Gender Differentials in the Housing Markets in Latin America

Transcription

1 Gender Dfferentals n the Housng Markets n Latn Amerca Nestor Gandelman Unversdad ORT Uruguay * September 2006 Second Draft Abstract The gender of the household head has often been treated as an exogenous determnant of housng tenure. We argue that several determnants of homeownershp also affect household headshp and that falng to explctly account for ths endogenety leads to nconsstent results. Usng ndvdual level data for Chle, Honduras and Ncaragua we show that although on average women have lower probablty of beng homeowners, those women that head ther famles (sngle, separated or dvorced) have larger probabltes of attanng homeownershp. Thus household level analyss should control for the endogenety of household headshp n order to properly address the gender effect on housng tenure. We estmate a bvarate probt countres and fnd evdence that all else equal female headed famles have lower probablty of ownng ther home n fourteen (out of seventeen) Latn Amercan countres. Wthout the endogenety control ths evdence was not present n eleven out of ths fourteen countres. * The authors wsh to thank Alexs Avcharan for excellent research assstance and the MECOVI program for help n the access to the household surveys data. Ths paper benefted from comments from Eduardo Lora, Arturo Galndo, Hugo Ñopo, Clauda Pras, Claudo Santbañez and Vctora Rodrguez. The usual dsclamer apples.

2 As stated n IADB (2004) Poverty s both cause and effect of poor housng condtons. Lack of effectve demand resultng from the low ncome of households s the underlyng cause that prevents the prvate provson of houses. Conversely, mprovng housng condtons can have a major nfluence on poverty allevaton through mprovements n the lvng standards of low ncome famles, and on poverty reducton va ncreased employment opportuntes. Therefore, understandng the determnants of housng tenure and potental gender dscrmnaton s mportant for poverty reducton polces. The study of the determnants of housng tenure and the concerns wth possble dscrmnaton has been on the research agenda even before approprate econometrc technques were commonly used. L (1977) s the frst paper that goes beyond lnear models and estmates a logt model to the determnants of homeownershp but does not consder the gender of the household head. Several types of varables have receved most of the attenton of the researchers: ncome and wealth, lfe cycle status, locaton and neghborhood attrbutes and a varety of socoeconomc ndcators. In partcular, much attenton has been gven to the racal or ethnc orgn of the father. There s substantal evdence of racal dscrmnaton n the access to mortgage credt and homeownershp. The gender economc dscrmnaton lterature has also spent lots of efforts to study the exstence of dscrmnaton on dmensons lke salares, promotons, etc. One of the most common strateges s to nclude as an explanatory varable a dummy for women and conclude that f the estmated coeffcent s sgnfcantly dfferent from zero, women receve a dscrmnatory (postve or negatve) treatment wth respect to men. It s therefore strkng the absence of comments on dscrmnaton n the studes of the determnants of homeownershp. The reason s that most studes fnd more favorable outcomes for female headed famles or do not fnd sgnfcant results at all. 1 Gven the outcomes, n other contexts, of the gender dscrmnaton 1 Van Leuvenstejn and Konng (2004) and Gandelman and Gandelman (2004) fnd that women have hgher probabltes of ownng ther household n the Netherlands and Uruguay respectvely. Chur and Jappell (2003) and Armah (1997) do not fnd gender dfferences n fourteen OECD countres and Ngera respectvely. Manrque and 2

3 lterature these results are surprsng. We argue that the determnants of women household headshp and those of homeownershp are correlated and therefore the specfcaton used n most studes has an endogenety problem that leads to nconsstent and often counterntutve results. If women s martal status s not exogenous to the tenure choce, then, even n the presence of dscrmnaton aganst women n the housng markets, a nave vew of the data may reflect that women headed households have hgher probabltes of ownng ther home. For nstance, those women that do not have a place were to lve, have lower ncome, have more chldren, etc. wll probably not dvorce ther husbands even f they want to. There s a selecton bas n whch women headed famles tend to have better socoeconomc ndcators than what they would have f female headshp were a completely random process. Ths explans, why a smple vew may fnd that women headed famles are more lkely to own they house. Thus, the gender of the household head can not be treated as other truly exogenous characterstcs lke race and ethnc orgn. To the best of our knowledge ths s the frst paper that focuses on the factors affectng homeownershp and household headshp jontly by explctly provdng an econometrc soluton to the endogenety ssues that arse by the jont determnaton of both varables. Our results for seventeen Latn-Amercan countres show that the bases are mportant and that female headed famles have a substantally worse market outcome n terms of homeownershp. I. Data Thanks to the collaboraton of the MECOVI program and the natonal nsttutes of statstcs we were able to have access to the household surveys of seventeen Latn Amercan countres. The countres ncluded n ths study are: Argentna, Bolva, Brazl, Chle, Colomba, Ecuador, Peru, Paraguay, Uruguay and Venezuela from South Amerca; Costa Rca, El Salvador, Guatemala, Ojah (2003) found that men are more lkely to own ther household but women tend to have hgher household expendture n Span. 3

4 Honduras, Ncaragua and Panama from Central Amerca and Mexco. Table A1 n the appendx presents detaled nformaton on the data sources. Table 1 presents the housng tenure structure for the countres covered n ths study. Argentna s the only country that does not dstngush those that own ther house and are stll payng for t from those that have already fnshed payng. On average 72% of all households own ther home, 14% rent and 13% use a house wth or wthout owners approval. Venezuela, Panama, Paraguay and Ncaragua have the largest share of homeowners and lower shares of renters. Colomba s the opposte case wth the lower ownershp rato. Ths may n part be due to the nternal forced mgraton that many Colombans faced on the last decades. The mortgage market seems to be more developed n Chle, Costa Rca, Panama and Uruguay, beng the only countres where more than 10% of households own ther home but are sll payng for t. 2 Own, already pad Table 1. Housng tenure Own, stll payng Rent User wth or wthout owner approval Cases Argentna 72.6% 14.8% 12.6% 26,285 Bolva 61.2% 2.2% 16.2% 20.4% 4,832 Brazl 69.7% 4.5% 14.8% 11.1% 107,840 Chle 61.8% 10.4% 11.2% 16.6% 68,153 Colomba 45.7% 6.8% 34.3% 13.2% 22,949 Costa Rca 65.1% 10.3% 13.4% 11.2% 11,032 Ecuador 63.2% 4.7% 17.8% 14.3% 18,959 El Salvador 64.2% 5.6% 11.2% 19.1% 16,808 Guatemala 59.9% 1.8% 18.6% 19.7% 2,784 Honduras 69.2% 3.7% 13.4% 13.7% 7,983 Mexco 67.3% 5.9% 14.0% 12.8% 22,130 Ncaragua 77.0% 0.6% 3.2% 19.2% 4,171 Panama 67.3% 11.0% 10.0% 11.7% 6,344 Paraguay 76.6% 1.3% 8.4% 13.7% 9,591 Peru 68.9% 0.4% 10.2% 20.5% 2,163 Uruguay 57.3% 10.6% 16.8% 15.3% 18,338 Venezuela 74.8% 6.2% 9.7% 9.3% 46,287 Source: Own elaboraton based on countres household survey 4

5 I.1 Measurement errors There are potentally dffcultes n measurng the two man varables of our analyss. Frst, the status of household head s self-declared. Female household headshp does not necessary mply that the martal status of the woman of the house s dvorced. In prncple, a woman household head could be sngle, marred, dvorced or a wdower. But gven the household self-declaraton of headshp, t s not surprsng that n practce n Latn Amerca there are very few cases of marred female household heads. To deal wth ths ssue we explored the use of alternatve objectve defntons of household headshp, e.g. assgnng the household headshp to the man ncome provder or to the more educated household member. We found no sgnfcant dfferences n the man results of the paper. The second measurement problem s that for most countres home ownershp s not observed at the ndvdual level but only at the household level,.e. we do not know whch member of the famly s the legal owner of the house. Therefore most of our analyss has to be carred out the household level rather than the ndvdual level as s more tradtonal n the dscrmnaton lterature. There s precse nformaton on the dentty of the household owner for Chle, Honduras and Ncaragua. For these three countres, we present our analyss at the ndvdual level that confrms female worse outcomes n terms of homeownershp and the endogenous nature of household headshp. The endogenety stressed n ths paper has to do wth the explct decson of women to head ther famly. Ths could take the form of a sngle mother or a dvorced or separated woman. Although wdows may also be female household heads they became so only after the passng of ther partner. Therefore besdes crmnal cases, women do not take a decson to become wdows so the endogenety wth homeownershp s not present. In our household level estmatons and summary statstcs we exclude households headed by wdows for all countres but Brazl and Ecuador that do not report cvl status. 2 In Uruguay the state owned Banco Hpotecaro del Uruguay has a market share of more than 80% of all mortgage housng credt (Gandelman and Gandelman 2004). As a result of a severe fnancal crss n 2002, currently ths source of home fnance s not avalable any more. 5

6 II. Housng dfferentals - Indvdual level analyss II.1 Methodology The tradtonal approach to estmatng the determnants of homeownershp s to postulate a structural equaton Own * = x' β + ε (1) * where Own = 1 f Own > 0 and ε s an error term assumed to dstrbute normal or logstc. All explanatory varables n x are assumed to be exogenous. We are n the presence of gender dfferental effects, f all other thngs equal, women have lower probablty of ownng ther home. In order to test ths gender dfferental treatment one of the regressors would be a gender dummy. When the estmaton s carred out at the ndvdual level there are no problems wth the gender varable snce, even n these days, sex as race or ethnc orgn s not a choce varable and could be taken as exogenous. II.2 Results The only three countres were we could observe who s the actual owner of the house are Chle, Honduras and Ncaragua. For ths three countres we present for each country n column A of Table 2 an average effect on the probablty of ownng ther home for a woman and n column B we desegregate ths effect by types of women. In partcular we dstngush the sngle women headng a famly, sngle women not headng a famly (e.g. daughters lvng wth ther parents), women lvng wth her couple (marred or not), dvorced or separated women and fnally wdows. As expected rcher, older and more educated people are more lkely to own ther home. After controllng for these varables, the probablty of a woman to own ther home s lower than the 6

7 probablty of men for the three countres and thus we are n the presence of gender dscrmnaton. But ths result s not homogenous for all types of women, when desegregatng the analyss, we observe that separated women or sngle female household heads have hgher probablty of ownng ther home. Thus, although we have already establshed for these three countres that women have lower probablty of ownng ther home, when observng data aggregated at the household level, we wll lkely have that female household headshp s assocated wth a larger probablty of homeownershp. Ths s not the true gender effect, t rather reflects the fact that those women that felt they could head ther famly have larger probabltes of achevng ownershp. Table 4. Determnants of the probablty of homeownershp ndvdual level data Chle Honduras Ncaragua A B A B A B Woman [0.012]*** [0.025]*** [0.028]*** Woman-Sngle-Not household head [0.024]*** [0.060]*** [0.093]*** Woman-Sngle- household head [0.033] [0.058]*** [0.258]** Woman Separated [0.029]** [0.074] [0.046]*** Woman Couple [0.015]*** [0.030]*** [0.034]*** Woman Wdow [0.024]*** [0.070]** [0.068] Income [0.008]*** [0.008]*** [0.001]*** [0.001]*** [0.007]*** [0.007]*** Age [0.000]*** [0.000]*** [0.001]*** [0.001]*** [0.001]*** [0.001]*** Schoolng [0.002]*** [0.002]*** [0.003]*** [0.003]*** [0.004] [0.004] Illterate [0.025]*** [0.025]*** [0.073]*** [0.074]*** [0.039]*** [0.040]*** Constant [0.023]*** [0.024]*** [0.041]*** [0.042]*** [0.039]*** [0.039]*** Observatons Note: Own=1 f ndvdual owns the house, Wom=1 for females, Schoolng s years of formal educaton, Illterate=1 f the ndvdual does not know how to read and wrte. Standard errors n brackets * sgnfcant at 10%; ** sgnfcant at 5%; *** sgnfcant at 1% 7

8 III. Housng dfferentals -Household level analyss III.1 Methodology When the estmaton s carred out at the household level, the Gender dummy wll equal one n the presence of a female household head but household headshp s not exogenous. For nstance, there s evdence that dvorces are affected by several ncome and welfare varables. Shroder (2002) revews the evdence on ndrect effects of housng assstance on the self-suffcency of asssted famles. He concludes that there s a strong assocaton of housng assstance wth sngle-adult household formatons. Other papers that report smlar evdence nclude Danznger et al. (1982) and Hannan and Tuma (1990). It s therefore natural to assume that some of the varables that ncrease the probablty of ownng a house also ncrease the probablty of observng women headed famles. If ths endogenety s neglected the estmated coeffcents of model (1) are nconsstent. Snce for most countres the nformaton about homeownershp s at the household level rather than the ndvdual level (.e. we know f a member of the household owns the house but not whom), we need to provde a remedy for the endogenety that arses at the household level analyss. Therefore to estmate the dfferental effect of household head by women we postulate a bvarate probt model n whch t s possble to test whether woman headshp and housng tenure are exogenous. The model s based on two structural equatons. * Own = β 1' x + γ1woman + ε 1 (2) * 1 2' x2 + γ 2Own 2 Woman = β + ε * * where Own and Woman are latent varables, Own and Woman are dchotomous varables that take the followng values: x 1 and * 1f Own > 0 Own = 0 otherwhse are vectors of exogenous varables, * 1f Woman > 0 Woman = 0 otherwhse x2 1 β and β 2 are vector of parameters, γ 1 and γ 2 area scalar parameters and the error terms are assumed to be dstrbuted bvarate normal wth mean 0, varance 1 and correlaton Cov ( ε ε ) = ρ 1, 2. Whle the bvarate probt model can be 8

9 dentfed based on the functonal form assumptons of the jont normal dstrbuton and therefore there s no need for any extra dentfcaton strateges some of the determnants of homeownershp should not affect the gender headshp regresson and vce versa. As shown n Greene (1998) and Greene (2003), despte the endogenety of woman headshp, a multple equaton specfcaton for two dchotomous varables lke equaton (2) can be consstently estmated by Full-Informaton Maxmum Lkelhood (FIML) methods. 3 The ntuton behnd ths result s that the four probablty terms that enter the lkelhood functon can be decomposed n the condtonal and the margnal dstrbuton for women. For nstance, ( Own = 1,Woman = 1 ) = P( Own = 1Woman = 1) P( Woman = 1) P. The loglkelhood functon to be maxmzed s gven by: l N ( β ) = [ d P + d P + d P d P ] where: d d d d = = = = OwnWoman P = P( Own = 1,Woman = 1) = Φ ( β1' x1 + γ, β 2 ' x2, ρ) 10 Own ( 1 Woman ) P = P( Own = 1,Woman = 0) = Φ ( β1' x1 + γ, β 2 ' x2, ρ ) 01 ( 1 Own ) Woman P = P( Own = 0,Woman = 1) = Φ ( β1' x1, β 2 ' x2, ρ) 00 ( 1 Own )( 1 Woman ) P = P( Own = 0,Woman = 0) = Φ ( β ' x, β ' x, ρ) ) d Φ (.,., ρ s the bvarate normal dstrbuton assumed for the perturbatons an Ths nce result of the bvarate probt model has already been used n emprcal work n varous areas. Greene (1998) studes the probablty of a gender economc courses at Lberal Arts Colleges, Whte and Wolaver (2003) focus on occupaton choce and mgraton and Greene, Rhne and Toussant-Comeau (2003) study the decson to patronze check-cashng busnesses and the decson to be unbanked. Fabbr, Monfardn and Radce (2004) focus on cesarean delvery utlzaton across publc and prvate hosptals. Ths last paper presents Monte Carlo evdence that testng the null of ρ = 0 s a vald test for exogenety. 3 A two-stage procedure parallelng 2SLS for lnear smultaneous equatons models wll yeld nconsstent results as dscussed n Wooldrdge (2003). 9

10 III.2 Bass statstcs Table 3 and 4 present descrptve statstcs of varables lkely to affect the probablty of becomng a homeowner and the probablty of a woman to head her own household. Some of the varables are for the household as a whole, some are characterstcs of the household head and some are characterstcs of the woman of the household. The frst two varables are dependent varables of our model at the household level. Own and Woman are dummy varables. Own takes the value of one when the household owns the house where they lve and 0 otherwse whle Woman takes the value 1 whether the household head s a woman and 0 otherwse. Smply lookng at the means, women represent a hgher percentage of renters than owners and users. In most cases but n Brazl, Ecuador, Venezuela and Ncaragua the percentage of owners s hgher n men headed households than n women headed households, at least n the frst two countres ths s probably due to our nablty to depurate the database from wdows. The varables of nterest can be classfed n the followng four categores: ncome, lfe-cycle status, locaton and neghborhood attrbutes and other socoeconomc characterstcs. We defne two ncome-related varables: total household ncome (IncomeHouse) and total ncome of the woman of the house 4 (IncomeWoman). There s not a clear pattern n current household ncome wth respect of owners and renters. In many countres the house ncome level s about the same for both groups. The household ncome of owners s hgher than renters n Colomba, Uruguay and Guatemala and s lower n Chle, Ecuador, Honduras, Ncaragua, Paraguay, Peru and Venezuela. The key dfferences n ncome are between owners and renters wth those usng wthout explct rghts over the house. The mean values of the IncomeWoman and IncomeHouse mply that the on average the ncome of the woman of the house accounts for approxmately 30% of total ncome. Venezuela s an exceptonal case where the mean value of IncomeWoman s 60% of the mean value on total household ncome. When breakng these averages by household head t transpres that when the household s headed by a man the share of women s ncome n total ncome s much lower (about 20%) and much hgher when s headed 4 She may be the household head or the household head s wfe. 10

11 by a woman. Women that potentally earn by themselves more money are lkely to feel more ndependent and therefore ths may affect the decson to reman marred or not. Ths s also clear from the comparson n absolute terms of IncomeWoman for those women that are household heads and those that are not. For most countres the average ncome of women headng ther household s more than double the ncome of women not headng ther household. The exceptons are Mexco, Paraguay and Venezuela were n any case the average ncome of women headng ther households s more than 40% the average ncome of women not dong so. We consdered three lfe-cycle status varables: age of the household head and age of the woman (AgeHead and AgeWoman), a dummy that takes the value of 1 f the household head s marred an 0 otherwse (Marred) 5 and the amount of chldren under 18 years old n the house (Chldren). In female headed famles AgeHead takes the same value of AgeWoman. In most Latn Amercan countres ownng a house s a famly achevement that can be attaned only after many years of efforts. Our tables show that owner household heads and the women of the house are about 10 years older than renters and users. 6 In couples men are usually older than women and on average our data mples a dfference between 2 and 5 years old (Argentna beng the mnmum and Ncaragua the maxmum). The age gap s larger for users followed by owners and renters. If a person does not beleve hs actual mate to be stable, t s natural that he may not be nterested n gettng nto a long-term contract as a housng mortgage credt or buyng a household that could be consdered a martal property n case of dvorce or separaton. He wll prefer a more flexble housng soluton lke rentng. The household head beng marred and the presence of chldren are proxes of famly stablty. The majorty of owners are marred (fgures gong up to 74% for Mexco and Bolva) whle only a mnorty of renters and users are (the share of marred household heads n renters and users s above 50% only for Chle, Mexco and Venezuela). Owners tend to have more chldren than renters and about the same amount than users. 7 Breakng the analyss n the household head gender dmenson, only a very small proporton (n most countres below 20%) of woman households heads are marred (Paraguay beng the country 5 Ths varable could not be defned for Ecuador and Brazl. 6 Haven taken wdows out of the sample ths age gap s lower than the age gap for the whole sample. 7 It may be surprsng that the average for Chldren s between 1 and 2 but t should be noted that ths s the average number of chldren per household and not per famly. 11

12 wth the hghest share of marred female household heads, 31%) and they tend to have less chldren than households were there s a couple present and the household head s a man. We defned two varables related to educaton level and both were dvded between the household head and the women of the house. SchoolngHead and SchoolngWoman are the years of formal educaton of the household head or of the woman of the house. 8 IllterateHead and IllterateWoman are dummy varables takng the value 1 f the household head or the woman of the house s llterate and 0 otherwse. On average, owners are older than renters are but renters are on average more educated than owners. Gven the mprovements n educaton levels over the last decades t s not surprsng that the younger groups are more educated that the older ones. Wth respect to locaton Cty s a dummy that takes value one f the house s located n a urban center and 0 otherwse and CaptalCty s a dummy that takes the value one f the house s located n the captal cty of the country. 9 Many tmes there are cultural dfferences between nhabtants of the urban centers and the rest of the country. Beng the latter more conservatve s reasonable to have a lower proporton or women headed n rural areas. 8 Argentna only reports schoolng levels and not actual years. We assume that those wth prmary ncomplete attend 3 years, those that dd not completed secondary school attended 8 years, those that dd not completed unversty studes had 13 formal years of educaton and fnally those wth unversty degrees were assgned 16 years of schoolng. 9 For Bolva CaptalCty takes the value 1 f the household s located ether n La Paz or Sucre. 12

13 Table 3. Summary Statstcs by housng tenure South Amerca Argentna Bolva Brazl Chle Colomba Own Rent Use Tot Own Rent Use Tot Own Rent Use Tot Own Rent Use Tot Own Rent Use Tot Own 100% 0% 0% 71% 100% 0% 0% 61% 100% 0% 0% 74% 100% 0% 0% 70% 100% 0% 0% 50% Woman 21% 29% 25% 23% 13% 23% 17% 15% 27% 30% 22% 27% 17% 20% 17% 17% 25% 29% 24% 26% IncomeHouse IncomeWoman AgeHead 50,6 38,1 41,6 47,4 46,2 35,5 37,0 42,4 48,4 39,5 41,3 46,3 51,2 39,9 42,9 48,3 50,9 38,9 40,1 45,1 Agewoman 48,3 36,9 39,1 45,4 43,6 33,6 34,5 40,1 45,4 36,9 37,9 43,4 48,1 37,2 39,6 45,4 47,5 36,2 36,5 42,0 Marred 64% 37% 45% 57% 74% 45% 54% 65% 70% 52% 57% 65% 50% 29% 30% 39% Chldren 1,3 1,0 1,5 1,3 2,3 1,9 2,0 2,1 1,2 1,1 1,4 1,2 1,3 1,3 1,4 1,3 1,3 1,3 1,6 1,4 SchoolngHead ,1 9,2 9,0 7,9 5,9 7,3 4,8 6,0 7,1 10,0 7,7 7,6 7,8 9,1 6,3 8,1 SchoolngWoman ,0 8,8 8,8 7,1 6,0 7,3 5,2 6,1 7,1 9,8 7,8 7,6 7,7 9,0 6,6 8,0 Cty 60% 87% 81% 69% 87% 98% 70% 87% 61% 88% 52% 62% Captalcty 6% 10% 5% 6% 22% 23% 27% 23% 54% 68% 32% 56% Ecuador Paraguay Peru Uruguay Venezuela Own Rent Use Tot Own Rent Use Tot Own Rent Use Tot Own Rent Use Tot Own Rent Use Tot Own 100% 0% 0% 68% 99% 0% 0% 76% 100% 0% 21% 72% 100% 0% 0% 66% 100% 0% 0% 83% Woman 21% 22% 19% 21% 22% 33% 20% 23% 11% 22% 18% 14% 21% 27% 23% 23% 40% 37% 30% 39% IncomeHouse IncomeWoman AgeHead 51,9 40,9 42,0 48,5 47,3 35,6 38,5 45,0 48,9 40,1 38,9 45,8 54,7 44,5 45,5 51,4 50,4 40,3 41,3 48,7 Agewoman 48,2 37,6 38,1 44,9 43,7 33,0 34,6 41,6 45,6 37,3 36,3 42,8 52,1 42,2 42,4 48,8 48,0 38,2 37,6 46,5 Marred 65% 37% 42% 59% 61% 39% 39% 54% 67% 45% 48% 60% 52% 55% 39% 51% Chldren 1,8 1,8 1,8 1,8 2,2 1,5 1,9 2,1 1,9 1,9 1,8 1,9 0,8 0,9 1,4 0,9 4,4 3,5 3,1 4,2 SchoolngHead 6,2 8,9 6,8 6,8 6,2 9,4 6,1 6,5 7,9 11,7 9,8 8,7 9,3 10,2 7,9 9,2 7,3 10,6 6,9 7,6 SchoolngWoman 6,1 8,7 6,8 6,6 6,1 8,9 5,8 6,3 6,3 11,0 8,6 7,3 10,1 11,1 9,3 10,1 7,7 10,7 8,3 8,0 Cty 49% 84% 47% 55% 50% 89% 43% 52% 67% 93% 85% 73% Captalcty 6% 22% 5% 7% 37% 51% 55% 42% 55% 67% 48% 56% Note: Own=1 f household owns the house, Wom=1 f household the head s female, IncomeHouse= total household ncome, IncomeWom= total ncome of the woman of the house, Age and Schoolng are evaluated for the household head and the woman of the house, Schoolng s years of educaton, Marred=1 f household head s marred, Chldren=amount of chldren under 18 n the house, Cty= 1 f the house s located an urban center and Captalcty=1 f t s located n the captal cty.

14 Table 3 (contnuaton). Summary Statstcs by housng tenure Mexco and Central Amerca Costa Rca El Salvador Guatemala Honduras Own Rent Use Tot Own Rent Use Tot Own Rent Use Tot Own Rent Use Tot Own 100% 0% 0% 75% 100% 0% 0% 69% 100% 0% 0% 61% 100% 0% 0% 72% Woman 19% 26% 18% 20% 24% 32% 25% 25% 14% 23% 13% 16% 19% 27% 21% 21% IncomeHouse IncomeWoman AgeHead 47,1 37,4 40,6 45,0 46,6 37,9 39,5 44,1 47,2 37,3 38,8 43,6 47,3 36,1 39,2 44,6 Agewoman 43,5 34,2 36,1 41,4 42,8 35,0 35,6 40,6 43,7 34,7 35,5 40,4 43,0 32,9 34,9 40,5 Marred 62% 36% 39% 56% 45% 28% 28% 40% 65% 41% 59% 59% 48% 29% 30% 43% Chldren 1,5 1,4 1,8 1,5 4,6 3,8 3,9 4,3 1,4 0,9 1,2 1,3 2,7 1,9 2,3 2,5 SchoolngHead 7,2 8,5 5,7 7,2 5,3 7,6 5,0 5,5 5,6 7,3 5,7 5,9 6,2 7,8 5,8 6,4 SchoolngWoman 7,4 8,1 5,8 7,3 4,9 7,1 4,8 5,1 4,4 6,3 4,6 4,8 6,2 7,6 5,9 6,4 Cty 59% 40% 75% 58% 65% 94% 67% 71% 47% 89% 39% 52% Captalcty 15% 25% 7% 15% Mexco Ncaragua Panama Own Rent Use Tot Own Rent Use Tot Own Rent Use Tot Own 100% 0% 0% 72% 100% 0% 0% 76% 100% 0% 0% 77% Woman 15% 23% 19% 17% 22% 22% 16% 21% 20% 26% 19% 20% IncomeHouse IncomeWoman AgeHead 47,7 36,9 39,2 45,0 46,8 39,8 37,7 44,7 48,2 39,6 40,2 46,3 Agewoman 44,6 34,3 36,7 42,1 42,4 34,2 33,6 40,4 44,4 36,8 35,7 42,7 Marred 74% 51% 59% 69% 46% 40% 33% 43% 34% 24% 21% 32% Chldren 1,7 1,4 1,6 1,6 5,8 4,3 4,7 5,5 1,8 1,2 1,9 1,7 SchoolngHead 7,6 9,8 8,4 8,1 4,4 8,9 4,5 4,5 7,5 10,3 7,5 7,8 SchoolngWoman 7,3 9,4 8,2 7,7 4,2 8,2 4,6 4,4 7,7 10,4 7,8 8,0 Cty 55% 98% 50% 55% 49% 85% 62% 54% Captalcty 13% 17% 11% 13% 8% 27% 10% 10% Note: Own=1 f household owns the house, Wom=1 f household the head s female, IncomeHouse= total household ncome, IncomeWom= total ncome of the woman of the house, Age and Schoolng are evaluated for the household head and the woman of the house, Schoolng s years of educaton, Marred=1 f household head s marred, Chldren=amount of chldren under 18 n the house, Cty= 1 f the house s located an urban center and Captalcty=1 f t s located n the captal cty. 14

15 Table 3. Summary Statstcs by household head South Amerca Argentna Bolva Brazl Chle Colomba Man Wom Tot Man Wom Tot Man Wom Tot Man Wom Tot Man Wom Tot Own 72% 64% 71% 63% 51% 61% 75% 75% 75% 70% 68% 70% 51% 48% 50% Wom 0% 100% 23% 0% 100% 15% 0% 100% 30% 0% 100% 17% 0% 100% 26% IncomeHouse IncomeWom AgeHead 47,8 45,6 47,3 42,6 41,3 42,4 44,3 50,6 46,2 48,3 48,7 48,3 45,2 44,8 45,1 AgeWom 45,3 45,6 45,3 39,8 41,3 40,1 40,3 50,6 43,4 44,6 48,7 45,4 40,9 44,8 42,0 Marred 70% 13% 57% 72% 26% 65% 75% 18% 65% 49% 12% 39% Chldren 1,3 1,1 1,3 2,2 1,5 2,1 1,4 1,0 1,3 1,3 1,1 1,3 1,4 1,2 1,4 SchoolngHead ,9 7,7 7,9 6,1 5,7 6,0 7,6 7,6 7,6 7,9 8,5 8,1 SchoolngWom ,0 7,7 7,1 6,2 5,7 6,0 7,5 7,6 7,6 7,8 8,5 8,0 Cty 67% 78% 69% 84% 93% 87% 60% 73% 62% Captalcty 6% 7% 6% 23% 27% 23% 54% 62% 56% Ecuador Paraguay Peru Uruguay Venezuela Man Wom Tot Man Wom Tot Man Wom Tot Man Wom Tot Man Wom Tot Own 68% 68% 68% 77% 74% 76% 74% 59% 72% 67% 62% 66% 81% 85% 83% Wom 0% 100% 21% 0% 100% 23% 0% 100% 14% 0% 100% 23% 0% 100% 39% IncomeHouse IncomeWom AgeHead 47,5 52,6 48,5 45,1 44,5 45,0 46,0 44,9 45,8 51,6 50,7 51,4 47,5 50,5 48,7 AgeWom 42,5 52,6 44,9 40,6 44,5 41,6 42,4 44,9 42,8 48,2 50,7 48,8 43,0 50,5 46,5 Marred 67% 31% 59% 62% 8% 54% 74% 12% 60% 76% 11% 51% Chldren 1,8 1,8 1,8 2,1 2,0 2,1 1,9 2,1 1,9 0,9 0,8 0,9 4,2 4,2 4,2 SchoolngHead 7,0 5,9 6,8 6,5 6,3 6,5 8,8 8,5 8,7 9,0 10,0 9,2 8,0 7,0 7,6 SchoolngWom 6,9 5,9 6,6 6,3 6,3 6,3 7,0 8,5 7,3 10,2 10,0 10,1 8,8 7,0 8,0 Cty 53% 62% 55% 49% 65% 52% 71% 86% 73% Captalcty 6% 10% 7% 41% 53% 42% 54% 63% 56% Note: Own=1 f household owns the house, Wom=1 f household the head s female, IncomeHouse= total household ncome, IncomeWom= total ncome of the woman of the house, Age and Schoolng are evaluated for the household head and the woman of the house, Schoolng s years of educaton, Marred=1 f household head s marred, Chldren=amount of chldren under 18 n the house, Cty= 1 f the house s located an urban center and Captalcty=1 f t s located n the captal cty. 15

16 Table 3 (contnuaton). Summary Statstcs by household head Mexco and Central Amerca Costa Rca El Salvador Guatemala Honduras Man Wom Tot Man Wom Tot Man Wom Tot Man Wom Tot Own 75% 72% 75% 70% 66% 69% 62% 55% 61% 72% 67% 71% Wom 0% 100% 20% 0% 100% 25% 0% 100% 16% 0% 100% 21% IncomeHouse IncomeWom AgeHead 44,9 44,6 44,9 44,0 44,5 44,1 43,7 43,3 43,6 44,2 45,1 44,4 AgeWom 40,4 44,5 41,3 39,1 44,5 40,6 39,8 43,3 40,4 39,0 45,1 40,4 Marred 67% 8% 56% 50% 8% 40% 67% 18% 59% 50% 13% 43% Chldren 1,5 1,5 1,5 4,5 4,0 4,3 1,3 1,2 1,3 2,6 2,2 2,5 SchoolngHead 7,1 7,3 7,2 5,7 4,9 5,5 6,0 5,5 5,9 6,3 7,0 6,4 SchoolngWom 7,3 7,3 7,3 5,2 4,9 5,1 4,7 5,5 4,8 6,2 7,0 6,4 Cty 61% 48% 58% 69% 82% 71% 48% 65% 52% Captalcty 14% 21% 15% Mexco Ncaragua Panama Man Wom Tot Man Wom Tot Man Wom Tot Own 72% 63% 70% 75% 81% 76% 78% 75% 77% Wom 0% 100% 17% 0% 100% 21% 0% 100% 20% IncomeHouse IncomeWom AgeHead 44,8 43,9 44,7 44,5 45,6 44,7 46,3 46,5 46,3 AgeWom 41,4 43,9 41,8 38,9 45,6 40,4 41,5 46,5 42,7 Marred 78% 22% 69% 52% 8% 43% 38% 7% 32% Chldren 1,7 1,4 1,6 5,6 5,2 5,5 1,8 1,6 1,7 SchoolngHead 8,0 8,3 8,1 4,4 5,0 4,5 7,5 8,7 7,8 SchoolngWom 7,6 8,3 7,7 4,3 5,0 4,4 7,8 8,7 8,0 Cty 50% 74% 55% 50% 70% 54% Captalcty 11% 18% 13% 8% 15% 10% Note: Own=1 f household owns the house, Wom=1 f household the head s female, IncomeHouse= total household ncome, IncomeWom= total ncome of the woman of the house, Age and Schoolng are evaluated for the household head and the woman of the house, Schoolng s years of educaton, Marred=1 f household head s marred, Chldren=amount of chldren under 18 n the house, Cty= 1 f the house s located an urban center and Captalcty=1 f t s located n the captal cty. 16

17 III.3 Results Table 5 and 6 present the results of the tradtonal probt estmaton for homeownershp and women household heads. Table 7 presents the estmaton of the bvarate probt model where we control for the endogenety of woman headshp. There are two dfferences n the ownershp regresson presented n Table 5 and n Table 7: the smultaneous estmatons n the case of the bvarate probt model and the number of observatons ncluded. Snce the bvarate probt model can be run only when there s nformaton for all varables n both equatons the number of observaton n Table 7 s lower than n Table 5 for all countres. In order to be sure that our results are not due to composton effects we also run the smple probt models restrctng the set of observatons to those consdered n Table 7. The results do not change qualtatvely. In tables A2, A3, and A4 n the appendx we report the same summary statstcs than n Tables 1, 2 and 3 but restrct to those observatons ncluded n the bvarate probt model. The pcture does not change ether. Thus, the changes n the estmatons from the smple probt model to bvarate probt must be due to the endogenety control. The man methodologcal result of ths secton can be seen by the reverse of the sgn of Women n the Homeownershp regressons for the cases of Argentna, Chle, Colomba, El Salvador, Mexco, Paraguay, Uruguay and Venezuela. Table 8 presents the margnal effect of Woman Headshp wth and wthout controllng for female headshp endogenety. Accordng to the smple probt models there s statstcally sgnfcant dscrmnaton n favor of women n Argentna, Chle, Colomba, Costa Rca, El Salvador, Ncaragua, Paraguay, Uruguay and Venezuela. The only country where there s

18 dscrmnaton aganst women at the tradtonal statstcal sgnfcance levels are Brazl, Ecuador and Peru. On the contrary the bvarate probt models shows evdence of dscrmnaton aganst women n all cases but n Costa Rca, Ncaragua and Venezuela. The probt model overestmates the margnal effect of women headshp on average by Consderng that around 72% of households own ther home an overestmaton n the margnal of ths magntude s really bg. The smple probt estmaton results suggest n nne countres a more favorably outcome n terms of homeownershp for female headed famles, a less favorably result for three countres and was nconclusve for fve countres. In the bvarate probt model, the three countres that were suggestng a worse condton for women and the fve countres that were prevously nconclusve show sgnfanct evdence of lower probablty of homeownershp for women. Of the nne countres whose probt model suggested that female headshp was assocated wth larger probablty of ownng ther home, after controllng for endogenety only three mantan ths concluson whle sx reverse sgn and suggest lower probablty for women headed famles. Therefore, n the bvarate probt model we recover the ntutve result that femaleheaded famles are not n a better stuaton than husband-wfe famles n what respect to homeownershp. The rest of the varables present for the most cases reasonable results. The hgher the ncome of the house the most lkely to become a homeowner n all cases but Bolva, Paraguay and Venezuela. In all cases we found that the hgher the ncome of the woman of the house the more lkely to head her own household, n lne wth the endogenety expected. 18

19 The lfe cycle varables also have the expected sgns for the most. The older the household head the more lkely to own hs house. Famly stablty s also assocated wth less flexble housng tenure optons as ownershp. The fact of beng marred sgnfcantly ncreases the probablty of becomng a homeowner n the probt model wth a margnal effect around The change n the sgn n the Marred varable n the bprobt model s probably produced by the same endogenety that changes the sgn of the woman varable. The number of chldren also s postvely related wth the probablty of ownng the house n sxteen out of the seventeen countres, the exempton beng Uruguay a country wth the lowest fertlty rates n Latn Amerca. Older women are more lkely to become household heads. Recallng that we have excluded wdows form the household level regressons, ths result s not so obvous and may suggest that even for those female household heads ther frst opton was a more tradtonal both parents famly, and after beng unsuccessful for some tme ther choose to head ther own famly. Wth respect to educaton we found more counterntutve results. At least n one of the estmaton methods the schoolng varable reflects that more educaton s assocated wth hgher probablty of beng a homeowner n Colomba, Uruguay, Argentna and Costa Rca. On the contrary n Bolva, Brazl, Chle, Ecuador, Peru and Venezuela more educaton s assocated wth lower probablty of becomng homeowners. In part ths result may be produced by an ncrease n the level of educaton of young cohorts that have lower probablty of ownng ther home as reflected n the age varable. We conjectured that more educated women have more labor opportuntes and therefore may fell less attached to an unsatsfactory marrage. Ths seems to be verfed for Argentna, Ecuador, Venezuela, Mexco and Ncaragua but not for Bolva, Brazl, Chle, Ecuador, Paraguay, Peru and Mexco. 19

20 Table 5 Determnants of the probablty of homeownershp (wthout controllng for endogenety) Argent Bolva Brazl Chle Colomb Ecuador Parag Peru Urug Vene Mexco Costa El Guatem Hond Ncarag Panama Rca Salvador Woman [0.022]*** [0.060]* [0.010]*** [0.017]*** [0.023]*** [0.025]*** [0.038]** [0.132]*** [0.030]*** [0.018]*** [0.028] [0.040]*** [0.027]* [0.076] [0.046] [0.064]*** [0.049] IncomeHouse [0.004]** [0.019]*** [0.004]*** [0.006]*** [0.005]*** [0.008]*** [0.005]*** [0.018] [0.016]*** [0.002]*** [0.010]*** [0.008]*** [0.005]*** [0.024] [0.016] [0.013] [0.013] AgeHead [0.001]*** [0.002]*** [0.000]*** [0.000]*** [0.001]*** [0.001]*** [0.001]*** [0.004]*** [0.001]*** [0.001]*** [0.001]*** [0.001]*** [0.001]*** [0.002]*** [0.001]*** [0.002]*** [0.001]* ** Marred [0.021]*** [0.047]*** [0.013]*** [0.021]*** [0.034]*** [0.098]*** [0.027]*** [0.017]*** [0.023]*** [0.034]*** [0.026]*** [0.058]*** [0.040]*** [0.052]*** [0.046]* ** Chldren [0.007]*** [0.012]*** [0.003]*** [0.005]*** [0.007]*** [0.006] [0.009]*** [0.026] [0.010] [0.004]*** [0.007]*** [0.011]*** [0.006]*** [0.020]*** [0.010]*** [0.010]*** [0.011]* ** SchoolngHead [0.003] [0.005]*** [0.001] [0.002]*** [0.002]** [0.002]*** [0.004] [0.010]*** [0.003]*** [0.002]*** [0.002]*** [0.004]*** [0.003] [0.006] [0.005]* [0.006] [0.005]* ** Constant [0.053]*** [0.135]*** [0.025]*** [0.076]*** [0.070]*** [0.058]*** [0.086]*** [0.196] [0.140]*** [0.040]*** [0.090]*** [0.106]*** [0.057]*** [0.184]*** [0.138]*** [0.122]*** [0.108]* * Observatons Note: Dependent varable: Own=1 f household owns the house, Woman=1 f household the head s female, IncomeHouse= total household ncome, AgeHead s the age of the household head and SchoolngHead s years of educaton of the household head, Marred=1 f the household head s marred, Chldren=amount of chldren under 18 n the house. Standard errors n brackets * sgnfcant at 10%; ** sgnfcant at 5%; *** sgnfcant at 1%

21 Table 5 Determnants of the probablty of female headshp Argent Bolva Brazl Chle Colomb Ecuador Parag Peru Urug Vene Mexco Costa El Guatem Hond Ncarag Panama Rca Salvador Own [0.029]*** [0.066]** [0.011]*** [0.018] [0.027]*** [0.027]*** [0.044]* [0.142]** [0.035] [0.021] [0.036]*** [0.046]* [0.031]** [0.085] [0.053]*** [0.071]* [0.056]** IncomeWoman [0.005]*** [0.013]*** [0.002]*** [0.002]*** [0.002]*** [0.006]*** [0.003]*** [0.026]*** [0.006]*** [0.001]*** [0.011]*** [0.004]*** [0.004]*** [0.016]*** [0.007]*** [0.009]*** [0.010]** * AgeWoman [0.001]*** [0.003]*** [0.000]*** [0.001]*** [0.001]*** [0.001]*** [0.001]*** [0.005]*** [0.001]*** [0.001]*** [0.001]*** [0.002]*** [0.001]*** [0.003]*** [0.002]*** [0.002]*** [0.002]** * Marred [0.028]*** [0.065]*** [0.016]*** [0.029]*** [0.037]*** [0.160]*** [0.034]*** [0.018]*** [0.032]*** [0.047]*** [0.034]*** [0.085]*** [0.054]*** [0.073]*** [0.063]** * Chldren [0.009]* [0.019]*** [0.004]*** [0.007] [0.010]*** [0.007] [0.010] [0.036] [0.014]* [0.004]*** [0.012] [0.014] [0.007]*** [0.030] [0.013] [0.011]** [0.013] SchoolngWoman [0.004] [0.007] [0.001]*** [0.002]** [0.003]*** [0.003]*** [0.005]*** [0.015] [0.004]*** [0.002]*** [0.004]*** [0.005] [0.003] [0.010] [0.007]*** [0.008]*** [0.006]* Cty [0.016]*** [0.018]*** [0.027]*** [0.039]*** [0.158]** [0.041] [0.042]*** [0.102] [0.065]*** [0.053]** * Captalcty [0.053]*** [0.097] [0.026]** [0.033]*** [0.061] Constant [0.070]*** [0.140]*** [0.026]*** [0.044]*** [0.064]*** [0.056]*** [0.084]*** [0.280]*** [0.091]*** [0.041]*** [0.108]*** [0.105]*** [0.064]*** [0.176]*** [0.111]*** [0.134]*** [0.120]** * Observatons Note: Dependent varable: Woman=1 f household the head s female, Own=1 f household owns the house. IncomeWoman, AgeWoman and SchoolngWoman are total ncome, age and years of formal educaton of the woman of the house, Marred=1 f the household head s marred, Chldren=amount of chldren under 18 n the house, Cty= 1 f the house s located an urban center and Captalcty=1 f t s located n the captal cty. Standard errors n brackets * sgnfcant at 10%; ** sgnfcant at 5%; *** sgnfcant at 1% 21

22 Table 7 Determnants of the probablty of homeownershp and woman household headshp Argent Bolva Brazl Chle Colomb Ecuador Parag Peru Urug Vene Mexco Costa Rca El Salvador Guatem Hond Ncarag Panama Home ownershp Woman [0.015]*** [0.062]*** [0.007]*** [0.012]*** [0.021]*** [0.017]*** [0.031]*** [0.089]*** [0.027]*** [0.014]*** [0.021]*** [0.038]*** [0.026]*** [0.052]*** [0.034]*** [0.051]*** [0.034]*** IncomeHouse [0.002]*** [0.015]*** [0.002]*** [0.006]*** [0.004]*** [0.005]*** [0.004]*** [0.003]*** [0.014]*** [0.001]*** [0.004]*** [0.004]*** [0.005]*** [0.024] [0.014]*** [0.003]*** [0.006]*** AgeHead [0.001]*** [0.002]*** [0.000]*** [0.001]*** [0.001]*** [0.001]*** [0.001]*** [0.004]*** [0.001]*** [0.001]*** [0.001]*** [0.001]*** [0.001]*** [0.002]*** [0.001]*** [0.001]*** [0.001]*** Marred [0.019]*** [0.049] [0.015]*** [0.024]*** [0.036]*** [0.097] [0.030]*** [0.015]*** [0.028]*** [0.037]*** [0.027]*** [0.058]** [0.041]*** [0.048]*** [0.044]*** Chldren [0.006]*** [0.012]* [0.003]** [0.005]*** [0.007] [0.005] [0.008]*** [0.024] [0.009]*** [0.004]*** [0.008]*** [0.010]*** [0.006]*** [0.020]*** [0.011]*** [0.009]*** [0.010] SchoolngHead [0.002]*** [0.002]*** [0.001]*** [0.002]*** [0.002]*** [0.002]*** [0.004] [0.009]*** [0.003]*** [0.002]*** [0.002] [0.004] [0.003] [0.006] [0.005] [0.002]*** [0.004] Constant [0.049]*** [0.138]*** [0.022]*** [0.071]*** [0.067]*** [0.040]*** [0.082] [0.199] [0.126]*** [0.045]*** [0.067]*** [0.090]*** [0.061]*** [0.190]*** [0.137]*** [0.074]*** [0.093]*** Female headshp Own [0.016]*** [0.047]*** [0.008]*** [0.016]*** [0.019]*** [0.016]*** [0.034]*** [0.087]*** [0.031]*** [0.015]*** [0.024]*** [0.039]*** [0.030]*** [0.053]*** [0.039]*** [0.051]*** [0.034]*** IncomeWoman [0.004]*** [0.011]*** [0.002]*** [0.001]*** [0.002]*** [0.005]*** [0.002]*** [0.019]*** [0.005]*** [0.001]*** [0.006]*** [0.003]*** [0.004]*** [0.014]*** [0.006]*** [0.006]*** [0.009]*** AgeWoman [0.001]*** [0.002]*** [0.000]*** [0.001]*** [0.001]*** [0.001]*** [0.001]*** [0.005]*** [0.001]*** [0.001]*** [0.001]*** [0.002] [0.001]*** [0.003]*** [0.001]*** [0.001]*** [0.001]*** Marred [0.024]*** [0.055]*** [0.015]*** [0.024]*** [0.033]*** [0.148]*** [0.031]*** [0.016]*** [0.030]*** [0.037]*** [0.030]*** [0.076]*** [0.047]*** [0.052]*** [0.055]*** Chldren [0.006]*** [0.013]*** [0.003]** [0.006]*** [0.007] [0.005] [0.008]*** [0.024] [0.011]** [0.003]*** [0.008]*** [0.011]* [0.007] [0.026] [0.011]*** [0.009]*** [0.010] SchoolngWoman [0.002]* [0.004]* [0.001]*** [0.001]*** [0.002]*** [0.002]*** [0.003]* [0.008]*** [0.003]*** [0.001]*** [0.003]* [0.004] [0.003] [0.006] [0.005] [0.004]*** [0.004] Cty [0.016]*** [0.013]*** [0.004]*** [0.142]** [0.004]*** [0.038]* [0.092] [0.046]*** Captalcty [0.004]*** [0.054]** [0.009]* [0.008]*** [0.020]* [0.049] [0.022] Constant [0.044]*** [0.111]*** [0.025]*** [0.032]*** [0.051]*** [0.040]*** [0.073]*** [0.243]** [0.064]*** [0.038]*** [0.066]*** [0.099]*** [0.061]*** [0.170]*** [0.095]*** [0.096]*** [0.089]*** rho [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] Observatons Note: Dependent varables: Own=1 f household owns the house and Woman=1 f household the head s female. IncomeHouse= total household ncome, AgeHead and SchoolngHead are age and years of formal educaton of the household head IncomeWoman, AgeWoman and SchoolngWoman are total ncome, age and years of formal educaton of the woman of the house, Marred=1 f the household head s marred, Chldren=amount of chldren under 18 n the house, Cty= 1 f the house s located an urban center and Captalcty=1 f t s located n the captal cty. Standard errors n brackets, * sgnfcant at 10%; ** sgnfcant at 5%; *** sgnfcant at 1%

23 Table 7 Margnal Effects of Woman headshp over the probablty of beng a homeowner Probt Bprobt Overestmaton Argentna ** *** 0,4912 Bolva *** 0,4677 Brazl *** *** 0,5087 Chle *** *** 0,5496 Colomba *** *** 0,4656 Costa Rca *** *** -0,2483 Ecuador *** *** 0,5062 El Salvador * ** 0,0352 Guatemala *** 0,4754 Honduras *** 0,4636 Mexco *** 0,5504 Ncaragua *** *** -0,2643 Panama *** 0,5268 Paraguay * *** 0,4987 Peru ** *** 0,4724 Uruguay *** *** 0,546 Venezuela *** *** -0,2973 * sgnfcant at 10%; ** sgnfcant at 5%; *** sgnfcant at 1% IV. Conclusons Although there s a large lterature on the determnants of housng tenure and although there s also a large lterature on women dscrmnaton there are no studes that pont that women -all the rest equal- have lower probabltes of ownng ther house. We argue that the housng tenure decson and the housng headshp decsons should not be treated as exogenous. Among the varables that enter the decson of a woman to dvorce her husband are ncome related ssues and famly lfe cycle dmensons that also affect the probablty of ownng ther house. If ths type of endogenety s not properly accounted, t leads to nconsstent and often counterntutve results. In ths paper, we use ndvdual level data on homeownershp from Chle, Honduras and Ncaragua and verfy the potental problems wth household level estmatons that do not