On the Negative Relationship between Labour Income Uncertainty and Homeownership:Risk aversion vs Credit Constraints.

On he Nega i e Rela ionship be ween Labo Income Unce ain y and Homeowne ship: Risk A e sion s. C edi Cons ain s Luis Diaz-Se ano♣ Na ional Uni e si y o I eland, Maynoo h IZA Bonn, CREB Ba celona Abs ac : In his pape we es o he i s ime whe he he d i ing o ce behind he nega i e e ec o income unce ain y on owne -occupancy p opensi ies is isk a e sion o c edi cons ain s. To disen angle his puzzle we es ima e educed o m equa ions using I alian da a. Consis en wi h he p e ious empi ical e idence in he US, ou esul s con i m ha in I aly bo h labo income unce ain y and c edi cons ain s exe a signi ican nega i e e ec on he p obabili y o homeowne ship. Howe e , ou main indings indica e ha he nega i e ela ionship be ween labo income unce ain y and homeowne ship is d i en by households’ isk a e sion. Keywo ds: Homeowne ship, income unce ain y, c edi cons ain s, Risk a e sion JEL classi ica ion: D1, R0, J0 ♣ Na ional Uni e si y o I eland, Maynoo h, Depa men o Economics (Rhe o ic House), Co. Kilda e, I eland. E-mail: [email protected] Tel: 353-1-7083793 | Fax: 353-1-7083934. 1 1. In oduc ion Undoub edly owning one’s dwelling is no only a signal o pe sonal success bu also one o he mos impo an ways o weal h accumula ion. Hence, ba ie s o homeowne ship ha e adi ionally been an impo an esea ch and policy issue. His o ically, in mos o he de eloped economies he e has been a wide ange o public policies implemen ed so as o p omo e homeowne ship. These ange om impo an ax deduc ions and gene ous subsidies o he less a o ed homebuye s o he public p o ision o a o dable dwellings aimed o low-income households. Howe e , he success o hese policies is condi ion by sca ce public esou ces and excess demand om he less a o ed popula ion g oups. Hence, a o dable lending has also been a main ocus in he mo gage ma ke s in mos o he de eloped economies. Du ing he 1990s he mo gage indus y has de o ed a g ea e o o design mo gage p oduc s o acili a e he access o homeowne ship. These ange om impo an inno a ions in a o dable mo gage lending ha educes down paymen s o a minimum amoun o mo gage paymen p o ec ion insu ance policies. The la e a e specially aimed a mi iga ing he de as a ing e ec ha income unce ain y exe s on he homeowne ship p opensi ies, hough a e wi h limi ed success (see e.g. P yce and Keoghan, 2002, and Fo d and Quilga s, 2001, in he UK, and Ross and Too ell, 2004, in he US). Gi en he lack o liquidi y o home pu chases and ha inancing he pu chase o one’s dwelling en ails unde aking a long las ing high le el o indeb edness, income unce ain y becomes i sel as impo an as he le el o income when deciding he enu e s a us. 2 P e ious empi ical li e a u e on his issue obse es an unequi ocal nega i e e ec o income unce ain y on he p obabili y o homeowne ship. Acco ding o heo e ical models, income unce ain y is expec ed o exe an e ec on he enu e decisions a e conside ing isk a e sion, howe e , his assump ion has ne e been empi ically es ed. In his pape we es o he i s ime whe he he d i ing o ce behind he nega i e ela ionship be ween homeowne ship and income unce ain y is households’ a e sion o he isk o a mo gage de aul o on he con a y i is d i en by c edi cons ain s. The ele ance o ca ying ou such a es comes om he plausible conjec u e ha households wi h mo e ola ile incomes migh be also mo e c edi cons ained han household wi h s eadie incomes. Disen angling his puzzle is c i ical in o de o design public policies, imp o e a o dabili y o lending, and de elop mo e e ec i e mo gage insu ance policies aimed a p omo ing homeowne ship among hose households mos a income isk. The emainde o he pape is s uc u ed as ollows. In sec ion 2 we e iew he li e a u e analyzing he e ec o income unce ain y and c edi cons ain s on homeowne ship. Sec ion 3 desc ibes he da a se and he empi ical amewo k. Sec ion 4 p esen s he empi ical esul s, and sec ion 5 summa izes and concludes. 2. O e iew o he li e a u e Du ing he pas decade he e ec s o income unce ain y on homeowne ship ha e ecei ed conside able a en ion by economis s. Howe e , heo e ical models inco po a ing his e ec sys ema ically end o p o ide ambiguous p edic ions; 3 income unce ain y is expec ed o exe a nega i e o non-nega i e e ec depending on di e ences in he cons uc ion and he assump ions unde lying each model. De Sal o and Eeckhoud (1982) p edic ed a nega i e ela ionship be ween housing consump ion and he p obabili y o unemploymen . Using a simila amewo k Tu nbull e al. (1991) ound ha he ela ionship be ween income unce ain y and homeowne ship is gene ally nega i e, howe e , i migh be non- nega i e i he expec ed labo income en ails compensa ing wage di e en ials o income isk. Fu (1995) analyzes he demand o housing unde he p esence o liquidi y cons ain s. He shows ha wi h inc easing liquidi y and high isk-a e sion housing in es men migh all wi h inc easing income unce ain y, howe e , wi h cons an isk-a e sion and i in es o s do no inc ease liquidi y his esul could swi ch o a posi i e e ec . Despi e he ambigui y shown by heo e ical models, he e a e some s udies ha explici ly es he e ec o income unce ain y on he homeowne ship p opensi ies and ound an unequi ocal nega i e e ec . Hau in and Gill (1987), Hau in (1991) and Robs e al. (1999) epo e idence based on US da a, and Diaz-Se ano (2004) does o Spain and Ge many. The e is also an ample ange o s udies in he US analyzing he e ec o c edi cons ain s on homeowne ship. This li e a u e s a s wi h Linneman and Wach e (1989) who, using c i e ia based on mo gage equi emen s ha se indus y s anda ds, buil indica o s e lec ing he deg ee o household’s income and weal h cons ain s ela i e o home pu chases. They obse ed a nega i e e ec o c edi cons ain s on he p obabili y o homeowne ship. 4 Using a simila amewo k Hau in e al. (1997) obse ed ha he home enu e choices among young Ame ican households a e also qui e sensible o c edi cons ain s. Rosen hal (2002) p oxy household’s c edi cons ain s using di ec esponses o ques ions aimed a asce aining whe he he household had had any eques o c edi u ned down o jus pa ially g an ed. He also obse ed a lesse p opensi y o homeowne ship among c edi cons ained households. Que cia e al. (2003) assessed he impac o a o dable lending ini ia i es on homeowne ship a es. They obse ed a posi i e e ec on homeowne ship bu also no ed ha a e con olling o bo owing cons ain s he impac is no he same ac oss he less a o ed popula ion g oups. Ba ako a e al. (2003) s udied he e olu ion o he e ec bo owing cons ain s du ing he 1990’s. They conclude ha weal h cons ain s ha e a la ge nega i e e ec han income cons ain s on homeowne ship, bu also ha his e ec is dec easing o e he las decade. Linneman e al. (1997) upda ed Linneman and Wach e and cons uc ed a simula ion model o measu e he e ec o policy changes go e ning cons ain s and mo gage in e es a es on he agg ega ed a es o homeowne ship in he US. They also ind weal h cons ain s o ha e a la ge impac han income cons ain s. Bou assa (1995) is one o he ew s udies p o iding empi ical e idence ou side he US. He eplica ed Linneman and Wach e using Aus alian da a and obse ed a nega i e e ec o c edi cons ain on he homeowne ship p opensi ies. 3. Da a, a iables and empi ical amewo k 3.1. Da a The da a we use in ou s udy comes om he I alian Su ey o Household 5 Income and Weal h (SHIW). I is a panel su ey (annual om 1977 o 1987 and biannual om 1989 o 2000) ca ied ou by Banca d’I alia (Cen al Bank o I aly). The su ey con ains de ailed in o ma ion on household cha ac e is ics, employmen , income, asse s, inancial habi s, he ype o home enu e and se e al ques ions ega ding he homeowne ship and he bo owing condi ions. Addi ionally, s a ing om 1995, he su ey also includes o a o y ques ions add essed o he s udy o speci ic issues. Fo ou pu poses, he 1995 and 2000 wa es con ain ques ions add essed o he household heads ha allow us o cons uc a measu e o indi idual isk a e sion. We use he panel om 1986 o 2000 o es ima e income unce ain y and he wa es co esponding o 1995, 1998 and 2000 o e alua e c edi quali y cons ain s and o examine he de e minan s o he homeowne ship p opensi ies. 3.2. Owning/ en ing use cos s Owning and en ing cos s is a ele an a iable when analyzing he housing enu e choices. In his pape his a iable is used in he es ima ion o he p obabili y o homeowne ship, bu also o pe o m linea eg essions on p e e ed housing alues ac oss I alian households. We de ine he cos o owning ela i e o en ing as [] () k k k V R C pm p R =++−+, (1) 6 whe e Vk and Rk a e he de la ed a e age house alues and annual en s in egion k, espec i ely, is he nominal mo gage a e, p is he p ope y ax a e, m is he main enance a e, and is he ma ginal ax a e. In equa ion (1) he nume a o speci ically e e s o he owne occupancy oppo uni y cos (Hende son and Ioannides, 1987). The cos o owning ela i e o en ing is compu ed o each o he 20 egions a ailable in ou da a1. De la ed a e age house alues o ecen pu chases, a e age annual en s, and a e age nominal mo gage in e es a es a e di ec ly aken om ou da a se . In I aly, he p ope y ax a e anges om 3 pe cen o i s ime homebuye s o 7 pe cen , and he ma ginal ax a e is 19 pe cen up o he maximum amoun o 3,615€. We use hese alues o es ima e owning cos s in equa ion (1). Following Robs e al. (1999) we assume a main enance a e o 1.5 pe cen . 3.3. Measu ing bo owing cons ain s To measu e o wha ex en a household is c edi cons ained we ollow Linneman and Wach e (1989). Using hei no a ion, he h eshold house alue ha household i should aim in o de no being income cons ained (VI) and weal h cons ained (VW) is: 0.35 , 5 IW i iii I VVW ==⋅, (2) 1 These egions a e Piemon e, Valle d’Aos a, Lomba dia, T en ino, Vene o, F iuli, Ligu ia, Emilia Romagna, Toscana, Umb ia, Ma che, Lazio, Ab uzzi, Molise, Campania, Puglia, Basilica a, Calab ia, Sicilia and Sa degna. 7 whe e is he mo gage in e es a e, I is he annual household income, and W is he ne household weal h. Since house alues a e only obse ed o homeowne s, we use a subsample o uncons ained homeowne s, hose wi h obse ed house alue lowe 85 pe cen o bo h VI and VW, o es ima e he ollowing housing demand equa ion * iii VX u β =+, (3) whe e V* is he p e e ed house alue, X is a ec o o household cha ac e is ics which also includes house p e e ences, β is a pa ame e ec o o be es ima ed, and ui is a andom e o e m. In a second s age we use ˆ β o impu e a * ˆi V o each household ei he homeowne o en e . Hence, we assume ha a household is income cons ained (IC) o weal h cons ained (WC) i * ˆ0.9 I ii VV>⋅ o * ˆ0.9 W ii VV>⋅, espec i ely. Equa ion (3) is es ima ed using a pooled sample co e ing he wa es 1995, 1998 and 2000. Since using pooled da a may lead o ine icien es ima es we jus selec he las wa e he household has pa icipa ed. The explana o y a iables (X) in equa ion (3) a e a se o a iables ega ding he household, i.e. a squa ed polynomial on household income and he numbe o child en; some cha ac e is ics o he household head, i.e. a squa ed polynomial on age, ma i al s a us and gende ; a se o geog aphical dummies, i.e. egion and ci y size; he cos s o owning ela i e o en ing and yea dummies. 8 The linea es ima ion by OLS o he p e e ed housing alue equa ion (3) is epo ed in able 1. Mos o he a iables conside ed a e signi ican a 1 pe cen o be e . Bo h he household income and he age o he household head show a posi i e bu dec easing e ec . Highe p e e ed house alues a e obse ed in he No h-Eas o I aly and he Islands, and i is dec easing wi h he ci y size. Inse able 1 a ound he e Addi ionally, ou da ase also p o ides some ques ions ha allow us di ec ly measu e bo owing cons ain s in he same way as in Rosen hal (2002). These ques ions a e: C54. Du ing he las 12 mon hs did you household apply o a bank o a inancial company o a loan o a mo gage? C55. Was he applica ion g an ed in ull, in pa o ejec ed? C56. Du ing he las 12 mon hs did you o ano he membe o you household conside he possibili y o applying o a bank o a inancial company o a loan o a mo gage bu hen change his/he mind hinking ha he applica ion would be ejec ed? F om answe s o C54-C56 we c ea e a dummy a iable ha akes 1 i he loan was denied, jus pa ly g an ed, o i any membe e ained om applying conce ned o being u ned down. We call his di ec p oxy o being c edi cons ained DCC. Addi ionally o IC and WC we also use DCC o examine he e ec o c edi cons ain s on he p obabili y o homeowne ship. Table 2 shows a 15 coe icien s o be es ima ed, and ν i is he e o e m. I we expand equa ion (7) o conside he e ec o bo h c edi cons ain s (CC) and labo income unce ain y (2i ε σ ) as measu ed in exp ession (5) we ha e *' 2 ˆ ( 0) ( 0) ( 1,..., ) ii i iii yIy IH CC i N ε δλσπν =>= + ++> = , (8) F om equa ion (8) we should expec λ <0 (e.g., Linneman and Wach e , 1989, Bou assa, 1995, Hau in e al., 1997, Rosen hal, 2002, and Ba ako a e al., 2003), and π <0 (Hau in, 1991, Robs e al., 1999, and Diaz-Se ano, 2004a). Es ima es o he pa ame e s in equa ion (8) coming om a single uni a ia e p obi model will be, howe e , inconsis en i ν i and he e o e m o a po en ial bina y equa ion whe e CCi is he endogenous a iable a e co ela ed (see Woolb idge 2002, p. 477). In his con ex , he bi a ia e p obi model would p o ide consis en es ima es. As men ioned ea lie , he main ocus o he pape aims o es whe he he e ec o labo income unce ain y ( 2i ε σ ) is d i en by household’s isk-a e sion ( o e.g. a mo gage de aul ) o on he con a y i is d i en by household’s c edi cons ain s (e.g. no access o he mo gage ma ke ). This es equi es es ima ing equa ion (8) o di e en popula ion g oups; c edi s. non-c edi cons ained, and isk s. non- isk a e se. The size and he signi icance o he es ima ed e ec s o π in equa ion (8) o each o hese popula ion g oups will allow us disen angling he puzzle. As in Rosen hal (2002), he model consis s o wo equa ions and can be exp essed as ollows 16 *' 11 111 ( 0) ( 0) ( 1,..., ) ii ii yIy IH i N δν =>= +> = *'2 22 2222 ˆ ( 0) ( 0) ( 1,..., ) ii iii yIy IH i N ε δσπν =>= ++> = (9) whe e y i1* is he la en a iable indica ing he p opensi y o be o no c edi cons ained, o isk o non- isk a e se, and yi2* is he la en indica o ega ding he p opensi y o homeowne ship, whe e 12 (, ) ii ν ν ~(0,0,1,1, )BVN ρ . The ma ixes H1i and H2i do no need o con ain he same a iables. The explana o y a iables in he enu e choice equa ions (8) and (9) a e a se o household head’s cha ac e is ics, i.e educa ion, a squa ed polynomial on age, ma i al s a us, and sel -employmen ; a se o household’s a iables, i.e. size, numbe o dependen membe s (no income ecipien s), household income; a se o dummies collec ing he c edi si ua ion (ou s anding bank deb s); a se o geog aphical dummies, i.e. egional dummies, ci y size and household’s loca ion; and as ou key a iables we include labo income unce ain y and a se o dummies ega ding c edi cons ain s. Howe e , since in he sys em o equa ions (9) he enu e choice equa ion is speci ically es ima ed o c edi o non-c edi cons ain households, dummies e e ing o c edi cons ain s a e d opped om his equa ion. In he sys em o equa ions (9) we ace bo h a censo ing and obse a ion ule o bo h yi1 and yi2, which lead us o conside he sample selec ion issue. Hence, we need o con ol o co ela ion be ween he e o e ms and he sequence o “choices”. Fo each enu e ou come we ha e h ee ypes o obse a ion: being (non-)c edi cons ained; being homeowne ; and no being homeowne . Analogously, we can d aw he same sequence in he case ha he i s la en 17 indica o (yi1*) e e s o he p opensi y o being o no isk a e se. The uncondi ional p obabili ies o his bina y ee a e gi en by: 12 21122 12 21122 111 (1 1) (, ,) (1 0) (, ,) (0)1() ii ii ii ii ii Py y H H Py y H H Py H δ δρ δ δρ δ ===Φ ===Φ −− ==−Φ ∩ ∩ (10) whe e Φ and Φ2 deno e he uni a ia e and bi a ia e s anda d no mal cumula i e dis ibu ion unc ions, espec i ely. And, he esul ing log-likelihood unc ion is gi en by () 1 2 11 2 21122 1 1 211 22 11 10 0 log ( , , ) log ( , , ) log 1 ( ) , i i ii i ii y y ii i yy y LogL H H HH H δδρ δδρ δ = = == = =Φ + +Φ −−+ −Φ ∑ ∑∑ (11) Finally, o es ima e he models (8) and (9) we es ic ou sample o homeowne s ha ing ou s anding mo gage paymen s and ha pu chased hei dwelling a e 1989. By applying he o me es ic ions we keep ou o he sample households ha ha e pu chased hei dwelling oo long ago, ha e no needed a mo gage o ha e inhe i ed he dwelling. Ob iously, hese households migh ne e expe ience a mo gage de aul , he e o e, hey a e no expec ed o ollow he same choice ules ha homeowne s ha a e cu en ly mo gage bo owe s. By conside ing hese households in he sample he ue ela ionship be ween income 18 unce ain y and he p obabili y o homeowne ship migh be “obscu ed”. Gi en ha he necessa y in o ma ion o know whe he a household is c edi cons ained o isk-a e se is only a ailable om 1995, o es ima e ou econome ic model we pool he co esponding c oss-sec ions o 1995, 1998 and 2000. In o de o a oid ine iciency we jus ake he las wa e he household has pa icipa ed. 4. Econome ic esul s Table 6 and 7 epo he econome ic es ima ion o he p obabili y o homeowne ship. Table 6 ocuses on he uni a ia e and bi a ia e p obi es ima es o e alua e he e ec o labo income unce ain y and c edi cons ain s on he p obabili y o homeowne ship (equa ion 8). Recall ha gi en he nonlinea na u e o he uni a ia e and bi a ia e p obi models, he es ima ed coe icien s lack any economic in e p e a ion and a e jus used o de e mine he sign o he ela ionship. Howe e , a his s age his is jus wha we a e in e es ed in. The c edi cons ain equa ion in he bi a ia e p obi model is included o a oid he inconsis ency o he pa ame e s in he homeowne ship equa ion. I is wo h no ing ha co ela ion be ween bo h equa ions u ns ou o be highly signi ican (0 ρ ≠), which means ha con olling o his co ela ion is c i ical. Households wi h olde household heads and wi h mo e dependen membe s ha e highe p opensi y o be c edi cons ained. Household income, educa ion o he household head and his/he sel -employmen s a us exe s a nega i e e ec on such a p opensi y. This equa ion also includes a se o dummy a iables collec ing he e ec o he ou s anding bank deb s o he pu chase o di e en goods. We ind 19 his se o a iables o be mo e impo an on weal h cons ain s han on income cons ain s. Ou main indings conce n he owne -occupancy equa ion. Conside i s he ole o he c edi cons ain s. Consis en wi h he p e ious e idence, we ind ha in I aly bo h weal h cons ain s (WC) and income cons ain s (IC) exe a signi ican and nega i e e ec on he p obabili y o homeowne ship, hough wi h a ema kably la ge e ec o he weal h cons ain s. The o me esul coincides wi h wha was obse ed in he US5. No e ha ou p oxy o c edi cons ain s (DCC) based on di ec ques ions shows also a signi ican and nega i e e ec . The es ima es coming om he bi a ia e p obi model also con i m ha he c edi cons ain pa ame e s in he homeowne ship equa ion a e qui e sensible o he omission o he endogenous na u e o he household’s c edi cons ain s p opensi ies. We ind a signi ican downwa ds bias in he c edi cons ain s pa ame e s coming om he uni a ia e p obi model, -0.38 s. -0.78 o income cons ain s (models 1 and 2, able 6) and – 3.33 s. –4.03 o weal h cons ain s (models 3 and 4, able 6). Tu ning ou a en ion o he e ec o income unce ain y on homeowne ship, consis en wi h he p e ious empi ical e idence o he US, Ge many and Spain, we also obse e a signi ican and nega i e e ec in I aly. O he di e ences ac oss models a e obse ed in he e ec o o he a iables conside ed in he owne - occupancy equa ion, bu he signs pe sis en ly emain in all models. Owne - 5 We ha e also ca ied ou a numbe o al e na i e speci ica ions. In models whe e income and weal h cons ain s a e simul aneously conside ed, income cons ain s ha e u ned ou o be s a is ically insigni ican . The e o e, he e ec o weal h cons ain s domina es o e income cons ain s. Al e na i ely, we ha e also es ima ed a i a ia e p obi model using he simula ed maximum likelihood me hod ha simul aneously es ima es he p obabili y o homeowne ship and he weal h and income cons ain s p opensi ies, and once mo e we obse e ha he weal h cons ain s e ec domina es o e he income cons ain s e ec . 20 occupancy is mo e likely in smalle ci ies (less han 500,000 inhabi an s), and ou o he ci y cen e and in isola ed a eas. As expec ed, household head age, amily income and being ma ied aise he p opensi y o own, whe eas household size and educa ion exe a nega i e e ec . This coun e in ui i e esul con as s wi h he obse ed in Spain, bu coincides wi h p e ious e idence o Ge many (see Diaz- Se ano, 2004). Inse able 6 a ound he e Tu ning o ou majo indings, we ocus now on he es ima es o he e ec o income unce ain y on he p obabili y o homeowne ship shown in able 7. Recall ha his es ima es comes om he sys em o equa ions (9). Al hough all models in able 7 include he same explana o y a iables (excep c edi cons ain s in he homeowne ship equa ion) han he es ima es epo ed in able 6, o he sake o simplici y we ocus on he es ima ed pa ame e s and e ec s associa ed o labo income unce ain y in he homeowne ship equa ion. To acili a e in e p e a ion and he compa ison be ween al e na i e models we also epo he a e age ma ginal e ec s (APE). The single e ec o labo income unce ain y in he homeowne ship equa ion can be compu ed as in he uni a ia e p obi as 2 22 22 2 () ii i H φ δσππ +⋅, whe e φ is he s anda d no mal densi y and π 2 is he pa ame e associa ed o he labo income unce ain y (Ch is o ides e al., 1997). We ha e es ima ed he APE o di e en popula ion g oups depending on whe he hey a e o no c edi cons ained, and whe he hey a e o no isk a e se. 21 These esul s a e c ucial o de e mine he na u e o he nega i e ela ionship be ween homeowne ship and labo income unce ain y. Fi s a all, we shall ema k ha he high s a is ical signi icance o he co ela ion e ms sugges s ha con olling o sample selec ion is c i ical o ob ain unbiased es ima es. Di e ences in he APE o labo income unce ain y be ween c edi and non- c edi cons ained a e ai ly modes , -0.067 s. -0.078 o weal h (un)cons ained, espec i ely, and –0.289 s. -0.236 o income (un)cons ained, espec i ely. Majo di e ences in his nega i e ela ionship a e epo ed when conside ing isk- a e se s. non isk-a e se households, -0.319 s. –0.041, espec i ely. Fo he isk a e se, a 10 pe cen inc ease in he a e age labo income unce ain y dec eases abou -3.25% he p obabili y o being homeowne , whe eas o he non- isk a e se his e ec has u ned ou o be s a is ically insigni ican . Fo income (un)cons ained households hese pe cen ages a e –2.95% s. –2.41%, espec i ely, and –0.68% s. –0.80% o weal h (un)cons ained households. These esul s sugges ha he nega i e link be ween income unce ain y and he homeowne ship p opensi ies is d i en by isk-a e sion. 5. Conclusions and discussion In his pape we ha e in es iga ed he e ec o labo income unce ain y and c edi cons ain s on he p obabili y o homeowne ship in I aly. Consis en wi h he p e ious empi ical e idence, we ind ha in I aly c edi cons ained households and wi h mo e ola ile incomes a e less likely o own hei dwelling. As in he US, we also obse e ha al hough bo h ypes o cons ain s a e impo an , he weal h cons ain s e ec domina es o e he income cons ain s e ec . 22 As he main ocus o his pape we ha e in es iga ed o he i s ime he unde lying na u e in he nega i e link be ween labo income unce ain y and homeowne ship. To ca y ou he es we ha e pe o med educed o m es ima es using he bi a ia e p obi model wi h sample selec ion. We obse e ha he gap in he es ima ed nega i e e ec o income unce ain y on homeowne ship be ween c edi and non-c edi cons ained households is imma e ial. Howe e , a ma kedly la ge gap is ound be ween isk and non- isk a e se households, and being indeed s a is ically insigni ican o he non- isk a e se. These esul s indica e ha he nega i e e ec o income unce ain y on he homeowne ship p opensi ies is d i en by household’s isk a e sion. The co olla y o ou esul s sugges s ha ins i u ions and he banking indus y should de o e g ea e e o s o p omo e homeowne ship among hose households mos a income isk. P obably, mo e e icien mo gage p o ec ion paymen insu ance policies should be designed in o de o mi iga e he de as a ing e ec o income unce ain y on he homeowne ship p opensi ies o he mo e isk-a e se households. We ind his is an impo an issue ha s ill is unde - esea ched. 23 Re e ences Ba ako a I., R.W. Bos ic, P.S. Calem, S.M. Wach e , 2003, Does c edi quali y ma e o homeowne ship?, Jou nal o Housing Economics 12, 318-336. Bou assa S.C., 1995, The impac s o bo owing cons ain s on home-owne ship in Aus alia, U ban S udies 32, 1163-1173. Ch is o ides L.N., T. S engos, R. Swidinsky, 1997, On he calcula ion o ma ginal e ec s in he bi a ia e p obi model, Economics Le e s 54, 203-208. DeSal o, J.S., L.R. Eeckhoud , 1982, Household beha io unde income unce ain y in a monocen ic u ban a ea, Jou nal o U ban Economics 11, 98-11. Diaz-Se ano L., 2004, Labo income unce ain y, isk-a e sion and homeowne ship, IZA discussion pape #1008, IZA Bonn. Fo d, J., Quilga s, D., 2001. Failing home owne s? The e ec i eness o public and p i a e sa e y-ne s. Housing S udies 16, 147-162. Fu Y., 1995, Unce ain y, liquidi y, and housing choices, Regional Science and U ban Economics 25, 223-236. Guiso L., M. Paiella, 2001, Risk a e sion, weal h and backg ound isk, CEPR discussion pape #2728, CEPR London. Ha og J., A. Fe e -i-Ca bonell, N. Jonke , 2002, Linking measu ed isk a e sion o indi idual cha ac e is ics, Kyklos 55, 3-26. Ha og J., E.J.S. Plug, L. Diaz-Se ano, A.J.C. Viei a, 2003, Risk compensa ion in wages: a eplica ion, Empi ical Economics 28, 639-647. Hau in D.R., P.H. Hende sho , S.M. Wach e , 1997, Bo owing cons ain s and he enu e choice o young households, Jou nal o Housing Resea ch 8, 137-154. 24 Hau in, D.R., H.L. Gill, 1987, E ec s o income a iabili y on he demand o owne -occupied housing, Jou nal o U ban Economics 22, 136-150. Hau in, D.R., 1991, Income a iabili y, homeowne ship, and housing Demand, Jou nal o Housing Economics 1, 60-74. Hau in, D.R., P.H. Hende sho and D. Kim, 1994, Housing decisions o Ame ican you h, Jou nal o U ban Economics 35, 28-45. Hende son, J.V. and Y.M. Ioannides, 1987, Owne occupancy: consump ion s. in es men demand, Jou nal o U ban Economics 21, 228-241. Hsiao, C., 1986, Analysis o panel da a. (Camb idge Uni e si y P ess, Camb idge). Linneman P.D. and S.M. Wach e , 1989, The impac s o bo owing cons ain s on homeowne ship, AREUEA Jou nal 17, 389-402. McGold ick K., 1995, Do women ecei e compensa ing wages o ea nings unce ain y?, Sou he n Economics Jou nal 62, 210-222. O alo-Magne, F., and S. Rady, 2002, Tenu e choice and he iskiness o non- housing consump ion, Jou nal o Housing Economics 11, 266-279. P yce, G., Keoghan, M., 2002. Unemploymen insu ance o mo gage bo owe s: is i iable and does i co e hose mos in need?. Eu opean. Jou nal o Housing Policy 2, 87-114. Que cia, R.G., G.W. McCa hy, and .M. Wach e , 2003, The impac s o a o dable lending e o s on homeowne ship a es, Jou nal o Housing Economics 12, 29-59. Robs , J., R. Dei z, and K. McGold ick, 1999, Income a iabili y, unce ain y and housing enu e choice, Regional Science and U ban Economics 29, 219-229. Ross, S.L., Too ell, G.M.B, 2004. Redlining, he communi y ein es men ac , and p i a e mo gage insu ance. Jou nal o U ban Economics 55, 278-297. 31 Table 6: Es ima es o he e ec o labo income unce ain y and c edi cons ain s on he p obabili y o homeowne ship (uni a ia e and bi a ia e p obi s). Uni a ia e P obi (Model 1) Bi a ia e P obi (Model 2) Uni a ia e P obi (Model 3) Bi a ia e P obi (Model 4) Homeowne ship Homeowne ship IC Homeowne ship Homeowne ship WC Coe . z-s a Coe . z-s a Coe . z-s a Coe . z-s a Coe . z-s a Coe . z-s a Cons an e m 1.3253 4.61 1.6403 5.88 2.0252 12.76 3.7505 9.07 4.1114 11.33 2.2036 7.74 Age -0.0160 -1.86 -0.0188 -2.25 0.0158 8.64 -0.0503 -3.70 -0.0481 -4.06 0.0090 6.87 Age squa ed 0.0001 1.11 0.0001 1.63 0.0004 3.21 0.0004 3.79 Household size -0.1056 -5.25 -0.1069 -5.53 -0.1015 -3.65 -0.1136 -4.61 Ma ied 0.2942 6.04 0.2640 5.53 0.3137 4.18 0.2572 3.79 Dependen 0.2617 10.80 -0.0124 -0.76 Yea s o Schooling 0.0148 2.37 0.0087 1.47 -0.0349 -6.88 -0.0255 -3.01 -0.0340 -4.51 -0.0361 -6.02 Sel -employed 0.1738 3.36 0.1939 3.77 0.0960 1.24 -0.3650 -5.55 -0.4545 -7.04 -0.4621 -7.62 Income unce ain y -0.5466 -5.55 -0.5239 -5.95 -0.2968 -2.73 -0.3123 -3.46 Family income 1.2⋅10-5 7.04 9.3⋅10-6 5.91 -0.0001 -38.89 3.9⋅10-6 3.01 8.8⋅10-7 0.87 -2.3⋅10-5 -19.23 Rela i e cos o owning -0.0005 -3.26 -0.0004 -2.67 -0.0003 -1.74 -0.0002 -1.34 Loca ion dummies (base-o he s) Isola ed – coun yside 0.2208 1.95 0.2294 2.07 0.0420 0.24 Town ou ski s -0.3659 -4.47 -0.3526 -4.37 -0.3660 -3.01 Be ween ou ski s and ci y cen e -0.3606 -4.39 -0.3312 -4.11 -0.3967 -3.29 Ci y cen e -0.4840 -5.71 -0.4489 -5.38 -0.5692 -4.62 32 Ci y size (base < 500,000 inhab.) >500.000 inhab. -0.3676 -6.60 -0.3456 -6.34 -0.3328 -4.88 C edi cons ain s DCC (Di ec answe om esponden s) -0.4970 -4.43 -0.4917 -4.49 -0.4963 -3.10 -0.4476 -3.27 IC (Income cons ained) -0.3881 -5.93 -0.7887 -9.94 WC (Weal h cons ained) -3.3366 -45.26 -4.0331 -51.64 Ou s anding bank deb dummies Pu chase o eal goods -0.1362 -0.22 0.2077 0.63 Pu chase o mo o ehicles -0.1321 -1.38 0.1401 2.35 Pu chase o u ni u e, elec ical appliances 0.2601 2.17 0.2290 2.88 Pu chase o non-du able goods 0.2742 0.92 0.4807 2.96 ρ Wald es H0:ρ=0 0.4485 55.56 0.7874 9.94 Sample size 5,845 No e: All he equa ions include yea and egional dummies (20 egions). 33 Table 7: Es ima es o he e ec o labo income unce ain y on he p obabili y o homeowne ship (bi a ia e p obi s wi h sample selec ion) Bi a ia e p obi wi h sample selec ion Coe icien z- alue ρAPE ∆p(y2i=1) wi h 10% inc ease in 2 ˆi ε σ Wald es H0:ρ=0 Sample size Income cons ained (IC) IC=1 -0.8081 -6.09 0.0450 -0.2894 -2.95% 1.16 5,845 IC=0 -0.5976 -6.44 -0.2362 -2.41% 1.16 5,845 Weal h cons ained (WC) WC=1 -0.1678 -1.55 -0.9827 -0.0669 -0.68% 1115.67 5,845 WC=0 -0.1968 -3.53 -0.0785 -0.80% 1115.67 5,845 Risk and non isk a e se Risk a e se=1 -0.8029 -4.70 0.0910 -0.319 -3.25% 7.89 2,944 Risk a e se=0 -0.1041 -0.35 -0.041 -0.42% 7.89 2,944