Disability and Labour Fource Participaton in Ireland 1995 - 2000

Disabili y and Labou Fo ce Pa icipa ion in I eland 1995-2000 B enda Gannon* Economic and Social Resea ch Ins i u e and Depa men o Economics, Na ional Uni e si y o I eland, Maynoo h Summa y This pape aims o analyse he e ec o disabili y on pa icipa ion in he labou o ce, using he I ish componen o he Eu opean Communi y Household Panel Su ey 1995-2000. A ange o panel models a e conside ed, bu o allow o any unobse ed in luences o s a e dependence in labou o ce pa icipa ion, ou p e e ed model is a dynamic panel model. We show how he es ima es o cu en disabili y a e changed once we con ol o he e ec o pas disabili y and p e ious pa icipa ion. We compa e base es ima es o disabili y wi h hose con olling o unobse ed he e ogenei y and pas pa icipa ion. The esul s sugges ha he base e ec o disabili y is o e es ima ed by be ween 40-60 pe cen o men and by 5-10 pe cen o women. Keywo ds Disabili y; labou o ce pa icipa ion; s a ic and dynamic panel models 6 Oc obe 2004 *Co espondence o: B enda Gannon, Economic and Social Resea ch Ins i u e, 4 Bu ling on Road, Dublin 4, I eland. Email: b enda.gannon@es i.ie WORK IN PROGRESS: This pape is no o publica ion and should no be quo ed wi hou p io pe mission om he au ho . 1 I. In oduc ion People wi h disabili ies ace many ba ie s o ull pa icipa ion in socie y, no leas in he labou ma ke , and he ex en and na u e o pa icipa ion in he labou ma ke has a mul i ude o di ec and indi ec e ec s on hei li ing s anda ds and quali y o li e. In s udying he e ec o disabili y on labou o ce pa icipa ion, we a e aced wi h a a ie y o analy ical challenges, such as he e ec o unobse ed cha ac e is ics o disabled indi iduals and he e ec o hei pas pa icipa ion in he labou ma ke . This pape uses panel da a me hods o con ol o hese ac o s and we es ima e he impac o disabili y on pa icipa ion, con olling o unobse ed he e ogenei y and pas pa icipa ion. In e na ionally, he i s gene a ion o econome ic s udies on he e ec o disabili y on labou o ce pa icipa ion eme ged a ound he la e 1970’s. Ba el and Taubmann [1] es ima e an OLS model o weekly hou s wo ked o analyse he e ec o heal h on ea nings and labou supply, whe eas Chi okos and Nes el [2], es ima e a Tobi model ela ing annual hou s wo ked o heal h his o y by looking a he deg ee o poo , good, imp o ed o de e io a ing heal h o e he p e ious en yea s. Mo e ecen esea ch emphasises he impo ance o he way heal h and limi a ions a e cap u ed, wi h he ype o heal h s a us a iable used leading o di e en pa e ns in e ms o labou o ce pa icipa ion. Wol e and Hill [3], o example, measu e heal h s a us using an index o limi a ion in daily ac i i ies, Madden and Walke [4] measu e heal h in e ms o hose who epo a longs anding illness o disabili y, while Me e and Schul z [5] also measu e heal h s a us using a heal h index. Using Labou Fo ce Su ey da a, Kidd, Sloane and Fe ko [6] analyse he e ec o heal h limi a ions on he kind o paid wo k possible in he UK. They con i m he p esence o subs an ial wage and pa icipa ion a e di e ences be ween disabled and non-disabled indi iduals. The ocus o p e ious policy o disabled people has been on he p o ision o se ices, whe eas mo e ecen ly, he e is a campaign o ci il igh s and he p o ision o legisla ion o equali y and ull pa icipa ion. Employe s and policy make s a e he e o e in e es ed in whe he o no disabili y has an e ec on pa icipa ion. In his pape , we aim o de e mine whe he i is disabili y ha de e mines he pa icipa ion decision, o i he e may be some o he unobse able cha ac e is ics in ol ed ha dis o he disabili y e ec s. 2 P e ious s udies analysing unobse ed indi idual e ec s in his con ex eme ged in he mid-eigh ies. Sickles and Taubman [7] we e one o he i s o use longi udinal da a in es ima ing e i emen decisions, and allow o unobse ed he e ogenei y in he e i emen unc ion. Es ima ing a bina y andom e ec s p obi model, hey allow o unobse ed a ec s by simul aneously es ima ing he heal h and e i emen equa ion, allowing he e o s o be co ela ed. They allow o co ela ion o he unobse ed e ec wi h he disabili y a iable, bu hey do no include he e ec o labou ma ke his o y. Thei indings show ha mo ing om poo heal h o good heal h dec eases he p obabili y o e i emen , bu hey do no show how he heal h e ec changes as a esul o allowing o unobse ed e ec s. Bound [8] looks a a e i emen equa ion in he c oss sec ional con ex , and shows ha i he e o s in he heal h and e i emen equa ion a e co ela ed, hen he e is an upwa d bias in he e ec o heal h. He aims o iden i y he e ec s o inancial incen i es on epo ing beha iou and e i emen decisions, and in es iga es i objec i e measu es may be used as a p oxy o subjec i e measu es o heal h. The au ho concludes ha he sel - epo ed measu e is no eliable in es ima ing he e ec o heal h on e i emen . K eide [9] also analyses wo k pa icipa ion wi h c oss sec ion me hodology and a i es a he same conclusion. He inds ha when he ue measu e o disabili y is used, he e ec on pa icipa ion is lowe , by 17.2% o men and 24.9% o women. Bo h Bound [8] and K eide [9] use c oss sec ion da a o es ima e he e ec o he ue e ec o disabili y on pa icipa ion, bu iden i ica ion o hei models equi es a a iable, ha a ec s heal h bu ha is no co ela ed wi h pa icipa ion. Kidd, Sloane and Fe ko [6] apply a decomposi ion app oach o c oss sec ion da a and ind ha app oxima ely 50 pe cen o he di e ence in pa icipa ion a es is due o unexplained e ec s. Ou da a o e he possibili y o analysing he ela ionship be ween disabili y and labou o ce pa icipa ion o e a signi ican pe iod a he han jus a a poin in ime, and allow us o use panel da a echniques in ou es ima ion. Using panel da a, we cap u e he e ec s o a iables ha a e pa icula o an indi idual and a e cons an o e ime. We can con ol o hese unobse ables by using a panel model and we he e o e do no need o include an iden i ying a iable. Labou o ce pa icipa ion may also be in luenced by pas pa icipa ion, whe e non-pa icipan s in he p e ious 3 yea may be less likely o pa icipa e in he cu en yea . Al hough his may be ue o all indi iduals, i may also be a speci ic cha ac e is ic o disabled people and lead o an inco ec in e p e a ion o he disabili y e ec . I may be ha disabili y educes he p obabili y o p e ious pa icipa ion, and he e o e indi ec ly in luences cu en pa icipa ion. Using panel da a, we can inco po a e his s a e dependence e ec and e-es ima e he e ec o disabili y on pa icipa ion. I may also be ha indi iduals epo a disabili y as an ex-pos jus i ica ion o no being in wo k in he p e ious yea . Again, we would expec he e ec o disabili y o be misin e p e ed, and can use he esul s o he dynamic model o disen angle he unobse ed he e ogenei y and pas pa icipa ion e ec s. Mo e ecen ly, Lindeboom and Ke kho s [10] also include he e ec o pas labou ma ke ou comes on cu en heal h in hei e i emen model. They ind ha o elde ly people, wo king in he p e ious pe iod only sligh ly dec eases he alue o heal h. They es ima e a mul inomial logi model, o acili a e he h ee di e en labou ma ke s a es compa ed o wo king, a ailable o indi iduals nea ing e i emen age in he Ne he lands. Al hough hey only ha e wo wa es o panel da a, by using in o ma ion on p e ious labou ma ke his o y, hey speci y an equa ion o ini ial pa icipa ion and es ima e he p obabili y o wo king ini ially. This is included in o he o e all likelihood unc ion om which unobse ed e ec s a e in eg a ed ou . They ind ha he e ec s o heal h a e exagge a ed o elde ly people in a simple mul inomial model, compa ed o hei p e e ed model. In his pape , we ollow a di e en app oach o Lindeboom and Ke kho s [10] mainly because we use six wa es o panel da a and can he e o e iden i y he e ec o pas pa icipa ion wi hin a less complica ed model. The main ocus in his pape is o model wo labou ma ke ou comes – pa icipa ion and non-pa icipa ion – and hence we concen a e on a bina y esponse a iable. In con as o Lindeboom and Ke kho s [10], we ollow an app oach by Woold idge [11] ha allows us o a oid speci ying a dis ibu ion o he ini ial pa icipa ion. The likelihood unc ion om ou app oach is easie o es ima e and se es he same pu pose in e ms o looking a he e ec o unobse ed he e ogenei y. Ou indings using I ish da a a e simila o hose o Lindeboom and Ke kho s [10] among o he s, in ha hei epo ed disabili y a iable o e es ima es he impac o disabili y on pa icipa ion in he Ne he lands. In addi ion, 4 we show exac ly how much unobse ed he e ogenei y con ibu es o a ia ion in pa icipa ion and how his changes he e ec o disabili y. Finally, we show he e ec o pas disabili y ( ia i ’s e ec on p e ious pa icipa ion), on cu en labou o ce pa icipa ion. II. Theo e ical F amewo k We model pa icipa ion i s ly as a s a ic p ocess wi h cu en pa icipa ion, and secondly as a dynamic p ocess. In ou dynamic model, we accoun o he ac ha he choice be ween consump ion and leisu e is conside ed as a li e ime decision, so we assume ha indi iduals maximise hei expec ed u ili y o e hei li e ime. Following Bound e al [12], in gene al, he pa icipa ion equa ion is based on he assump ion ha indi iduals maximise a u ili y unc ion gi en by: ),,(max jjj j T j ZLCUE     [1] whe e Cj and Lj a e consump ion and leisu e in pe iod j espec i ely. Zj is a ec o o as e shi e s and includes disabili y. The u ili y unc ion is maximised subjec o an in e empo al budge cons ain : jjjjjj A CHWA )1()( 11     [2] whe e Wj is he wage, Hj is hou s o wo k, Aj ep esen s asse s, and j is he a e o in e es . In his pape , ou empi ical model shows how indi iduals compa e he u ili y be ween wo s a es – pa icipa ion and non-pa icipa ion. Sol ing his model p o ides an exp ession o op imal leisu e as a unc ion o W, H, Aj and Zj. Much o he li e a u e on he e ec o heal h on labou o ce beha iou has ea ed heal h as an exogenous as e shi e . We ake his app oach, and hence do no speci y a heal h p oduc ion unc ion. In his con ex , we ob ain es ima es om a educed o m modela and concen a e on how he disabili y e ec changes once we allow o unobse ed indi idual e ec s and s a e dependence in labou o ce pa icipa ion. 5 III. Da a The da a on disabili y and labou o ce pa icipa ion in I eland a e om he Li ing in I eland Su ey 1995-2000.b The Li ing in I eland Su ey is he I ish componen o he Eu opean Communi y Household Panel, conduc ed by he ESRI o Eu os a . We wish o ocus on indi iduals o wo king age, hence we exclude hose aged 65 and o e . In he Li ing in I eland Su ey, de ailed in o ma ion on cu en labou o ce s a us was ob ained. Fo cu en pu poses his allows us o dis inguish be ween hose who we e a wo k, o unemployed bu seeking wo k – who we will coun as ac i e in he labou o ce – and all o he s, whom we will coun as inac i e. The pe cen age o hose unemployed bu seeking wo k is qui e low anging om 7.5% in 1995 o 2.8% in 2000, gi ing a panel a e age o 5.1%. Fo his eason, we do no include hem as a sepa a e ca ego y in ou dependen a iable. A measu e o disabili y can also be cons uc ed om he Li ing in I eland su ey on he basis o indi idual esponses o he ollowing ques ion: “Do you ha e any ch onic, physical o men al heal h p oblem, illness o disabili y?” I may well be, ha no only he p esence o such an illness o disabili y bu also he ex en o which i limi s o es ic s a pe son, is impo an . To cap u e his, we use esponses o a ollow-up ques ion conce ning he impac o he disabili y o dis inguish a) hose epo ing a ch onic illness o disabili y and saying ha i limi s hem se e ely in hei daily ac i i ies b) hose who epo a ch onic illness o disabili y and saying i limi s hem o some ex en , and c) hose who epo such a condi ion bu say i does no limi hem a all in hei daily ac i i ies. The ex en o which esponden s say hey a e limi ed ela es o hei daily ac i i ies a he han wo k, bu simila measu es ha e been shown o ha e signi ican disc imina o y powe in e ms o labou o ce pa icipa ion in esea ch elsewhe e (e.g. Malo [13]). Fu he mo e, as Table 1 shows he e a e di e en a es o pa icipa ion o 6 each sub-g oup, so i is impo an ha we dis inguish be ween he di e en le els o disabili y, in ou analysis o labou o ce pa icipa ion. The e ec s o disabili y on labou o ce pa icipa ion may di e among indi iduals, depending on o he cha ac e is ics, o example age o educa ion. Since disabili y may be co ela ed wi h o he a iables, we include measu es o age, educa ion, egion, unea ned income, age o younges child and ma i al s a us. These a iables a e de ined in de ail in Table 2 and summa y s a is ics a e p o ided in Table 3. The younges indi iduals in his sample a e aged 16 and he numbe o obse a ions o males and emales a e 7,188 and 7,670 espec i ely. IV. Panel models and Resul s S a ic Pooled P obi Model: Using he Li ing in I eland Su ey 1995-2000, we es ima e a ange o panel models o cap u e he e ec o disabili y on pa icipa ion. We exclude 1994 because he ques ions ega ding heal h p oblems and limi a ions di e ed om 1995 and subsequen yea s. Fi s ly, we es ima e a s a ic pooled model, assuming ha he e o s a e independen o e ime and unco ela ed wi h he explana o y a iables. This model p o ides us wi h base es ima es, wi h which we can compa e esul s om models ha inco po a e unobse ed he e ogenei y and s a e dependence. The log likelihood unc ion o he pooled panel da a is simila o ha o he c oss sec ional p obi : ))(1log()1()(log)(log ' 111 ' 1  i T i N i T i i N ixFyxFyL    [3] and maximising his ac oss all i wi h espec o , we ob ain he pooled p obi es ima o . The s anda d e o s a e adjus ed o accoun o se ial co ela ion in he e o s a he indi idual le el. The main a iables o in e es a e, disabili y and he associa ed limi a ions in daily ac i i ies, bu we also con ol o o he ac o s ha may 7 be co ela ed wi h disabili y, as men ioned ea lie . In addi ion, i is likely ha pas disabili y has a di ec e ec on cu en pa icipa ion, so we include lagged a iables o he h ee ypes o disabili y. Pooling all a ailable da a o he yea s 1995 o 2000, and es ima ing a s anda d p obi model, we ob ain es ima es om he pooled balanced sample.c We p esen esul s om his pooled s a ic model in Table 4, Columns 1 and 4, o men and women espec i ely. These esul s a e p esen ed as pa ame e co- e icien s, bu we will la e discuss some o he main esul s in e ms o pe cen age e ec s. The e ec s o cu en disabili y a e qui e high o bo h men and women, educing he p obabili y o cu en labou o ce pa icipa ion signi ican ly. A a i s glance, disabili y has a g ea e nega i e e ec on he labou o ce pa icipa ion p obabili y o men, compa ed o women. Al hough he e ec o a se e ely limi ing disabili y is less o women han men, i is s ill subs an ial. In he case o men, e en hose wi h no limi a ions ha e a sligh educ ion in he p obabili y o pa icipa ion. Fo women, we see ha he p obabili y o pa icipa ion o hose wi h no limi a ions, is no signi ican ly di e en om women wi h no disabili y. The gap be ween he e ec s o se e e and some limi a ions is qui e la ge o men and e en mo e p onounced o women, sugges ing ha se e e disabili y has a mo e nega i e e ec on women’s pa icipa ion. Pas disabili y, in he p e ious yea , also has a subs an ial e ec on cu en pa icipa ion, and is no much lowe han he e ec o cu en disabili y. This applies in he case o se e e and some limi a ions, o bo h men and women. Simila o cu en disabili y and se e e limi a ions, we see ha indi iduals who p e iously had a se e ely limi ing disabili y ha e a much lowe p obabili y o cu en pa icipa ion, compa ed o hose wi h no p e ious disabili y. In e ms o he o he explana o y a iables (see Table A1), we see ha labou o ce pa icipa ion inc eases wi h age up o 34 (compa ed o hose aged 55-64), bu he e ec alls sligh ly a e he age o 44. Those wi h seconda y o hi d le el educa ion ha e a g ea e p obabili y o pa icipa ing in he labou ma ke . As expec ed, we see ha women wi h child en a e less likely o pa icipa e, and his e ec ge s smalle as he younges child is olde . The opposi e e ec is ound o men, whe e child en inc ease he p obabili y o pa icipa ion, in pa icula when he younges child is ei he aged less han 4, o in he olde age g oup o 12-18. 8 The esul s om he s a ic pooled model aise wo impo an ques ions. The i s in e es ing ques ion is whe he o no pas disabili y a ec s cu en pa icipa ion di ec ly, o does i wo k h ough a sepa a e channel by nega i ely a ec ing pas pa icipa ion? I so, we would expec o see ha pas pa icipa ion in luences cu en pa icipa ion, and he e ec o pas disabili y should disappea . This would sugges ha pas disabili y s ill does ha e an e ec on cu en pa icipa ion, bu does so by a) di ec ly in luencing pas pa icipa ion and he e o e, b) indi ec ly a ec ing cu en pa icipa ion. The second ques ion a ising om hese esul s is whe he o no he con ol a iables app op ia ely accoun o any unobse ed cha ac e is ics o disabled people ha also in luence hei labou o ce pa icipa ion decision? Again, i his we e no ue, we would expec ha he ac ual e ec o cu en disabili y should be lowe . We now explo e a dynamic model o pa icipa ion ha inco po a es bo h pas pa icipa ion and unobse ed e ec s. S a e Dependence and Unobse ed He e ogenei y: In o de o dis inguish be ween he wo e ec s – unobse ed indi idual e ec s and pas pa icipa ion - we now include a lagged dependen a iable in o he model.d In gene al e ms he ollowing likelihood is de i ed and maximised; iiiiiiii i T i iiiiiTiiTiiTiiTi dx xy xyy dx xxyy xxyy   )|(),,|()],,,|([ )|(),,,...|,...,(),,...,|,...,( 001, 1 1000           [4] We mus speci y ),|( 0iii xy  - known as he ini ial condi ions p oblem. Heckman [14] sugges s app oxima ing ),|( 0iii xy  and hen speci ying )|( ii x  . Then )|,...,( 0iiTi xyy is ob ained by in eg a ing ou he unobse ed e ec . The main di icul y in his app oach is in speci ying he dis ibu ion o ini ial pa icipa ion. We he e o e ollow an al e na i e app oach sugges ed by Woold idge [11] whe e we conside : iiiiiii T iiiiiTi dxy xyyy xyyy  ),|(),,|,...,(),|,...,( 0010 1   [5] 15 disabili y and pa icipa ion, and people wi h hese disabili ies may e-join he labou o ce. The incen i e e ec s o disabili y bene i s may also play a ole he e and hese ac o s will be in es iga ed in u u e esea ch. V. Conclusions People wi h disabili ies ace many ba ie s o ull pa icipa ion in he labou ma ke , wi h se ious implica ions o li ing s anda ds and quali y o li e. This pape has analysed he ac o s associa ed wi h pa icipa ion o non-pa icipa ion in he labou ma ke , using da a on people epo ing ch onic illness o disabili y in a la ge-scale I ish ep esen a i e su ey. The esul s o he panel analysis p esen ed in his pape , b ing ou he scale o he impac on labou o ce pa icipa ion, o ha ing an illness o disabili y ha limi s he indi idual se e ely in hei daily li e. We con olled o s a e dependence and unobse ed he e ogenei y by es ima ing a dynamic model wi h co ela ed andom e ec s. The esul s show ha unobse ed he e ogenei y con ibu es subs an ially o he base e ec o disabili y o men, and o some ex en o women. In ou p e e ed model, (pooled dynamic) disabled men wi h a cu en se e e limi a ion a e now only 9 pe cen age poin s less likely o pa icipa e compa ed o non-disabled men. Howe e , he e ec o pas pa icipa ion is qui e high, a 40 pe cen age poin s. Fo women, ou p e e ed model is he dynamic model wi h co ela ed andom e ec s. Those wi h a se e ely limi ing disabili y ha e a lowe p obabili y o pa icipa ion by 26 pe cen age poin s, compa ed o women wi h no disabili y. The e ec s o some and no limi a ions a e less subs an ial. The e ec o pas pa icipa ion is lowe in he model o women, educing cu en pa icipa ion by 13 pe cen age poin s. The in e ac ion o disabili y, educa ion and pa icipa ion o women, should be explo ed u he . In his pape , we aimed o p o ide mo e accu a e es ima es o he e ec o disabili y on pa icipa ion. Howe e , we acknowledge some limi a ions. In pa icula , i he epo ing o disabili y in he su ey is p one o measu emen e o , we canno es ima e he ue e ec o disabili y on pa icipa ion. This may help o explain he subs an ial con ibu ion o unobse ed indi idual e ec s, bu wi hou ex ending he model o allow o measu emen e o in epo ing beha iou , ou esul s on he e ec o disabili y on pa icipa ion a e no conclusi e. Again, his will o m pa o u u e 16 esea ch whe e we will model labou o ce pa icipa ion and disabili y, while con olling o epo ing beha iou . 17 Foo no es: a. Ou speci ica ion includes a measu e o unea ned income bu does no include a con ol o wages. Co ec ly accoun ing o he ela ionship be ween disabili y and wages is a opic o u u e esea ch. b. Ano he da a sou ce is a special module on disabili y included wi h he Qua e ly Na ional Household Su ey (QNHS) in he second qua e o 2002, which ocused on he ex en and na u e o es ic ion o ac i i ies o people wi h disabili ies and hei labou o ce s a us. Simila analyses o disabili y labou o ce pa icipa ion in a c oss sec ional con ex , we e ca ied ou using QNHS da a and we a i e a simila conclusions ob ained om he Li ing in I eland 2000 da a. c. In his pape , we a e assuming ha al hough he e is a i ion in he sample be ween 1995 and 2000, i does no bias he esul s o he e ec o disabili y on pa icipa ion. This is especially e iden in he pooled model, whe e we ollow Woold idge [11] and es o he e ec o a i ion using in e se p obabili y weigh s on he pooled model o he unbalanced sample. The p obabili y o being in each wa e is no in luenced by disabili y, o bo h men and women. In he pa icipa ion model o men he e was no change in he o e all co-e icien s. Fo women, we ind ha he e is a sligh o e es ima ion o he e ec o se e e disabili y, changing he co-e icien in he unbalanced sample om –0.3678 in he o iginal pooled model o –0.4203 in he weigh ed pooled model. Howe e , in his pape we assume o e all ha a i ion is no a p oblem in biasing es ima es o disabili y, and ocus on he balanced sample h oughou . d. We could in oduce a lag o wo yea s o pa icipa ion, and hen include ini ial pa icipa ion and p e ious pa icipa ion as he wo ini ial alues. Howe e , his inc eases da a equi emen s and wi hou a la ge T, we canno a o d o be so lexible in he dynamics o he model. Fu he mo e, he ansi ion ma ix p obabili ies o pa icipa ion in each yea show ha he a e o change om pa icipa ion o non-pa icipa ion o ice e sa is he same o each pai o yea s. The co ela ion be ween pa icipa ion and p e ious 18 pa icipa ion is 0.79, likewise he co ela ion be ween p e ious pa icipa ion and lagged p e ious pa icipa ion is 0.79. 19 Re e ences 1. Ba el, A. and Taubman, P. Heal h and Labou Ma ke Success: The Role o Va ious Diseases. The Re iew o Economics and S a is ics 1979; LXI (1): 1-8. 2. Chi okos, T. and Nes el, G. Fu he E idence on he Economic E ec s o Poo Heal h. The Re iew o Economics and S a is ics, 1985; LXVII(1): 61-69. 3. Wol e, B. and Hill, S. The E ec o Heal h on he Wo k E o o Single Mo he s. The Jou nal o Human Resou ces 1995; 30[1]: 42-62. 4. Madden, D. and Walke , I. Labou Supply, Heal h and Ca ing: E idence om he UK. UCD Wo king Pape 1999; 99/28. 5. Me e, C. and Schul z, T. Heal h and Labou Fo ce Pa icipa ion o he Elde ly in Taiwan. Economic G ow h Cen e Yale Uni e si y Discussion Pape 2002: 846. 6. Kidd, M, Sloane,P. and I. Fe ko. Disabili y and he labou ma ke : an analysis o B i ish males. Jou nal o Heal h Economics 2000, 19:961-981. 7. Sickles, R. and P. Taubman. An Analysis o he Heal h and Re i emen S a us o he Elde ly. Econome ica 1986; 54[6]: 1339-1356. 8. Bound, J. Sel -Repo ed e sus Objec i e Measu es o Heal h in Re i emen Models. The Jou nal o Human Resou ces 1990; XXVI[1]:106-138. 9. K eide , B. La en Wo k Disabili y and Repo ing Bias. The Jou nal o Human Resou ces 1999; XXXIV[4]:734-769. 10. Lindeboom, M. and M. Ke kho s. Subjec i e Heal h Measu es, Repo ing E o s and he Endogenous Rela ionship be ween Heal h and Wo k. Wo king Pape 2002. 20 11. Woold idge, J. (2002), Econome ic Analysis o C oss Sec ion and Panel Da a. 12. Bound, J, Schoenbaum, M., S ineb ickne , T.R. and T. Waidmann. The Dynamic E ec s o Heal h on he Labo Fo ce T ansi ions o Olde Wo ke s. Labou Economics, 1999, 6(2): 179-202. 13. Malo, M. and Ga cia-Se ano, C. An Analysis o he Employmen S a us o he Disabled Pe sons Using he ECHP Da a, Second D a ; 2003. 14. Heckman, J. J. The Inciden al Pa ame e s P oblem and he P oblem o Ini ial Condi ions in Es ima ing a Disc e e Time-Disc e e Da a S ochas ic P ocess, in S uc u al Analysis o Disc e e Da a wi h Econome ics Applica ions, Manski, C.F. and D. McFadden. Camb idge, MA:MIT P ess, 179-195. Fo ma ed 21 Table 1 Labou Fo ce S a us by le el o es ic ion o hose wi h Ch onic Illness o Disabili y, age 15-64, Li ing in I eland Su ey 1995-2000 Se e e limi a ion Some limi a ion No limi a ion No ch onic illness o disabili y Men Pa icipa ion 34.92 58.02 81.45 91.59 Non-pa icipa ion 65.08 41.98 18.55 8.41 N 189 655 318 6026 Women Pa icipa ion 13.82 31.82 44.65 55.15 Non-pa icipa ion 86.18 68.18 55.35 44.85 N 123 707 318 6522 22 Table 2 Va iable de ini ions o Dependen and Independen Va iables Va iable De ini ion LFP =1 i pa icipa ing in he labou ma ke , =0 o he wise Disabled wi h se e e limi a ion =1 i disabled and se e ely limi ed in daily ac i i ies, =0 o he wise Disabled wi h some limi a ion =1 i disabled and limi ed o some ex en in daily ac i i ies, =0 o he wise Disabled wi h no limi a ion =1 i disabled and no limi ed in daily ac i i ies, =0 o he wise (Base ca ego y=No disabili y) Age 15-24 =1 i aged 15-24 yea s, =0 o he wise Age 25-34 =1 i aged 25-34 yea s, =0 o he wise Age 35-44 =1 i aged 35-44 yea s, =0 o he wise Age 45-54 =1 i aged 45-54 yea s, =0 o he wise (Base ca ego y=aged 55-64 yea s) BMW =1 i li ing in Bo de , Midlands, Wes egion, =0 o he wise (Base ca ego y=Res o Coun y) Seconda y Educa ion =1 i highes le el o educa ion comple ed is seconda y, =0 o he wise Thi d Le el Educa ion =1 i highes le el o educa ion comple ed is hi d le el, =0 o he wise (Base ca ego y=No quali ica ions o highes le el o educa ion comple ed is p ima y) Ma ied =1 i ma ied o li ing wi h a pa ne , =0 o he wise Age Younges Child<4 =1 i age o younges child is less han 4, =0 o he wise Age Younges Child>=4 and <12 =1 i age o younges child is g ea e han o equal o 4 and less han 12, =0 o he wise Age Younges Child>=12and <18 =1 i age o younges child is g ea e han o equal o 12 and less han 18, =0 o he wise (Base ca ego y=No child en) Unea ned Income =Ne Household Income – Ne Indi idual Disposable Income (Ne Indi idual Disposable Income includes ne incomes om wo k, social wel a e paymen s and child bene i . Ne Household Income agg ega es indi idual da a o household le el) No e: The egional classi ica ions a e based on he NUTS (Nomencla u e o Te i o ial Uni s) classi ica ion used by Eu os a . 23 Table 3 Summa y S a is ics o all Va iables Va iable Pe cen age o Sample in each Ca ego y Men Women LFP 86.6 51.9 Disabled wi h se e e limi a ion 2.6 1.6 Disabled wi h some limi a ion 9.1 9.2 Disabled wi h no limi a ion 4.4 4.1 No Disabili y 83.8 85.0 Age 15-24 12.3 10.1 Age 25-34 16.4 17.2 Age 35-44 26.2 27.1 Age 45-54 24.4 25.6 Age 55-64 20.7 20.0 BMW 24.7 21.9 Seconda y Educa ion 51.8 59.0 Thi d Le el Educa ion 16.7 13.3 No educa ion o p ima y only 31.4 27.6 Ma ied 68.7 73.3 Age Younges Child<4 12.5 13.3 Age Younges Child>=4 and <12 21.3 24.5 Age Younges Child>=12and <18 15.2 17.7 Unea ned Income 228.64 (240.13) 389.5 (307.7) N 7188 7670 No e: Fo unea ned income we p esen he mean and s anda d de ia ion (in pa en heses) 24 Table 4 Panel Model Resul s Men (co-e icien s) Women (co-e icien s) Pooled S a ic Random e ec s dynamic ( e-scaled) Pooled Dynamic Pooled Random e ec s dynamic ( e-scaled) Pooled Dynamic Lag LFP 0.7511** (0.1194) 1.687** (0.0918) 0.7494** (0.0835) 1.7974** (0.0623) Disabled wi h se e e limi a ion -1.2368** (0.1314) -0.6639** (0.2653) -0.5653** (0.2218) -0.9173** (0.1736) -0.8256** (0.2827) -1.1359** (0.2393) Disabled wi h some limi a ion -0.7886** (0.0814) -0.5159** (0.1594) -0.4757** (0.1285) -0.3296** (0.0755) -0.3137** (0.1283) -0.4210** (0.1106) Disabled wi h no limi a ion -0.2066** (0.1042) -0.3464** (0.2161) -0.3397** (0.1380) -0.0175 (0.0928) -0.1811** (0.1497) -0.2732** (0.1326) Lagged Disabili y Disabled wi h se e e limi a ion -1.0555** (0.1275) -0.2534 (0.2593) -0.0765 (0.2465) -0.6203** (0.1626) -0.1470 (0.2863) 0.0102 (0.2643) Disabled wi h some limi a ion -0.5802** (0.0783) 0.0259 (0.1592) 0.1796 (0.1302) -0.2742** (0.0714) -0.0056 (0.1303) 0.0514 (0.1177) Disabled wi h no limi a ion -0.0925 (0.1175) 0.0887 (0.2254) 0.1298 (0.1461) -0.0290 (0.0962) -0.0495 (0.1566) -0.0464 (0.1363) Ini ial condi ion LFP in 1995 1.2059** (0.2096) 0.6399** (0.0944) 0.8984** (0.1353) 0.6315** (0.0626) Random e ec ( ime a e ages) Disabled wi h se e e limi a ion -0.8815** (0.5948) -0.9013** (0.4588) -0.3077 (0.7211) -0.2653 (0.5607) Disabled wi h some limi a ion -0.7265** (0.3237) -0.7146** (0.2371) -0.1387 (0.2744) -0.1209 (0.2041) Disabled wi h no limi a ion 0.3616 (0.5068) 0.2146 (0.3297) 0.4464* (0.3844) 0.5171* (0.3087) Cons an 0.4642** (0.1332) -0.8210** (0.2167) -1.0449** (0.1332) -0.5446** (0.1074) -0.1118** (0.1595) -1.5214** (0.0945) N 5930 5930 5930 6330 6330 6330 Pseudo R2 0.2772 0.5371 0.1700 0.5303 Rho 0.4684** 0.3984** No e: 10.0,*05.0**  pp . No e: 10.0,*05.0**  pp (Signi icance in andom e ec s models a e based on -s a s on base co-e icien s).