,buthasbeendocumentedbyanumberofstudies.Inparticular,Kotliko¤andGhokale(1992)estimateanage-productivitypro leforvariouscategoriesofworkersbydisentanglingaworkerswagefromhismarginalproductivitytheideaisthatdeferredpay-mentsforincentivereasonscreatesuchawedge.Insuchaworld,aworkersmarginalproductivitypro leovertimeisdi¤erentfromhiswagepro le.Theformercanneverthelessbeestimatedbyusingthefactthatundercompe-tition,nonetpro tsshouldbegeneratedbyhiringanadditionalworker.Hence,thepresentdiscountedvalueoftheworkersmarginalproductshouldbeequaltothatofhiswages,andKotliko¤ ndsthatundersomecondi-tionsthiscanbeusedtorecovertheage-productivitypro le.Hisresultsarestriking.Theyimplythatformostcategoriesofworkers,productivitypeaksataroundage45,thenfallstolevelsthatareestimatedtoathirdofthatpeakatage65.Sinceitistypicallyfoundthatcompensationriseswithagethroughoutthelifecycle,these ndingsimplythatrelativetotheirproductiv-2 ity,olderworkersare"overpaid",whiletheywere"underpaid"duringtheirprimeage.This ndingisconsistentwiththetheoreticalliteraturewhicharguesthat rmsusedeferredcompensationasanincentivedevice(Lazear,1990).Otherstudies(surveyedine.g.Skirbekk(2003))relyondirectmea-suringofhowabilityevolveswithage.Theycon rmthe ndingsofKotliko¤andGhokale,althoughbydesigntheycannotexpressthemasamonetaryequivalent.Figure1,takenfromAvolioandWaldman(1994),showshowdi¤erentmeasuresofabilityevolvewithage.Theyallpeakaround20-25,suggestingthatifproductivityincreasesfrom25to45,itisduetoexperienceratherthanability.Third,relativetoyoungerworkers,olderworkershaveahumancapitalwhichismorespeci ctotheircurrentjob/ rm,andlessgeneral,i.e.lesstransferabletoother rmsandothersectors.Thisisbecausetheyhavespentagreaterfractionoftheirlifeacquiringthosespeci cskillsthroughlearning-bydoingrelativetoacquiringmoregeneralonesintheeducationalsystem.Empirically,onewaytomeasurethate¤ectistolookatthewagelossofdisplacedworkersintheirfuturejobs,whichshouldbegreater,thegreaterthespeci ccomponentoftheirhumancapitalinthejobtheylost.Indeed,existingstudies(Ruhm(1991),Jacobsonetal(1993)Cohenetal(1997)RosoliaandSaint-Paul(1998))often ndalargelossforolderworkers.However,suchalossmayalsoreectgreaterrentsratherthangreaterspeci cskills,especiallyinlightoftheaboveargumentthatdeferredcompensationgeneratessuchrentsforolderworkersforincentivereasons.Wereturntothatissueinourempiricalstudybelow.Thesethreekeycharacteristicsofolderworkershaveanumberofimpli-cationsfortheworkingofthelabormarketandespeciallyforhowtheimpactoflabormarketinstitutions,suchasthosewhichprevailinEurope,onthoseworkers.Letusdiscusstheseimplications.Ifworkerswereo¤eringtheirservicesinapurespotmarket,theywouldbepaidtheirmarginalproductateachpointintime.Wewouldthenobservewagespeakingataround45andthenfallingtoreachpossiblymuchlowerlevelstowardtheendofonescareer.However,aspointedabove,thisisnotwhatwesee,sincewagesgoupwithage.Ifweassumethatthisdiscrepancyis3 duetoincentiveproblems,weseethatolderworkerswholosetheirjobloseanimportantrent.Thatis,whiledeferredpaymentsarenotdirectlyallocativeatthelevelofthe rm/workerpair(thetimingofpaymentsisdisconnectedfromthatoftheworkersmarginalproduct),theyarenotneutralwhenoneconsidersmobilitytoanotherposition,whichismorecostlyandthereforemoredeterred,thegreatertheworkerswagerelativetohisproductivityandthereforethegreaterhisage.Thisiscompoundedbythelossofspeci chumancapital,andalsobyathirde¤ect,whichisthatgiventheshortexpectedtenureinanypositionthattheywill nd,thereislittleroomforengagingindeferredcompensation.Insomesense,thatisfortunate,becauseitmakesitlesslikelythattheemployerlowersthewageuponhiringtoincreaseitlater.Atthesametime,italsomeansthattheolderworkersaremorelikelytoendupinjobswhereincentiveproblemsarelessimportant.Thisistruenotonlybecausedeferredcompensationismoredi¢cultbutmoregenerallybecauseitismoredi¢culttousepaytoelicitincentivesforworkerswithshortexpectedtenure1.Overall,thismeansthatolderworkers,unlesstheyarestillemployedata"lifelong"position,aremorelikelytobeemployedinthe"secondary"sectorofjobsthatareeasilymonitoredandthustypicallyinvolvelessautonomyandresponsibility,andalsolowerwages2.Asecondimplicationisthatunemploymentbene tsarelikelytobemoredamagingtotheemploymentrateofolderworkersthattothatofyoungerworkers.LjunqvistandSargent(1998)haveanalysedtheconsequencesforaggregateunemploymentofthefactthatunemploymentbene tsareindexedbackwardonwages.Thisbackwardindexationimpliesthattheunemployedwillbeespeciallypickyifthedistributionofjobo¤ersisassociatedwithlowerwagesthantheprecedingone.Theyshowthatattimesof"turbulence",i.e.intensesectoralreallocation,joblossesareassociatedwithlossesof rm-andsector-speci chumancapital,sothatthewagesthattheunemployedcangetinthenewsectoroftheeconomy,wheretheyareyettolearnthetrade,are 1Thisisageneralpredictionofthe"shirking"modelofe¢ciencywages(seeShapiroandStiglitz(1984)).2SeeSaint-Paul(1996).4 lowcomparedtotheirpreviousones,andthereforelowcomparedtotheirunemployentbene ts.Thisiswhy"turbulence"hasalargeadverseimpactonunemploymentincountrieswithgenerous,backward-indexed,unemploymentbene ts;ontheotherhand,thee¤ectismuchsmallerincountrieswheresuchbene tsarenotgenerous.Theirargumentisespeciallysalientfortheolderworkers.Ourdiscussionaboveimpliesthatevenintheabsenceof"turbulence",thewagestheycanclaimafterhavinglostajobarelikelytobesubstantiallysmaller,becauseofthedeferredcompensatione¤ect,thelossofspeci chumancapitale¤ect,andthegreaterlikelihoodofworkinginthesecondarysector.Furthermore,astheirproductivityfallswithage,thewedgebetweentheirreservationwageasdeterminedbytheirbene tlevelandtheo¤erstheygetislikelytogrow.Thuswenotonlyexpectalowerexitratefromunemployment,butalsoastronger"durationdependence"thephenomenonbywhichexitratesfromunemploymenttendtofallwiththelengthoftheunemploymentspell3.Wewillreferbelowtothise¤ectthefactthatunemploymentbene tspushtheresevrationwageofolderworkersuprelativetothewagestheycanexpectontheirfutureo¤ersastheentitlemente¤ect.Third,hiringand ringcostswilla¤ecttheemployabilityofolderwork-ersdi¤erentlyfromyoungerworkers.Thisisanimportantquestionbecausealargefractionofthosecostsareregulatoryespeciallythe ringcoststhatarecreatedbyemployentprotectionlegislation.Furthemore,employ-mentprotectionispaidspecialattentioninthedebateoverthemarketforolderworkers.Somecountries(e.g.France,asdiscussedbelow)havespe-cialprovisionsthatincreaseemploymentprotectionbeyondacertainage.Furthermore,inmostcasesentitlementsareincreasingwithtenure,whichautomaticallybene tsolderworkers.Ifwe rstconsider ringcosts,weseethattheirdeterrente¤ectonem-ployabilityislowerforolderworkers.Thereasonisthat,shouldthe rmconsiderlayingo¤theworker,itcanwaitforhisretirementinsteadofpay-ingthe ringcost.Thatvalueofwaitingisobviouslymuchhigherifthe 3Unfortunately,myownattemptsto ndsuchastatisticallysigni cante¤ectintheFrenchcasefailed.5 workerisyoung.So,somewhatparadoxically,a ringcostthatwouldbeim-poseduniformlyonallworkerswouldmakeolderworkersmoreemployablerelativetotheyoungerones.Ofcourse,thestoryisentirelydi¤erentif ringcostsgoupwithage.In-deed,mostEuropeancountrieshaveemploymentprotectionlegislationsthatbecomemorestringentasthetenureoftheworkerincreases.Thisislikelytobecorrelatedwithage.Furthermore,anumberofcountriesimposeaddi-tionaltaxesonlayo¤sforolderworkers.Anotableprovision,forexample,istheFrenchso-called"Delalande"contribution,whichimposesanadditional ringtaxonworkerswhoareolderthan50.Followingstandardeconomics,weexpectsuchataxtoreduceboththehiringratesandthe ringratesofthiscategoryofworkers.Furthermore,itisalsolikelythattheexpectationofthetaxwillreducethehiringrateforworkersyoungerthan50,andisalsolikelytoincreasetheirlayo¤rateas rmsmaywanttoanticipateonafuturelayo¤decisionconcerningthoseworkerstoavoidpayingtheadditionaltax.Behageletal.(2008)usechangesintheDelalandetaxtoestimateitse¤ectonthehiringrateofolderworkers.Inparticular,in1992thesystemwaschangedsoastoexempt rmsfrompayingthetaxwheneverworkerswereolderthan50atthetimetheywerehired.Clearly,thiswasmeanttoo¤setthenegativee¤ectsofthetaxuponhiring;butatthesametimeitmeantthatthosehiredafter50lackedtheextraprotectionoftheothers.Thiswouldimplythatworkershiredafter50wouldbelaido¤beforeworkersofthesameagehiredpriortothatage,thusreinforcingthedualnatureofthelabormarket.Inanycase,theestimatesofBehageletalbasedonthatnaturalexperimentcon rmthepresumptionthatimposingextraprotectiononaselectedgroupofworkershasasubstantialnegativeimpactontheirhiringrate.They ndthatthereformincreasedthejob ndingrateofthe50+group,relativetoyoungerworkers,by0.5percentagepointsonamonthlybasis.Thisisalargee¤ect.Forexample,withamonthlyjoblossrateof0.1%andjob ndingrateof1.5%,wegetanunemploymentrateinsteadystateequalto0.1/1.6=6.25%.Ifwenowreducethemonthlyjob ndingrateby0.5%,theunemploymentratejumpsto10%.Thee¤ectisacompoundof"direct"e¤ectsthereturnfromhiring6 somebodyisloweriftherearegreater ringcostsandsubstitutione¤ectsIpreferthelow ringcostapplicantthehighone.Whileitisdi¢culttodisentanglethetwokindsofe¤ects,theauthorsresultssuggestthatstrongsubstitutionisatwork:thereformnotonlyreducedthegapinhiringratebetweenth50-andthe50+agegroups,itinverteditssign:afterthereform,itwaseasierto ndajobforaworkeraged51thanaworkeraged49.Finally,theauthorsarealsoabletoestimatethee¤ectoftheDelalandetaxonlayo¤s;they ndthatitreducesthelayo¤rate.Theyalso ndaslightpositivee¤ectonthelayo¤ratebeforeage50butitisverysmallandstatisticallyinsigni cant.Turningnowtohiringcosts,weseethattheyhaveastrongere¤ectonaworkersemployability,theolderthatworker.Thisisbecausethehiringcostislessdamagingtothe rm,thelongerthedurationofthejoboverwhichthe rmcanrecoupit.Clearly,thatdurationislowerforolderworkers.Thusauniformhiringcostsreducedthedemandforolderworkersrelativetoyoungerones.Thelowerexpecteddurationofthejobreducesthetotal(discounted)valuetothe rmofemployingtheworker:thisiswhatwewillcalltheendgamee¤ect.Asaresultthe rmiswillingtopaylesstoemploytheworker,implyingthatolderworkersaregoingto ndjobsinsectorswheretrainingandrecruitmentcostsarelow.Below,Iwilldocumenttheendgamee¤ectbothinthecontextofasmalltheoreticalmodelandofanempiricalcomparisonbetweenFranceandtheUnitedStates.Thisisinaccordancewiththerestoftheliterature.Forexample,inarecentpaper,Hairaultetal.(2008)capturethee¤ectsofhiringcostsinasearch-matchingmodel,andtheydocumentapositiverelationshipbetweentheageofretirementandtheemploymentrateofolderworkers.ThispositiverelationshipisdepictedonFigure2.Theysupplementthiswitheconometricevidencebyconstructingaworker-speci cvariablewhichmeasuresthedistancetoretirementandshowthatthisvariablehasapositivee¤ectontheprobabilityofbeingemployedinaFrenchdataset,controllingfortheusualvariables.7 3LessonsfromasimplemodelWhilesomeoftheaspectsdiscussedaboveareeasytograsp,aformalmodelmaybeusefulinordertounderstandtheinterplaybetweenhiringand ringcostsandtheendgamee¤ect.Inthissection,Iconsidersuchamodel.AssumetimeiscontinuousandthatthedurationofhumanlifeisT:Whenhired,workersareinahighproductivitystateandproduceaowofoutputy=yH:Letsdenotetheworkersage.Withprobabilityp(s)perunitoftime,theworkersfallsintoalowproductivitystatesuchthaty=yLyH:Theworkerremainsinthatstateuntilheleavesthejoborretires.Inprinciplep(s)canvarywithage,andcouldreectthehump-shapeddependenceofproductivitywithrespecttoagedocumentedbyKotliko¤andGhokaleandothers.Wagesare xedandequaltow:Tohireaworkera rmmustpayahiringcostH:Furthermore,a ringcostFmustbepaidandweallowittobetime-dependent.Morespeci cally,the ringcostisassumedtobeequaltoF0forssandF1F0forss:IfF1=F0weareinthespecialcaseofa ringcostwhichisindependentofage.Firmsmaximizethepresentdiscountedvalueofpro ts,underaconstantrealinterestrateequaltor:Theyhaveworkersofdi¤erentagesarrivingrandomly,andforeachworkerdecidewhetherornottohirehim.Clearlygiventheconstantreturnstoscaleimplicitinourassumptionsregardingtheprocessforproductivity,allthesehiringdecisionsareindependentfromoneanother.Wewanttocharacterizethehiringanddismissaldecisionsof rmsde-pendingontheageoftheworker.Fortheproblemtobeinteresting,itmustbethatyLwyH:Otherwise,onewouldeithernothireanybodyornot reanybody.Letusthenassumethatthisinequalityholds.We rstcharacterizethedismissaldecision.Foraworkerofages-277;ssuchthaty=yL;the rmcankeephim8 untilretirementwhichyieldsa(negative)presentdiscountedpro tequaltoK(s)=ZTs(yLw)er(ts)dy=yLw r1er(Ts):(1)Ontheotherhand,the rmcangetridoftheworkerrightaway,whichinvolvesanimmediatecostequaltoF1andthusanetpro tequaltoF(s)=F1:Thusthe rmwillkeeptheworkerifandonlyifwyL r1er(Ts)F1:Foraworkerinthelowstatewithagess;thetwooptionsarethesame:keeptheworkeruntilretirementandgetapro tK(s)givenbytheRHSof(1),vs. retheworkerrightawayandhaveapro tF(s)=F0:Theintermediaryoptionofkeepingtheworkertogetridofhimimmediatelypriortotheriseofthe ringcostatagescannotbeoptimal.Itwouldyieldanetpro tequaltoW(s)=Zss(yLw)er(ts)dyF0er(ss)=yLw r1er(ss)F0er(ss)=F0+F0+yLw r1er(ss):(2)Clearly,forthatoptiontodominateimmediatedismissal,wewouldneedthatW(s)F0;implying,byvirtueof(2),thatF0wyL r:ButsincewyL risthepresentdiscountedvalueoflossesforalowproductivityworkerthatwouldstayforever,thislatterinequalityinturnimpliesthatthe rmwillbebetter-o¤keepingtheworkeratagesratherthangetridofhim.Inshort,iftheowequivalentoflossesisgreaterthanthe ringcost,thenimmediatedismissalisoptimal;otherwise,waitinguntiltheworkerretiresisoptimal.Theintermediateoptioncannotbeoptimalunlessthelossesthemselvesaretime-varyingwhichwehaveruledoutinourmodelforsimplicity.9 Figure3plotsthe"dismissal"frontierFFinthe(s;yL)plane:itgivesthelowestvalueofyLsuchthatthe rmpreferstokeeptheworker,asafunctionofage.Itisdownardslopingandhasaverticalasymptoteats=TandyL=1;reectingthefactthat rmsaremorelikelytowait,theoldertheworker.Thestepats=sreectstheincreasein ringcostswhenonepassesthatagethreshold.TheintersectionofthisfrontierwithahorizontallineattheactualvalueofyLdeterminesacriticalage,~s;afterwhichtheworkerwillnotbe redifhefallsinthelowproductivitystate.Conversely,allworkersyoungerthanthatwillbe rediftheyfallinthelowproductivitystate.Thecriticalagemaybelargerthan,equalto,orsmallerthan,theregulatorythresholds:Itwillbeexactlyequaltoitforawholerangeoflowproductivitylevels,duetotheverticalportionofFFats=s:TighteningemploymentprotectionlegislationincreasingF0,F1;orbothshiftsthedismissalfrontierdownandlowersthecriticalage.ThisisillustratedonFigure4.Whilethemodelrulesoutany"preventive"dismissalinanticipationofthethresholdforlow-productivityworkers,thismayhappenforhigh-productivityworkers.InAppendix1Isolveforthe rmsmaximumex-pecteddiscountedpro tinthecasewherepisconstantandintheregimewhere~ss:ItisshownthatforF0smallenoughandF1andwlargeenoughwithintheparameterzonecompatiblewiththisregime,itisindeedoptimalto rehighproductivityworkersimmediatelybeforetheyreachthecriticalages:Thenextstepistocharacterizethehiringdecision.Todoso,IneedtocomputeJH(s);thevaluetothe rmofemployingaworkerofagesinthehighproductivitystate.Thisinturngivesusthe"employabililty"ofaworkerofages:IfitisgreaterthanthehiringcostH;thenanapplicantofageswillbehired,otherwisetheapplicantwillbeturneddown.Therefore,thegreaterthevalueoftheworkerJH(s);themoretheworkerisemployable.ThevariationofJH(s)withsisanalysedintheAppendix,andtherearethreepossibilities:1.ThesimplestcaseiswhenJH(s)fallswithsalongthelifecycle.This10 isthecasedepictedonFigure54.Workersarealwayslessemployablewhentheygetolder.Typically,thisismorelikelytooldwhenpissmall,yHislarge,F1issmall,andwyLissmall.Firingcoststhendonotplayabigroleinthe rmshiringdecisions:oneisnotverylikelytofallintothelowproductivitystate,thelossesmadeinthatstatearenotverylarge,thepro tsmadeinthehighproductivitystatearelarge,andthelevelofthe ringcostislow.Thepro leofemployabilitythenresemblestheonethatwewouldgetintheabsenceofemploymentprotection;itisdecreasingsimplybecausewhentheworkerisolderthe rmexpectstoreappro tsfromtheworkeroverashorterperiodoftime.2.InthesituationshownonFigure65,employabilitygoesupuntilitreachesamaximumage^s;whichisgreaterthan~s;meaningthatthemostemployableworkersarenot redwhentheyfallintothelowproductivitystate.Thissituationprevailsif ringcostsareimportant,thusifpislarge,F1large,yHsmallandwyLlarge,andalsoiftheincremental ringcostforolderworkers,F1F0;issmall.Insuchasituationthepossibilityofhavingtoine¢cientlyretaintheworker,orgetridofhimandpaythe ringtax,shouldproductivityfall,playsanimportantrole.Thisgivesapremiumtoolderworkerswhoarenotexpectedtostaylong,andthisinturn,throughforward-lookingexpectations,generatesanupward-slopingpatternofemployabilityuptoage^s:Theprocesshaslimits,though,sincehighproductivityworkerswhoaretoooldstaytoolittletimeinthe rmtogeneratealargeNPV:thevalueofthejobeventuallyfallstozeroassincreasesbeyond^s:3.InthesituationdepictedonFigure7,employabilityisM-shaped6.Thatis,itfallsuntilonereachesthethresholdages;thengoesupuntilage^s;andthenfallsagain.Thissituationprevailsinthesamecircumstancesastheprecedingone,exceptthatthegapF1F0isnowlarge.Thismeansthatworkersgetlessandlessemployablewhentheyapproachthethreshold. 4This gurehasbeendrawnforT=50;s=35;w=1;yH=2;yL=0:5;F0=0:5;F1=1:5;r=0:05;p=0:1:5This gurehasbeendrawnforT=50;s=35;w=0:8;yH=1:5;yL=0:5;F0=2:8;F1=3;r=0:05;p=0:1:6This gurehasbeendrawnforT=50;s=35;w=0:8;yH=1:5;yL=0:5;F0=0:5;F1=3;r=0:05;p=0:1:11 However,afterthat,ageisagainapositivefactorsinceitreducesthelikeli-hoodofhavingtopaythe ringcost.Butthisisagaineventuallydefeatedbythevanishingoftheindividualsremainingprofessionallife.Finally,itisalsousefultocomparethepro leofemployabilityfordi¤erentlevelsofemploymentprotection.ThisiswhatisdoneonFigure8,wherethreesituationsarecompared:a"exible"onewithno ringcost,a"rigid"onewithauniform ringcostequalto150%oftheannualwage,andanasymetricalonewitha ringcostequalto100%ofthewagebefores=35andto200%ofthewagetherafter.Thegraphcon rmsthatrigiditiesreduceemployabilitymostlyforyoungandprime-ageworkersratherthanolderworkers,andthatthestepreducesemployability,relativetoauniformsystem,aheadofthedatethesteptakesplace(bysome veyearsinthissimulation).4Howdoolderworkersfareinaexiblevs.arigidlabormarket?Anempiricalcom-parisonbetweenFranceandtheUS.InthissectionIperformanempiricalcomparisonoftherelativeemployabilityofolderworkersina"exible"labormarket(theUnitedStates)anda"rigid"one(France).Tokeepthediscussionshort,Ionlyconsiderthosedimensionswherethetwocountriesaresigni cantlydi¤erent.However,acompletesetofgraphsandtablesisavailableintheworkingpaperversionofthisarticle.4.1Employmentrates:theendgamee¤ectThe rstvariableIlookatistheemploymentrate.Thisvariablereectsbothlaborsupplyandlabordemand,andthereforecapturesdi¤erentforcesfromindicatorslikeunemploymentorunemploymentduration,whichtellussomethingaboutthefunctioningofthelabormarket.Anemploymentratecanbeloweitherbecauseunemploymentishighorbecauselaborsupplyislow,andthelattercanbeloweitherbecauseofdistortionsorbecauseoftaxes.Ontheotherhand,itisoftenarguedthattheincentivestoregister12 asunemployedarelowifunemploymentbene tsarenotgenerousenough,whichblursthedistinctionbetweenbeingunemployedvs.outofthelaborforce.Wecannotsolvetheseissuesherebutwecanlookatbothindicators.Cohenetal.(1997)haveshownthatforsu¢cientlynarrowlyde ned(bysexandeducation)cells,employmentratesareparadoxicallyslightlyhigherinFrancethanintheUSforprime-ageworkers.Thus,mostofthelowerperformanceofFranceoverallcomesfrom(i)nonprime-ageworkers(ii)thecompositione¤ectduetolowereducationallevelsinFrance.Morerecentdatafromthe2005USCPSandthe2005Frenchlaborforcesurveycon rmsthose ndigs:employmentratesaremarginallyhigherinFranceforprime-agedmen.Butforolderworkers,thepictureisreversed.Whileemploymentratesareunderstandablymuchlowerinthe60-65agerangesincetheo¢cialretirementageinFranceis60theyarealsolowerinthe55-60range7.ThisisreportedinTable1:employmentratesforthosecatgoriesofworkersarealllowerinFrancethanintheUS1,andthecategorieswheretheolderworkersareleastemployedinFrancerelativetotheUS,arelowandmedium-skilledmen.Tohavesomeclueabouthowtointerpretthese ndings,wecanrefertothemodelabove.Itimpliesthatemployabilityfallssharplyasonenearsretirementage(seeforexample gure8).Thustheloweremployabilityofthe56-60agegroupinFrancecomparedtotheUSisprobablyduetotheearlierretirementageinFrance.Whenthatagewasbroughtdownto60thee¤ectonemploymentforpeopleimmediatelybelowthatagewasnottakenintoaccountthesenumberssuggestitissubstantial.Wehavealsoseenthatemploymentprotectionlegislationrunsintheoppositedirection,bymakingtheolderworkersrelativelymoreemployable.Thusthenumberssuggestthatthise¤ectisnotstrongenoughtoo¤settheendgamee¤ect. 7Forwomen,thepictureismorecomplex.Forhigh-schooldrop-outs,employmentratesaresubstantiallyhigherinFrance,implyingthateventhe55-60rangehasahigheremploymentratethanitsUscounterpart.Thesameistrueforcollegegraduates,butthatisnowduetothefactthatemploymentratesarevirtuallyidenticalacrossallactegories,exceptofcoursethe60-65agerange.Forhighschoolgraduatesandcollegedropouts,thepatternisthesameasformen:higheremploymentratesforprime-ageworkersinFrance,butmuchloweronesforthe55-60range.13 4.2Unemployment:themid-lifediscountWenowturntounemploymentrates(Table2).These guresarefairlylowforolderworkers;forworkersolderthan60inFrance,veryfewareactivelylookingforajobandunemploymentrates(notreported)areverylow.Themoststrikingfeaturehereisassociatedwithmatureratherthanolderworkers:ThereisasharpriseinunemploymentrateformenwithatleastahighschooldegreeinFranceduringtheirforties.Wewillrefertothisasthemid-lifediscount.ThisphenomenondoesnottakeplaceintheUnitedStates.Itismostsalientfortheintermediateskillsgroup,kickinginat41forhighschoolgraduatesand46forworkerswithsomecollege.Itismilderandchieylimitedtothe41-46agegroupforcollegegraduatesandvirtuallynon-existentforhighschooldropouts.SincethemidlifediscountoccursinFranceandnotintheUS,itisnaturaltoassumethatitisduetolabormarketrigidities.Whileinstitutionssuchastheincreaseinemploymentprotectionwithtenureandtheexistenceofastepatacertainagemayplayarole,abiggerrolecanpresumablybeascribedtotheentitlemente¤ect,aswehavediscussedabove.At45workersenjoytheirpeakproductivity,whichincludesagreatdealofjob-speci chumancapital,andinadditionderivelargerentsfromdeferredcompensation.Thustheentitlemente¤ectislikelytobelargeforthisgroup.Thishypothesisissomewhatcon rmedbythefactthatthephenomenonisabsentforhigh-schooldropouts.Theyaremorelikelytoworkineasilymonitoredroutineornoncognitivetasks,whichreducestheroleforincentivepaymentschemes,includingdeferredcompensation.Andthescopeforlearningbydoinginthosejobsislower,whichsuggestsaatterevolutionofproductivityduringonescareer.Therefore,atmid-lifetheseworkerswagesarelikelytobelessdi¤erentfromtheirmarginalproduct(inboththeircurrentandfuturejobs)thanforotherworkers,whichmakestheentitlemente¤ectofbackwardindexationofunemploymentbene tslesssalient.Interestingly,themid-lifediscountdoesnotseemtoa¤ectwomen.In-stead.theyexperienceabnormallyhighunemploymentrateinFrancecom-paredtotheUSduringtheirfertileyears,between25and45.Thisispresum-14 ablytheresultofgenerousprovisionsformaternityleaveinFrance,whichreducesthedemandforwomeninthoseages.Thee¤ectmaybesostrongastodwarfthemid-lifediscount.Butthemid-lifediscountisalsolikelytobesmallerforwomenonaverage,sincetheyarebothlesslikelytoengageincontinuouscareers(thusacquiringlessspeci chumancapital)andmorelikelytoworkinthe"secondary"sector.4.3LabormarkettransitionsTheprecedingdatagivesusastaticpictureofthelabormarket.Wealsowanttoknowhowtheolderworkersfareinadynamicsense.ForthiswecomparethetransitionratesbetweenFranceandtheUS.Astylisedrepresentationofthesetwoeconomiesholdsthatbecauseofrigidities,boththejoblossandthejob ndingratesarelowerinFrancethanintheUS.Thisisindeedwhattheearlierliteraturefound,e.g.Cohenetal.Howeveritalsofoundthatsomegroups,liketheyoung,wereusedasabu¤erofexibilityandthattheirjoblossrates,inparticular,weremoresimilartotheUS.Wewanttoknowifthisstylizedvisionholdsformorerecentdata,andinparticularhowtheolderworkersfare:aretheyusedina"exible"orina"rigid"way.Oneissueisthatlabormarketowsarequitesensitivetothebusinesscycle.Andthebusinesscyclewasnotthesamein2005inFranceandtheUS.WhileunemploymentwasfallingintheUS,itwasstillrisinginFrance;itstartedtofallin2006.Therefore,weperformthecomparisonoftransitionratesfortwomoresimilaryears,keeping2005forFrancebutusing2002fortheUnitedStates.I rstdiscussaggregatetransitionratesbyageandthendisaggegatethembysexandeducation.Istartwiththejoblossrate.Inrelationtotheabovediscussionaboutthemeasurementofemploymentandunemployment,therearetwowaystomeasureit:onecanusetheowfromemploymenttounemployment,oralternativelytheowfromemploymenttonon-employment(thesumofunemploymentandnonparticipation).Ideally,the rstoneshouldcapture15 involuntaryjoblosses,andthesecondoneaddsvoluntaryones.Inpractice,thedistinctionisblurredbythehighincentivestoregisterasunemployedinFrance(eveninthecaseofavoluntaryquits),andthelowonesintheUS.SoIlookatbothmeasures.Figure9depictstheemployment-to-unemploymentowinbothcountries.AstrikingfactisthatthismeasureofthejoblossrateisnotlargerintheUS.Themeasuresaresimilarforprime-age,higherinFrancefortheyoung,andlowerfortheolderworkers.Thissuggeststhatolderworkers,despitereformslikethe1992adjustmenttotheDelalandecontributionmentionedabove, rmlyremainintheprotectedsector.Incontrast,theyoungnowhaveajoblossrateevenhigherthanintheUnitedStates.Figure10showsthejoblossrate,includingtheemployment-to-nonparticipationow.Thepictureissomewhatmoreconsistentwithconventionalwisdom:thejoblossrateishigherintheUSforprime-ageworkers;itissimilarinbothcountriesfortheyoung;anditisnowhigherinFranceforolderworkers.Thisisclearlyduetotheowtononparticipation,andthereforermostlycapturestheroleofretirementandearlyretirementschemes.Thisisnotsurprising,butthenumbersaretelling:theysuggestthatforthe56-60agegroups,whohavenotreachedtheo¢cialretirementageyet,theseschemesaccountforsome15%oftherelevantworkforceretiringeachyear,ascomparedto10%intheUnitedStates.Inowturnonthejob ndingrates,thatarereportedonFigure11.Thedataarestriking.Overall,job ndingratesaretwentypercentagepointsgreaterintheUnitedStatesthaninFrance,say50%vs.30%yearly.Andthegapwidensastheworkersgetolder,risingtothirtypercentagepointsforthe51-55agegroup,whilethejob ndingratefallstoaverysmall5%peryearforthe56-60agegroupinFrance.Thisleadsustoreconsidersomewhattheexistenceofthemid-lifedis-count:itdoesnotshowupaslargeunemploymentforolderworkersbecausetheincidenceofunemployment(theemployment-unemploymenttransitionrate)amongthoseworkers,islow.Butifwelookatthedurationofunem-ployment(theinverseofthejob ndingrate),we nditisextremelylowforthoseworkers;andwealsohaveseenthattheincidenceissomewhat16 arti callylowbecauseofthelargerateatwhichtheywithdrawfromtheworkforce.Boththesupplysideandthedemandsideconspireingeneratingalowjob ndingrateforolderworkers:onthesupplyside,thereistheentitlemente¤ect.Onthedemandside,wehavetheendgamee¤ect.Notehoweverthatifweusethejob ndingrateofthe60-65agegroupintheUnitedStatesasameasureoftheendgamee¤ectinaexibleeconomy,wedo ndthatitexists,howeveritaccountsforsayareductioninjob ndingratesfrom55%ayearto40%ayear.Thisissubstantial,butnotcomparabletothemassivecollapseinjob ndingrateexperiencedbyFrenchworkersastheyreach55.Thegapisnotlikelytobeexplainedbytheentitlemente¤ectalone:itisprobablyaslargeforthe46-50group(otherwisethemid-lifediscountwouldnotarise),andyetthisgrouphasamuchhigherjob ndingrate.Anadditionalfactorprobablycomesfromtheminimumwage.Onecouldtrytotestforthisbylookingatmoredisaggregateddata,howeverthatisdi¢cultforthejob ndingratesincethereareveryviewunemployedworkersinmanysexxagecells.Butthescantevidencethatwehavesuggeststhatthejob ndingratesareespeciallylowfortheolder,lowskilledworkers.Indeed,intheFrenchlaborforcesurvey,thevastmajorityoftheunemployedabove50arehigh-schooldropouts.Forthesmallnumberofolderunemployedworkerswithgreaterskills,thejob ndingratesaretypicallysubstantiallyhigher,althoughtherearesofewobservationsthatweshouldbecautiousindrawinganyconclusion.Overall,thesedatagiveapictureofasubstantialexcessprotectionforjoblossinFranceforolderworkers,withthecounterpartofextremelylowjob ndingratescomparedtoothergroups.4.4WagesWenowturntotheanalysisofwages,startingwiththeevolutionofwagesoverthelifecycleinbothcountries.Figure12depictsthetimepro leofwages,foreacheducationalgroup,17 inFrance8.We ndatypicalpatternofwagesrisingwithage,andtherateofincreaseshowsnosignofslowingdown.Oneexceptionishighschooldropouts,whohaveaprettyatpro leandexperienceaslightfalltowardtheendoftheircareer(consistentwithmyargumentaboveregardingtheentitlemente¤ectbeingweakerforthoseworkers).Figure13showsthewagepro leforeacheducationalcategoryforthe"exible"US.Weseethatthereisahump-shapedpatternandthatitistypicallymorepronounced,themoreeducatedtheworker.Infact,forhigh-schooldropouts,wagesareessentiallyatovertime:thereseemstobenohumancapitalaccumulation,andnodepreciationthereof.Thissharpcontrastbetweenthetwocountriessuggestthatwage-settinginstitutionsinFranceleadtoreturnstoagethataresubstantiallytoohighrelativetothemarketoutcome:wagesshouldeventuallybefalling,despitethetendenciestodeferredcompensationandhumancapitalaccumulation,andinaccordancewiththedirectevidenceonproductivity.Thefallshouldstartataround50(whichispreciselytheagewheretheDelalandecontri-butionkicksin).ButinFrancetheincreasecontinues.Forexample,intheUS,malecollegegraduatesaged61-65earn10%lessthanthoseaged45-50.InFrancetheyearn33%more.Takenatfacevalue,thissuggeststheyare43%overpaid.Consequently,oneneedsaverylargeincremental ringcosttoo¤setthehighincentivesof rmto rethem,andonewouldneedasharpreductionintheirunemploymentbene treplacementratiotobringtheirreservationwageinlinewithwhattheycouldearninafuturejob.4.5WagelossesAlotofthediscussionabovearoundthee¤ectsofunemploymentbene tsandtheirretrospectiveindexationonwagesrevolvesaroundtheexistenceofhighrentsforolderworkers.Wehavearguedthatthoserentsmayoriginateintheirdisproportionateaccumulationofjob-speci chumancapitalaswellastheuseby rmsofdeferredcompensationinordertoelicitincentives.Thereareothersourcesofrentsaswell:Theymayoriginateinacollective 8Forspacereasonthedataareagaincon nedtomen.Asimilarpatternarisesforwomen.18 bargainingstructurethatwouldimposeanagepro leforearningsdiscon-nectedfrommarketforces.Or,itmaybethatolderworkershavehadmoretimetoacquirerents,forexamplebyinvestinginsecuringajobintheunion-izedsector.InthissectionIprovidesomeevidencebylookingatthewagelossesofdisplacedworkers.Intheory,therentofaworkershouldbede nedasthe(expected)presentdiscountedvalueofhisincomestreamunderhiscurrentjob,minushisex-pectedpresentdiscountedvalueofincomeshouldhebecomeunemployed.Underacompetitivelabormarket,thedi¤erencebetweenthetwoshouldbezero.Inpractice,tomeasurethatwewouldneedtoknowthedetailsoftheprocessesgoverningincomebothunderemploymentandunderunemploy-ment.Whilethisisnotinprincipleunfeasible,itisrepletewithproblems,soweuseinsteadthedi¤erencebetweentheageearnedfollowinganunem-ploymentspellandthewageofsimilarworkerswhohavenotexperiencedthatspell.Thisisonlyanimperfectmeasureoftherentbecauseitrestsonthewagesofthoseworkerswhodo ndjobs.Thistendstoreducethewagelossascomparedtotherent,sincewageo¤ersthatarebelowthereservationwagearerejected,whichtendstoincreasethedurationofunemployment,thusdepressingthevalueofbeingunemployed.Thusthetruerentmaterial-izesnotonlyinthefuturewagebutalsointhedurationofunemployment,andthatpartwhichclearlyisa¤ectedbylabormarketinstitutionsisnotreectedinourmeasureofwageloss.Nevertheless,ifanythingthisleadsourmeasuretounder-estimatetherentinthe"rigid"(=highunemploymentduration)countrycomparedtotheexible(=lowunemploymentduration)country.Thusifwe ndasubstantialwagelossforolderworkersintherigidcountry,thisisanunderstatementoftheproblemgeneratedforthoseworkersbythebackwardindexationofunemploymentbene ts.Toestimatethee¤ectofage,IestimateanearningsfunctiontowhichIaddagedummiescrossedwithadummyequalto1iftheworkerhasexpe-riencedanunemploymentspellintheprecedingyear.Twocommentsareinorder:First,thisapproachconstrainstheestimatedwagelosstobethesameacrosseducationalcategories;thisisadrawbackbutitsavesondegreesoffreedom.Second,theestimatesmaybebiasedifthereisacorrelationbe-19 tweenunobservedabilityandthelikelihoodoflosingonesjob;tocorrectforthistheliteraturehasusuallylimitedtheanalysistojoblossesthataresupposedlyexogenoustotheworkersunobservedability,suchasthosedrvenbyplantclosing.Here,however,thiswouldsubstantiallyreducethenumberofavailableobservations.Notehoweverthatmycomparisonbetweentwocountriesremainsvalidaslongasthebiasesinducedbyunobservedhetero-geneityareassumedtobethesame;furthermore,theestimatesarequitesimilartothoseoftheliterature,suggestingthebiasissmall9.Table3reportstheresultsformenandwomen.InbothFranceandtheUnitedStates,theestimatedwagelossissubstantialandofthesameorderofmagnitudeasfoundbytheearlierliterature.Thelossissigni cantlyhigherintheUnitedStatesthaninFrance,whichisatvariancewithsomeoftheearlier ndings10.Finally,therentseemsincreasingwithageinFrance,butintheUSitishighestforthe41-50agegroup.Inanutshell,Table3tellsusthatrents(asmeasuredbywagelosses)aregreaterintheUSthaninFrance,andthattheyrarehump-shapedwithageintheUSbutincreasingwithageinFrance.Howcanwemakesenseofthoseresults?Inbothcountries,rentsaregreaterforolderworkersoverall.Thisisconsistentwithourdiscussionaboveasboththee¤ectofdeferredcompensationandthatofspeci chumancapitalaccumulationtendtogenerateanupwardpro leofrents.Thedi¤erencesbetweentheUSandFrancearemoredi¢culttounderstand.Ifonlythesetwomarketforceswerepresent,wewouldexpecttogetthesameresultsinthetwocountries.Rigiditiesobviouslyintroducedi¤erencesbetweenthetwocountries,butitisnotstraightforwardtopredicttheire¤ect11.One 9Itshouldbeaddedthattheplantclosingapproachhasproblemsofitsown:Totheextentthatwageshaveaplantspeci ccomponent(whichwouldbetrueifwageformationobeyedsomerent-sharinglogic),theyarelikelytobecorrelatedwiththelikelihoodofplantclosing.Andsuchacorrelationwillalsoariseifworkersofsimilarunobservableabilitytendtoworkinthesameplants,asthetheoryofassignmentpredictsundercomplementaritybetweenworkerquality.10Notably,Cohenetal.(1997).11Notethatthetheoreticalpredictionregardingthecomparisonofwagelossesbetweenrigidandexiblecountriesisambiguous.Ontheonehand,greatergenerosityofunem-ploymentbene tspushedthereservationwageup:thistendstoreducethewagelossfromdisplacement,andatthesametimetolengthenthedurationoftheunemploymentspell.20 interpretationisthatintheUS,wherewagesaremoreinuencedbymarketforcesandwageinequalityisgreater,thereturntospeci chumancapitalisgreater,implyinggreatere¤ectsonwageswhensuchhumancapitalislostordepreciated12.Atthesametime,collectivebargaininginFranceforcesanupwardpro leofrentswithage.Onnet,thisgeneratesgreaterrentsintheUSexceptfortheagecategorywherethelattere¤ectisstronger,i.e.theolderworkers.5ConclusionandpolicyperspectivesAboveIhavearguedthatanumberoffeaturesofEuropeanlabormarketinstitutionsareparticularlyharmfulfortheelderly.Thisbringsthekeyquestion:howshouldpolicyhandletheseissues?Atemptinganswerwouldbetoincreaseemploymentprotectionfortheelderly,whichhasindeedoftenbeendone.Whyisthattempting?Becausetheendgamee¤ectisafundamentalcharacteristicofthelabormarketfortheelderly.Onecanreduceitsstrengthbyeliminatingregulatorycomponentsofhiringcostswhateverthejusti cationforsuchcosts,theexistenceoftheendgamee¤ecttellsusthattheyparticularlyharmolderworkersandsuggestssuchregulationsshouldbealleviatedforthem.Butthebulkofhiringcostsareintrinsicandnotgeneratedbyregulation.Thusitistemptingtosaythatnoderegulationofhiringcostswillworkandthattheonlythingonecandoistohavehigheremploymentprotectionforolderworkers.Yetthisisfarfromsatisfactory,foratleastthreereasons.First,wehaveseenabovethattherearereasonstobelievethatdi¤erentialemploymentpro-tectionprovisionsgeneratestrongsubstitutionse¤ects.Second,thiswould Ontheotherhand(thiswouldbeapparentfromadynamicsearchmodel),theprobabilityof ndingajobislowerinarigidcountry,whichmakesitmorecostlyfortheworkertoturndownano¤er;thispushesthereservationwagedown.SomesimulationsreportedinSaint-Paul(2000)suggestthatthetwoe¤ectscouldeaslycanceleachother,deliveringasimilarwagediscountinaexibleandarigidcountry.12Notethatthehump-shapedpatternofwagelossesintheUSmimicksthatofwages.Thisisconsistentwithourinterpretationprovidedthat(i)asubstantialfractionoftheevolutionofwagesoverthelifecycleisaccountedforbyspeci chumancapital,and(ii)asubstantialfractionofthatspeci chumancapitalislostwhentheworkerlosesisjoband/orduringthesubsequentunemploymentspell.21 leadtolockingolderworkersintheirjobsforalongtime,whichisprobablyudesirablefromtheviewpointofallocativee¢ciency.Third,nomatterhowhighemploymentprotectionis,therewillalwaysbeamassofunemployedolderworkers,ifanythingbecausetheiremployerwentbankruptorbecausetheyhadtoresignduetosomeadversepersonalshocksuchashavingtomove.Clearly,thesewouldbeparticularlyharmedbydi¤erential ringcostsforolderworkers.ThisisindeedtherationaleforthewaivingoftheDela-landecontributionforworkershiredafter50discussedabove,butitbegsthequestionofhowtohandletheendgamee¤ect.Toresolvethistension,itisusefultostartrecognizingthattheendgamee¤ectisgenuineinthatitissimplysociallyine¢cienttopayhighhiringcostsforaworkerwhowillretirefouryearsafterbeinghired.Thisshouldruleoutpoliciesthatsubsidizethehiringofolderworkers;ifolderworkershavetochangejobs,itmustbeinactivitieswherehiringcostsarenottoohigh.Togetagraspofhowane¢cientpolicywouldlooklike,wemayspecu-lateonhowthelabormarketforolderworkerswouldoperateinaworldofperfectlycompetitivelabormarkets.Insuchaworld,olderworkerswouldbemorelikelytosticktotheircurrentjobbecauseoftheendgameprob-lem.Thismeansthattheywouldhavetobearwithwagecutsinresponsetoproductivityfallingwithageandinresponsetonegativeshockstothatproductivity.Butsomeofthemwouldneverthelesslosetheirjobsandtheymighteitherworkinsectorswithlowhiringcosts,thatareprobablynotpayingmuch,ordecidetousetheiraccumulatedwealthtoretire.Thuswewillsee(i)decliningwageswithage,(ii)asusbstantialwagelossuponun-employment(reectingthedevaluationofspeci chumancapitalassociatedwithjobloss),(iii)lowerjoblossrates,(iv)useofpersonalsavingstomakeupforwagelosses,and nally(v)retirementcontingentonhavinganadverselabormarketshock.Ironically,wecouldmakethepointthat,totheextentthatthewagesofolderworkersare"toohigh"relativetothatbenchmarkinEurope,institu-tionslikeadditionalemploymentprotectionandpre-retirementareexactlywhatisneededtoreplicatethecompetitiveoutcomeintermsofemployment22 patternsdespitethatwagesgivethewrongsignals.However,thisadmin-istrativesolutiondoesnotguaranteethattherightpeopleareallocatedtothesituationsofcontinuingintheirjobandretiring,and,moreworryingly,thethirdtierofolderworkersthosewhomovetolow-productivityjobsiseliminatedaltogether,beingpushedintopre-retirementschemesorotherwelfareprograms.Onewaytoimproveonthatwouldbetomovetoasystemofmoreexiblewagesandlessgenerousunemploymentbene ts(andinparticularreconsidertheirbackwardindexationonwages),whileintroducingacompen-sationsystemthatallowsolderworkerstosupplementtheirlaborincome.Anattractivesolutionistoletthemfreetochoosetheirretirementageinanactuariallyfairway,whiledisconnectingentirelythepaymentofpensionsfromlabormarketparticipation.Thus,thosewhoseetheirwagefallingatsay58,couldstartdrawingtheirpension,eitherinpartorfully,andcu-mulateitwiththeirlaborincome.Anotherapproachwouldbetoreplacegenerousunemploymentbene tsbya"wageinsurance"schemethatwouldsupplementtheincomeofolderworkersiftheyhavelosttheirjobandendupinalow-payingjob.Thiswouldnotbeofgreathelptotheunemployed,butevidencefromanglo-saxoncountriessuggeststhatatlowbene tlev-els,unemploymentdurationislow,sothatsubstantialinsuranceisderivedfromhighjob ndingrate.Furthermore,asystemofunemploymentsup-portaccountcouldbeintroduced,whichwouldamounttoeasingthecreditconstraintontheunemployed.Onecommonworryisthatincreasingtheretirementagedoesnotwork,becauseolderworkershaveaverylowjob ndingrateandarethus"unem-ployable".Totheextentthatthisisduetotheendgamee¤ect,ouranalysissuggeststhatthemerefactofraisingtheretirementagewillincreasethejob ndingrateforworkersinagivenagecategory.Ontheotherhand,theendgamee¤ectcannotbeeliminatedanditwillnowapplytotheolderworkerswhoareatthesamedistancefromretirement;sincethoseareevenlessproductive,thee¤ectwillbeevenstronger,whichreinforcesmyclaimthatcumulatingpensionswithworkmustbeanimportantfeatureofanyworkablepensionreform.23 Shouldtheemploymentoftheolderworkersbesubsidized,asisthecaseintheNetherlands?Euwalsetal.(2008),inanextensivediscussionoftheDutchsystemanditsreformprospects,arguethatthisisprobablycounter-productive.Inadditiontotheethicalproblemsoftaxtreatmentslinkedtoindividualcharacteristics,inapuremarketoutcometheemploymentrateofolderworkerswouldnaturallybelowerthanforprime-ageones.Theonlydis-tortionthatexacerbatesthatisthebackward-indexedunemploymentbene t,butthenitmakessensetoreconsiderthedesignofunemploymentbene ts.Andthereissomecontradictionbetweentheattempttoimprovethe nanc-ingofpublicpensionsbyworkinglonger,andatthesametimeintroducinganotherredistributiveschemeinfavoroftheelderly.Thewageinsurancesystemproposedaboveisnotasperverselyredistrib-utiveastheDutchsubsidy,especiallysinceitmaybene tyoungerwork-ersandolderonesalike(butthelatterwouldbene tmore),andwouldbecompensatedbylowerunemployentbene ts(whichwouldalsobebetterinbudgetaryterms).Andallowingtocumulatepensionsisevensimplerand nanciallyanddistributivelyneutraloverall,thegoalsofthosesystemsistoallowconsumptionsmoothingalongthelifecycleandacrossstatesofnature,inthefaceof nancialmarketimperfections,andthatiswhatunem-ploymentsupportaccountsandactuariallyfairaccesstopensionsprovide.Euwalsetal.agreewiththisbutseemmorecautiousaboutsuchunemploy-mentsupportaccounts,basedontheviewthatcreditconstrainsarenotsoimportantintheNetherlands.Ifthisistrue,thenunemploymentsupportaccountsareoflittlevalue,but,ashownbyHasslerandRodriguez-Mora(1998),soisunemploymentinsurance:giventherelativelylowdurationofunemploymentspellsinaexiblelabormarket,borrowingandlendingallowworkerstoachieveagreatdealofinsuranceontheirownintheabsenceofunemploymentbene ts.Fewcountrieswouldconsiderreducingemploymentprotectionforolderworkers.YetEuwalsetal.(2009)pointtoaDutchstudywhichshowsthatthiscouldreduceunemploymentdurationforolderworkersconsider-ably.Furthermore,theeconomiccaseforincreasingemploymentprotectionwithage/tenureisnotstrong.Currentsystemsofadditionalemployment24 protectionforolderworkersareaveryimperfectattemptatprovidingcon-sumptionsmoothinginlightoflowerre-employmentprobabilitiesforolderworkers.Butoncemoree¢cientinstrumentssuchasthosesuggestedaboveareputinplace,themeritofemploymentprotectionrestsatbestonwhetherthereisadiscrepancybetweentheprivatecostoflaborasperceivedbythe rmandthetruesocialopportunitycostoflabor.Thegreatertheformerrelativetothelatter,thegreatertheincentivestoine¢cientlydismisstheworker.Employmentprotectionisoftenunderstoodasataxtocorrectsuchadiscrepancy(SeeforexampleBlanchardandTirole(2006)).Abovewehavediscussedthreesourcesofhighwagesforolderworkers.Oneisspeci chumancapitalaccumulation;itincreaseswagesbecauseitsreturnaretyp-icallysharedbetweentheworkerandthe rm,notthebecausethe rmsperceivedcostofemployingtheworkerhasincreased.Anotherisacollec-tivebargainingstructurewhichismorebindingforolderworkers.Itmayindeedincreasethewedgebetweentheprivateandsocialopportunitycostoflabor,butthenonemayequallyconsiderachangeincollectivebargaining.Finally,thereisthee¤ectofdeferredpayments.Butsuchpaymentsarenotallocative:InLazears(1990)model,forexample,theirtimingisdeterminedbyincentiveconsiderations,whiletheseparationdecisionisdeterminedbyacomparisonoftheworkersproductivityandtheprivateopportunitycostofwork.Therefore,inthatsetting,greaterwagesfortheelderlypersedonotgenerateexcesssiveincentivesfordismissalsrelativetootherworkers.Hencethereisnocompellingreasontoargueforadditionalemploymentprotectionforolderworkers.REFERENCESAvolio,B.J.andD.A.Waldman(1994).VariationsinCognitive,Per-ceptual,andPsychomotorAbilitiesAcrosstheWorkingLifeSpan:Exam-iningtheE¤ectsofRace,Sex,Experience,Education,andOccupationalType.,PsychologyandAging,Vol.9,No.3,pp.430-442.Behaghel,Luc,BrunoCréponandBéatriceSédillot(2008),"Theperversee¤ectsofpartialemploymentprotectionreform:ThecaseofFrencholderworkers",JournalofPublicEconomics92,696721Blanchard,OlivierJeanandJeanTirole,"TheOptimalDesignofUnem-25 ploymentInsuranceandEmploymentProtection:AFirstPass",JournaloftheEuropeanEconomicAssociation,vol.6,n1,2007,p.45-77.Cohen,Daniel,ArnaudLefrancandGillesSaint-Paul(1997),"FrenchUnemployment:Atransatlanticperspective",EconomicPolicyEuwalsRob,RuuddeMooij,andDanielvanVuuren(2009),"RethinkingRetirement:Fromparticipationtowardsallocation",CPBworkingpapern80,DenHaag.Hairault,Jean-Olivier,FrançoisLangotandThepthidaSopraseuth,(2008)"DistancetoRetirementandOlderWorkersEmployment:TheCaseForDe-layingtheRetirementAge",mimeo,UniversityofParisIJacobson,Louis,LaLonde,Robert,andSullivan,DanielG,(1993)."Earn-ingsLossesofDisplacedWorkers,"AmericanEconomicReview,83(4),685-709.Kotliko¤,LawrenceandJ.Gokhale(1992),"EstimatingTheAge-ProductivityPro leUsingthePresentValueofLifetimeEarnings",QuarterlyJournalofEconomicsLazear,Edward,JobSecurityProvisionsandEmployment,QuarterlyJournalofEconomics105:3(August1990):699-726.Ljunqvist,LarsandThomasSargent(1998),"TheEuropeanUnemploy-mentDilemma",JournalofPoliticalEconomy,106,3,514-550Rosolia,AlfonsoandSaint-Paul,Gilles(1998),"Thee¤ectofunemploy-mentspellsonsubsequentwagesinSpain",workingpaper,UniversitatPom-peuFabraandCEPR.Ruhm,Christopher(1991),AreWorkersPermanentlyScarredbyJobDisplacementsAmericanEconomicReview,Vol.81,No.1,319-24.Saint-Paul,Gilles(1996),DualLaborMarkets.CambridgeMA:MITPress-(2000),"LanatomieduchômageenEspagne;unecom-paraisonaveclaFranceetlesEtatsUnis",EconomieetStatistique,332/333,159-167Shapiro,Carl,andStiglitz,JosephE.(1984)."Equilibriumunemploy-mentasworkerdisciplinedevice".AmericanEconomicReview74:433-44.26 Skirbekk,Vegard(2003),"Ageandindividualproductivity:aliteraturesurvey",MaxPlanckInstituteworkingpaper.27 APPENDIXInthisAppendix,wesolveforthevalueofthe rmunderaconstantp:Wealreadyknowthatinthelowproductivitystate,thevalueofthejobtothe rmisequaltoJL(s)=yLw r(1er(Ts))fors~s;andtoJL(s)=F(s)fors~s;withF(s)=F0ifssandF(s)=F1ifs-381;s:Letusfocusontheregimewhere~s-381;s:WethenhavethatF1=wyL r(1er(T~s)):(3)ConsidernowJH(s);thevalueofemployingaworkerofagesinthehighproductivitystate.TheBellmanequationfors~sis(r+p)JH(s)=yHw+pyLw r(1er(Ts))+dJH(s) ds:(4)Thesolutionofthisdi¤erentialequation,whoseterminalconditionisJH(T)=0;isJH(s)=yHyL r+p1e(r+p)(Ts)+yLw r(1er(Ts)):(5)ThelasttermissimplyJL(s);the rsttermistheexpectedpresentdis-countedvalueoftheadditionalpro tsmaderelativetothelowproductivitystateaslongastheworkerremainsinahighproductivityone.Notealsothatthisequationalongwith(3)impliesthatJH(~s)=yHyL r+p1e(r+p)(T~s)F1:(6)Forss~s;theBellmanequationis(r+p)JH(s)=yHwpF1+dJH(s) ds:Theterminalsolutionofthisdi¤erentialequationisgivenby(6).Conse-quently,thesolutionisJH(s)=yHyL r+p1e(r+p)(Ts)+yLw r+p(1er(Ts)ep(~ss))pF1 r+p:(7)28 Considerthedecisionto reahighlyproductiveworkerjustbeforeages:ThiswillbeoptimalprovidedJH(s)F0:Thisismorelikelytohold,thelowerF0andthegreaterF1:ThesmallestpossiblevalueofF0iszero,andthelargestonecompatiblewiththisregimeistheonesuchthat~s=s;i.e.F1=wyL r(1er(Ts)):Substitutinginto(7),weseethatinsuchacasetheinequalityJH(s)F0=0holdsprovidedyHyL r+p1e(r+p)(Ts)wyL r(1er(Ts)):Themaximumadmissiblevalueofwisw=yH;anditcanbecheckedthatatthispointthisinequalityholds,sinceforanya0theexpression1eaz zfallswithz:Finally,forsstheBellmanequationis(r+p)JH(s)=yHwpF0+dJH(s) ds;(8)andtheterminalconditionisavalue-matchingconditionwith(7)ats=s:Wegetthefollowingsolution:JH(s)=yHyL r+p1e(r+p)(Ts)+yLw r+p(1er(Ts)ep(~ss))pF0 r+pp(F1F0) r+pe(r+p)(ss)(9)Equations(5),(7)and(9)de netheage-pro leofthe rmsnetexpectedpro tsfromemployingaworkerinthehighproductivitystate.IfJH(s)H;thenanyworkerofageswouldbehired.Ifthatinequalityfails,thentheworkerwouldnotbehired.Theseformulasarevalidintheparameterzonewhere~ss;orequivalentlyF1wyL r(1er(Ts)):Di¤erentiatingtherelevantformulaforJH(:)inthethreezones,wegetthefollowing:1.Ifs~s;wehavethatJ0H(s)_(wyL)(yHyL)ep(Ts):2.Ifss~s;thenJ0H(s)_(wyL)(yHyL)ep(T~s):3.Finally,ifss;thenJ0H(s)_(wyL)(yHyL)ep(T~s)er(Ts)ep(~ss)p(F1F0)29 Consequently,if(wyL)(yHyL)ep(T~s);orequivalentlywyL(yHyL)1rF1 wyLp=rthenJ0H0throughout.Ifthisinequalitydoesnothold,thentherearetwopossibilities:-If(wyL)(yHyL)ep(T~s)]TJ/;༕ ;.9;U T; 12;.425; 0 T; [00;er(Ts)ep(~ss)p(F1F0);thenH0]TJ/;༕ ;.9;U T; 12;.425; 0 T; [00;0fors^s=T+1 plnwyL yHyLandH00fors^s:-If(wyL)(yHyL)ep(T~s)er(Ts)ep(~ss)p(F1F0);thenH00forss;H0-277;0forss^s;andH0]TJ/;༕ ;.9;U T; 34;.166; 0 T; [00;0fors]TJ/;༕ ;.9;U T; 34;.166; 0 T; [00;^s:Forthesakeofcompleteness,letusalsodescribethesolutionifF1]TJ/;༕ ;.9;U T; 34;.166; 0 T; [00;wyL r(1er(Ts)):Inthiscase,workerssuchthatssarenot redwhenfallingintothelowproductivitystate,andwehavetwofurtherpossibilities:A.IfF0wyL r(1er(Ts));thenallworkerssuchthatsslosetheirjobwhentheirproductivityfalls.ThenJHisgivenby(5)foralls-277;s:Forss;itfollowstheBellmanequation(8)withtheterminalconditionJH(s)=yHyL r+p1e(r+p)(Ts)+yLw r(1er(Ts)):ThesolutionisJH(s)=yHyL r+p1e(r+p)(Ts)pF0 r+p1e(r+p)(ss)+yLw r(r r+p+p r+pe(r+p)(ss)er(Ts)ep(ss)):B.IfF0wyL r(1er(Ts));thenworkerswhofallintothelowstateareretainedi¤s~s;with~snowsolutiontoF0=wyL r(1er(T~s));(10)and~ss:Foralls-308;~s;JH(s)satis estheBellmanequation(4)withtheterminalconditionJH(T)=0:Henceitisagaingivenby(5).Foralls~s;itsatis es(8)withterminalconditionJH(~s)=yHyL r+p1e(r+p)(T~s)F0:30 ThesolutionisJH(s)=yHyL r+p1e(r+p)(Ts)F0 r+pp+re(r+p)(~ss)+yLw r+p1e(r+p)(~ss):31 life cyle; source: Avolioand Waldman (1994) ss countries. Source: Hairaultet al. (2008) syLFigure 3: the firing frontier s* syLFigure 4: effect of tighter employment protection s* s~ Figure 9: E-U transitions0,01,02,03,04,05,06,07,08,021 to 2525 to 3031 to 3536 to 4041 to 4546 to 5051 to 5556 to 6061 to 65 France United States Figure 10: E--non E transition0,05,010,015,020,025,030,035,040,021 to 2525 to 3031 to 3536 to 4041 to 4546 to 5051 to 5556 to 6061 to 65 France United States Figure 11: job finding rate0,010,020,030,040,050,060,070,021 to 2525 to 3031 to 3536 to 4041 to 4546 to 5051 to 5556 to 6061 to 65 France United States Figure 12: median wage, France, men0,05,010,015,020,025,030,035,021 to 2525 to 3031 to 3536 to 4041 to 4546 to 5051 to 5556 to 6061 to 65 High School dropouts High school Some college College graduates Figure 13: median wage, USA, men0,05,010,015,020,025,021 to 2525 to 3031 to 3536 to 4041 to 4546 to 5051 to 5556 to 6061 to 65 High School dropouts High school Some college College graduates High school Some college College degree Country France USA France USA France USA France USA 56-60 47.9 (1.5) 61-65 8.6 (1.1) Table 1 Employment rates of older workers by educational levels, men, US and France. High school Some college College degree Country France USA France USA France USA France USA 31-35 11.6 (1.1) 36-40 7.3 (0.8) 41-45 5.9 (0.7) 46-50 6.2 (0.7) 51-55 6.3 (0.7) 56-60 3.4 (0.6) Table 2 Unemployment rates, men, France and USA France USA 16-30 -0.041 (0.015) 31-40 -0.128 (0.021) 41-50 -0.199 (0.026) 51-65 -0.201 (0.039) ers, men and women, France and USA.