# RECORD 1b4b730b1b7e5c68c755a7ae46919b1120fc8c06 # TITLE: Housing Markets in Uruguay: Determinants of Housing Demand and Its Interaction with Public Policies # YEAR: 2011.0 VENUE: nan DOI: 10.18235/0011348 # AUTHORS: Natalia Garabato; Magdalena Ramada-Sarasola # PAGES: 70 CANDIDATE TABLE PAGES (1-based): [15, 21, 24, 25, 27, 28, 33, 34, 36, 37, 38, 40, 58, 64, 65, 66, 67, 68, 69, 70] # ABSTRACT: This paper analyzes the determinants of housing demand for Uruguay and the extent to which housing policies have an impact on their target population. The paper first analyzes the determinants of housing demand, following an approach based on Rosen's (1974) two-step procedure consisting of fitting a hedonic price regression in 34 different geographical units (or markets) to estimate a housing demand function. The determinants of formality and ownership choices were examined using a multinomial logit framework. Determinants of these choices include both household demographic attributes and access to and use of public housing programs and other social programs. Policy recommendations are offered on the basis of the finding that a price and income-inelastic formal housing market greatly contrasts with a rather price and income-elastic informal housing market. ===== PAGE 1 (score 0, numeric-paren 0, table-words 0, var-words 0) ===== Housing Markets in Uruguay: Determinants of Housing Demand and its Interaction with Public Policies Natalia Garabato Magdalena Ramada-Sarasola Department of Research and Chief Economist IDB-WP-258 IDB WORKING PAPER SERIES No. Inter-American Development Bank October 2011 ===== PAGE 15 (score 31, numeric-paren 18, table-words 3, var-words 1) ===== 12 Table 1. Determinants of Income: Box-Cox Regression Coef. P> ) ( 2 df χ Elasticity Transformed Age 279.851*** 0.000 0.228 Not Transformed Male 0.001 0.002 0.001 Caucasian 0.127*** 26.103 0.078 Primary 0.495*** 459.675 0.304 Secondary 1.084*** 1197.778 0.667 Technical 1.303*** 925.828 0.801 Tertiary 2.162*** 1750.546 1.329 Postgraduate 2.939*** 546.598 1.806 Inc. Retiree -0.688*** 305.375 -0.423 Inc. Informal -1.304*** 1109.138 -0.802 Inc. Private -0.329*** 72.744 -0.202 Inc. Public -0.257*** 38.080 -0.158 Inc. Other -0.996*** 112.402 -0.612 MVD 0.391*** 483.331 0.241 With Partner 0.815*** 848.897 0.501 Divorced/Widower 0.196*** 44.223 0.120 Constant -157.161 Coef. P>|z| λ -1.66*** 0.000 θ 0.053*** 0.000 N 17,030 LR 2 χ 8,259.95 P> 2 χ 0.000 Test HO: P> 2 χ θ =λ =-1 0.000 θ =λ =0 0.000 θ =λ =1 0.000 *p<0.1, **p<0.05, ***p<0.001 As can be seen in Table 1, likelihood-ratio tests reject the linear, log-linear and inverse- multiplicative specifications at the 1 percent significance level, being θ and λ 0.053 and -1.66, respectively, and hence supporting the choice of a Box-Cox specification. Given the obtained coefficient for λ income relates to age in an inverse U-shape, as expected. Transformed age has a positive elasticity of 0.228, a result that is slightly lower than the one obtained by Fontenla et al. (2009) for Mexico. Further, we obtain a positive effect on the household’s total disposable ===== PAGE 16 (score 4, numeric-paren 0, table-words 0, var-words 4) ===== 13 income when the head of household head is male and Caucasian. The percentage changes on a household’s total income for a male head of household (as opposed to a female one) and from having a Caucasian head of household (as opposed to another race) are 0.1 percent and 7.8 percent, respectively. The effect stemming from gender is so little since gender refers to the head of household, while income includes the whole household’s income, i.e. the income received by each household member, plus the entire household’s income from other sources such as remittances. Further, we obtain increasing elasticities for each consecutive completed education level, with an impact of almost tripling income for finished postgraduates. The impact of completing primary education is 30.4 percent, and a secondary degree generates an income premium of 66.7 percent with respect to households in which heads have not completed primary education. The results for variables capturing the type of source of income are also aligned with economic intuition. An informal source of income has a negative impact of income when compared to owning a business (omitted case), while public and private salaried are those displaying the smallest difference in permanent income when compared to business owners. Owning a business as the main source of income earns a premium of over 80 percent over having an informal source of income, while households headed by retirees and pensioners have 42 percent less total disposable income. As expected, cities in the upcountry display smaller permanent income levels than Montevideo. The income premium of living in the capital is 24 percent. Finally, biparental households have an income premium of almost 50.1 percent, while monoparental households with a divorced or widowed household head earn 12 percent more income than those led by singles not engaged in a stable life-partnership. The highly significant results obtained, as well as their correctly predicted direction, support the approach of estimating permanent income through the above regression, while temporary income is defined as the residual. The estimated permanent income, noted as y , is used in the next estimation stages. 3.2 Estimation of Hedonic Prices For all 34 geographic units defined above, we estimate hedonic price regressions testing among different specifications of the Box-Cox transformation. In all regressions the household head’s ===== PAGE 21 (score 42, numeric-paren 27, table-words 1, var-words 3) ===== 18 between the formal and the informal market can be explained by the fact that a higher price elasticity is to be expected for low-income households, and it also mirrors the higher degree of mobility in informal markets. In these markets, supply seems to match better the quantity of demand since it is more flexible as it is not regulated. Given Malpezzi (1999) and Zabel’s (2004) results on the high degree of variation of price elasticities and its sensitivity to model design, our results are in line with expectations, and it can be argued that the informal market behaves as a more homogeneous market in the sense that households in informal markets are more prone to relocate to other informal housing arrangements irrespective of neighborhood or area. In the formal market, where there is less supply than demand and where less housing substitutes are available, demand becomes price inelastic. Table 3. Housing Demand: Box-Cox Regression Total sample Formal Informal Coef. Elasticity Coef. Elasticity Coef. Elasticity Not Transformed Temporary Income 0.000*** 0.032 0.000*** 0.038 0.000*** -0.023 Male -0.209*** -0.087 -0.155*** -0.063 -0.364** -0.132 With Partner 0.019 0.008 0.011 0.004 -0.102 -0.037 Children<14 -0.046*** -0.013 -0.006 -0.001 -0.119*** -0.060 # adults in HH 0.007 0.007 0.014 0.013 0.062 0.057 Education HH head 0.125*** 0.113 0.137*** 0.123 -0.028 -0.018 Caucasian 0.255*** 0.106 0.243*** 0.098 0.067 0.024 Use Housing Policies -0.011 -0.004 -0.117*** -0.047 - - Financial Access 0.243*** 0.101 0.179*** 0.072 -0.283 -0.102 Social assistance -0.355*** -0.147 -0.346*** -0.140 -0.109 -0.039 Const. 3.472 3.92 7.299 Transformed Price Index -0.368*** -0.372 -0.368*** -0.368 -0.728*** -0.792 Permanent Income 0.106*** 0.373 0.091*** 0.325 0.105*** 0.570 Age 0.194*** 0.200 0.2*** 0.205 -0.12** -0.134 Coef. P>|z| Coef. P>|z| Coef. P>|z| λ 0.228 0.000 0.231 0.000 0.297 0.000 σ 1.183 0.000 1.178 0.000 1.417 0.000 N 16,939 13139 1054 LR 2 χ 5375.7 3661.77 359.33 P> 2 χ 0.000 0.000 0.000 Test HO: P> 2 χ P> 2 χ P> 2 χ λ =1 0.000 0.000 0.000 λ =0 0.000 0.000 0.000 λ =1 0.000 0.000 0.000 *p<0.1, **p<0.05, ***p<0.001 ===== PAGE 22 (score 5, numeric-paren 0, table-words 0, var-words 5) ===== 19 For the overall market, income elasticities obtained for temporary and permanent income are in line with those found in the literature. In fact, while temporary income is found to have an elasticity of 0.032, permanent income shows an elasticity of 0.373, which, although being smaller than the usually expected 0.6 to 0.8 income elasticities, is at least above the minimum income elasticities reported in literature for developed countries, which are around 0.36 (Mayo, 1981). Here again, the informal market yields an income elasticity that is higher and closer to what is expected—0.57—pointing again to the fact that the informal market seems to be working as a more consolidated housing market. These results seem to be robust to changes in the definition of neighborhoods and proper housing markets6 In addition, our results seems to be consistent with the findings from Garabato and Ramada-Sarasola (2011) on the types of qualitative deficits faced by households in Uruguay and the lack of use of housing programs by the eligible population. In this sense, Uruguayan households seem to display a certain inertia with respect to improving their housing conditions, and this may be also reflected in the low elasticity found with respect to income. Given that the regression displays a static relationship between income and housing, this may be also the result of neglecting a time lag or taking into account income evolution, rather than income levels. , as well as to different ways of segmenting housing demand according to income strata or geographic areas. When looking at household composition in terms of its members, we find that the larger the number of children, the less housing is demanded—and less income is available to demand housing given the costs children entail. This result seems to be highly driven by the informal market, where the elasticity is more than double the one found in the whole sample. In the formal market, no significant effect is found. In addition, the usual negative impact on housing demand of being married, reported in the literature, is being captured by the presence of the variable accounting for the number of children. Once that has variable been controlled for, as well as total income (which implicitly includes the number of earning adults in the household), neither the partnership variable nor the number of adults in the household yields significant results. In the overall sample, male heads of household demand 8.7 percent less housing than their female peers, Caucasian heads of household demand 13 percent more housing, and bi- parental households demand 10.6 percent more housing than their single counterparts. Further, we find that age has a positive impact on housing demand, with an elasticity of almost 0.20, both 6 Results are available upon request. ===== PAGE 24 (score 28, numeric-paren 18, table-words 1, var-words 2) ===== 21 housing demand. None of these effects can be observed among informal housing markets. The fact that the use of housing programs has a significantly negative impact on the amount of housing demanded, even after controlling for the effect stemming from the household's socioeconomic attributes, may imply that housing programs grant access to housing solutions with fewer attributes than those that would have been demanded if the housing unit had been acquired by other means, all other things being equal. 3.3.2 Housing Demand in Montevideo Given the low price elasticities found for housing demand when taking into account all Uruguayan cities with more than 5,000 inhabitants, the question arises whether the result is driven by low mobility from one urban area to the other across the country. We looked at the results of a much more homogeneous market, in which analogous housing units should behave as substitutes irrespective of them being in different neighborhoods. Table 4. Housing Demand: Box-Cox Regression for Montevideo Montevideo Montevideo Formal Montevideo Informal Coef. Elasticity Coef. Elasticity Coef. Elasticity Not Transformed Temporary Income 0.000*** 0.030 0.000*** 0.038 0.000*** -0.033 Male -0.137** -0.037 -0.038 -0.011 -0.31* -0.094 With Partner -0.035 -0.010 0.006 0.002 -0.043 -0.013 Children<14 -0.049** -0.008 0.031 0.004 -0.129** -0.053 # people in household 0.057** 0.036 0.075*** 0.053 0.092 0.073 Education HH head 0.092** 0.059 0.124** 0.092 -0.011 -0.006 Race White 0.328*** 0.089 0.330*** 0.099 -0.073 -0.022 Use Housing Policies -0.034 -0.009 -0.189*** -0.057 Financial Access 0.340*** 0.092 0.187** 0.056 -0.41 -0.125 Social assistance -0.432*** -0.117 -0.359*** -0.108 0.034 0.010 Const. 5.078 5.421 9.814 ===== PAGE 25 (score 17, numeric-paren 10, table-words 1, var-words 1) ===== 22 Table 4., continued Montevideo Montevideo Formal Montevideo Informal Coef. Elasticity Coef. Elasticity Coef. Elasticity Transformed Price Index -0.291*** -0.319 -0.322*** -0.350 -0.843*** -0.897 Permanent Income 0.047*** 0.321 0.042*** 0.239 0.072*** 0.498 Age 0.200*** 0.208 0.217*** 0.223 -0.139** -0.153 Coef. P>|z| Coef. P>|z| Coef. P>|z| λ 0.337 0.000 0.305 0.000 0.340 0.000 σ 1.638 1.399 1.642 N 7,129 5,410 775 LR 2 χ 1,980.21 1,423.59 251.05 P> 2 χ 0.000 0.000 0.000 Test HO: P> 2 χ P> 2 χ P> 2 χ λ =1 0.000 0.000 0.000 λ =0 0.000 0.000 0.000 λ =-1 0.000 0.000 0.000 *p<0.1, **p<0.05, ***p<0.001 Although the elasticities for the overall market and for the formal market seem to decrease, i.e., become more inelastic, the difference between them is not significant. In the case of the informal market, where mobility across the country is even more difficult than for the formal market, the change goes in the expected direction, with elasticities becoming significantly larger in absolute terms. In other words, within the informal market results taking into account the whole country were downward, biasing price elasticity. For the formal Montevideo housing market, no difference can be found and price elasticity remains low. Still another difference is the effect of male heads of household demanding less housing, which becomes less pronounced in Montevideo. 3.3.3 Housing Demand by Owners and Tenants Arguably, income elasticity should be greater for tenants, who can adapt their housing solutions to changes in income faster and face fewer transaction costs than homeowners. In fact, in Uruguayan homeowners incur transaction costs both for selling and for buying a new housing unit, among them income taxes, notary costs, and real estate brokerage fees. Coordinating the timing of selling and buying simultaneously to transfer the value of one housing unit to the next ===== PAGE 26 (score 11, numeric-paren 0, table-words 2, var-words 7) ===== 23 also should increase the difficulty faced by owners in adapting their housing solution to both their needs and changes in income. On the other hand, owners could refurbish and renovate their housing units as a reaction to an increase in income. This could offset the effect of transaction costs for improving their housing conditions in the case of owners, provided they have access to credit. If there is no working lending market, it may be difficult for owners to borrow a large amount of money to adapt their housing unit, despite their current income having increased. This reasoning warrants estimating the demand separately for owners and tenants to see whether and how their price and income elasticities differ. As can be seen in Table 5, the housing demand of tenants is more elastic to changes in housing price, which supports the inference that tenants have more ability to react to changes in prices given they are the only ones who actually act as real economic agents in the market. While for owners there is a so-called imputed rent, if prices go up, the odds of an owner moving due to the movement in prices are rather small. In addition, our results support the inference that tenants are more income elastic than homeowners. Both elasticities are around 0.3, i.e., aligned with the previous results presented above. The difference, although not large, is significant. Given that the elasticities are similarly, one could also conclude that for owners, the lack of a well-functioning mortgage market may be keeping the income elasticity low. For tenants—where this situation does not apply—restrictions on the supply side, inertia, and other types of transaction costs, such as finding proper rental guarantees, may explain the low elasticity. Further differences become apparent between owners and tenants. First, while for owners neither the fact of having a partner, nor the number of grown-ups in the households affects housing demand, it does so for tenants. According to our results, couples that rent demand less housing, other things being equal, than single people. Given that the variable on the number of children is non-significant,8 8 This result is probably related to the high correlation between the two variables and a smaller number of observations in the subsample for tenants. this may be capturing the effect of couples demanding less housing given that they have to spend on other items, such as schooling and health care for their children. The interesting fact is that this is not observable for owners, who do not appear to reduce their demand for housing in order to account for family-related expenses when the housing demand is felt to be related to a more permanent type of housing, i.e., when buying. Older heads of ===== PAGE 27 (score 31, numeric-paren 21, table-words 1, var-words 1) ===== 24 household also tend to demand more housing when buying than when renting, displaying an elasticity to age that almost doubles the one for tenants. Table 5. Housing Demand: Box-Cox Regression for Homeowners and Tenants Homeowners Tenants Coef. Elasticity Coef. Elasticity Not Transformed Temporary Income 0.000*** 0.037 0.000*** 0.038 Male -0.164*** -0.064 -0.076 -0.034 With Partner 0.02 0.008 -0.136** -0.061 Children<14 -0.009 -0.002 -0.004 -0.001 # people in household -0.014 -0.013 0.106*** 0.101 Education HH head 0.143*** 0.121 0.108** 0.115 Race White 0.249*** 0.097 0.197*** 0.088 Use Housing Policies -0.209*** -0.081 0.212*** 0.095 Financial Access 0.097** 0.038 Social assistance -0.381*** -0.148 -0.276*** -0.123 Const. 4.434 4.103 Transformed Price Index -0.351*** -0.346 -0.384*** -0.410 Permanent Income 0.087*** 0.318 0.101*** 0.345 Age 0.151*** 0.154 0.077** 0.080 Coef. P>|z| Coef. P>|z| λ 0.238 0.000 0.216 0.000 σ 1.240 0.93835 N 10,714 2,425 LR 2 χ 2714.87 895.05 P> 2 χ 0.000 0.000 Test HO: P> 2 χ P> 2 χ λ =1 0.000 0.000 λ =0 0.000 0.000 λ =-1 0.000 0.000 *p<0.1, **p<0.05, ***p<0.001 Note: Only the formal housing market was considered for the regression above. Finally, while, ceteris paribus, access to housing through housing programs reduces the amount of housing demanded for households becoming owners of a housing unit, the result is also significant, in the other direction, for tenants. In other words, tenants that make use of the rental guarantee fund are able to demand more housing. ===== PAGE 28 (score 54, numeric-paren 37, table-words 1, var-words 1) ===== 25 3.3.4 Housing Demand by Age Cohorts Segmented housing demand can further analyzed. Given the high degree of inertia present in Uruguayan households, which appears to be preventing them from adjusting their housing needs to changes in income, coupled with the aging of the Uruguayan population, it seems natural to ask if inertia may be related to heads of household overconsuming housing at older ages and not changing their housing solution after a certain age—say, 50—when they reach the maximum amount of housing they can demand. Table 6. Housing Demand: Box-Cox Regression by Age Cohorts Age <=30 Age 31-50 Age 51-65 Age 66+ Coef. Elasticity Coef. Elasticity Coef. Elasticity Coef. Elasticity Not Transformed Temporary Income 0.000*** 0.029 0.000*** 0.038 0*** 0.045 0*** 0.033 Male -0.144** -0.094 -0.041 -0.026 -0.325*** -0.099 -0.225** -0.074 With Partner -0.179** -0.117 -0.038 -0.024 0.113 0.034 0.103 0.034 Children<14 -0.021 -0.013 0.002 0.001 -0.121** -0.012 0.012 0.000 # people in household 0.182*** 0.228 0.017 0.028 0.01 0.008 -0.006 -0.004 Education HH head 0.077* 0.128 0.066** 0.107 0.204*** 0.140 0.188** 0.112 Race White 0.099 0.065 0.163*** 0.104 0.297*** 0.091 0.312*** 0.103 Use Housing Policies 0.07 0.046 -0.106*** -0.068 -0.18** -0.055 -0.113* -0.037 Financial Access 0.288** 0.188 0.139** 0.089 0.156 0.048 0.162* 0.053 Social assistance -0.132** -0.086 -0.213*** -0.136 -0.421*** -0.128 -0.74*** -0.243 Const. 2.553 2.846 5.429 4.863 Transformed Price Index -0.32*** -0.333 -0.369*** -0.370 -0.397*** -0.390 -0.364*** -0.363 Permanent Income 0.112*** 0.215 0.183*** 0.349 0.063*** 0.325 0.068*** 0.298 Age 0.339** 0.325 0.111* 0.109 0.155 0.160 0.17** 0.189 Coef. P>|z| Coef. P>|z| Coef. P>|z| Coef. P>|z| λ 0.118 0.000 0.114 0.000 0.299 0.000 0.281 0.000 σ 0.662 0.725 1.582 1.437 N 873 4,306 3,829 4,131 LR 2 χ 194.660 1113.880 1113.880 885.280 P> 2 χ 0.000 0.000 0.000 0.000 Test HO: P> 2 χ P> 2 χ P> 2 χ P> 2 χ λ =1 0.000 0.000 0.000 0.000 λ =0 0.004 0.000 0.000 0.000 λ =-1 0.000 0.000 0.000 0.000 *p<0.1, **p<0.05, ***p<0.001 We estimated the housing demand regression for four age cohorts that should proxy four different stages of a household. In the first stage—heads under 30 years of age—households are new and have to face several sources of costs, e.g., access to durable goods, which may reduce ===== PAGE 29 (score 2, numeric-paren 0, table-words 0, var-words 2) ===== 26 their income elasticity. In the second stage—heads of household between 30 and 50 years of age—households are stabilizing their housing situation and accessing something permanent. It is also the stage in which income varies the most and where expenses related to children are the highest. In the third stage, households should have a smoother income pattern, and while income varies less, housing needs should decrease since in this stage children leave home. Finally, in the fourth stage—heads of household 65 or older—income may be reduced due to retirement. This should be coupled with a reduction in housing demand. The regression results support most of these inferences. According to our results, the function of income elasticity depending on age displays a clear inverse U pattern, with a maximum elasticity found in stage 2, the stage in which income has more variation, decreasing both for the earlier and the later stages. The results also point towards higher elasticities for households in the fourth age cohort than for those in the first one. This does not support the idea of households overconsuming housing in older ages. In a country like Uruguay, this is an important policy issue. According to our results, households are less income elastic in their constitution stage. This is because in this stage, income is also used to access other goods— automobiles, durable goods—and housing may not be the first priority to adapt, once a minimum amount of housing has been provided for. In the next stage, income elasticity increases by 60 percent, coupled with the fact that income increases during this stage, given that this is the most productive age cohort. Housing needs also increase, as this is the stage in which households grow the most. After that, income elasticity decreases, although not as much as expected. This may imply that households continue to reduce their housing consumption in older ages, although clearly not as much as needed, since elasticity decreases for the last two cohorts, where housing needs also decline and income is rather stable. Finally, it also appears that access to financing is most important in terms of increasing housing demand for younger households. The importance of access to financing decreases as households age. In addition, while access through housing programs is not significant for the youngest households, it also has elasticities which in absolute terms decrease with age. This is interesting since it implies that housing programs seem to improve for older age cohorts, in terms of restricting less the housing demand of households, other things being equal. ===== PAGE 33 (score 32, numeric-paren 25, table-words 2, var-words 1) ===== 30 decided to enter the informal market because they had no economic means to stay in the formal market, while another 50 percent says that the trigger was the need to leave their parents’ household (Amarante and Caffera, 2003). In that sense, it could be argued that entering informality can be modeled as a decision in which demographic attributes play an important role. Most of the variables display the expected sign and are highly significant in explaining formality choices. In particular, the older and more educated the head of household, the more likely he/she is to be in the formal market. An increment of ten years in age increases the probability of being in the formal market by 0.01, and escalating one educational category (for example from primary to secondary) increases the likelihood of choosing the formal market by 0.023. The coefficient for the male dummy shows that women are 0.014 more likely to be a part of the formal market. Finally, although years of tenure significantly increases the odds of formality, the size of that effect is not great. Table 7. Formality Choices: Logit Estimation ME for Formal Market/ (se) Age 0.001*** (0.000) Male -0.014*** (0.003) Caucasian (d) 0.021*** (0.004) Education level 0.023*** (0.002) Inc. Retiree -0.032** (0.012) Inc. Informal -0.123*** (0.029) Inc. Private -0.053** (0.018) Inc. Public -0.045** (0.020) Inc. Other 0.001 (0.022) MVD -0.063*** (0.004) With Partner (d) 0.006** (0.003) # children under 14 -0.006*** (0.001) Fin. Access 0.026*** (0.003) ===== PAGE 34 (score 13, numeric-paren 6, table-words 1, var-words 4) ===== 31 Table 7., continued ME for Formal Market/ (se) Transfers 0.005** (0.003) Social assistance (d) -0.022*** (0.004) Years of Tenure 0.000*** (0.000) Pred. Prob. at mean .971 Log Likelihood -2753.259 Pseudo 2 R 0.2667 N 14,193 *p<0.1, **p<0.05, ***p<0.001 ME: Marginal Effects. Marginal effects are calculated for the formal market With respect to sources of income, all categories decrease the odds of being in the formal housing market, when compared to owning a business. As expected, those that are more prone to be in the informal housing market are the ones with informal sources of income. In fact, having an informal source of income increases the odds of informality by 0.123. Surprisingly, private and public salaried workers are more prone to be in the informal market than retirees and pensioners, despite the latter being a group with a smaller average income. This can be reflecting the fact that older people are less prone to enter an irregular housing market. Indeed, most households in irregular settlements are headed by very young people, who see their irregular situation as transitory. Still, as Garabato and Ramada-Sarasola (2011) have shown, for half the population in irregular housing settlements, informality is not transitory. Household composition also has an influence over formality decisions. Every additional child under the age of 14 increases the probability of choosing the informal market by 0.006, and monoparental households are more likely to be found in informal markets. Finally, those households receiving social assistance are 0.022 more likely to be part of informal housing markets, which is expected, as these households are probably in a vulnerable financial situation that hampers their ability to access the formal market. Households receiving income through transfers or remittances are 0.005 more likely to be in the formal market than those who do not, and the impact of having access to external financing sources is significantly positive, increasing the likelihood of formality by 0.026. This ===== PAGE 35 (score 6, numeric-paren 0, table-words 2, var-words 2) ===== 32 again supports the inference that there is a formality premium that households may not be willing or able to pay. 5.2 The Ownership Choice Table 8 shows the marginal effects estimates and standard errors for the two ownership models presented in the previous section. In the case of the formal market, households only have a dichotomous choice to make (to rent or to own) and hence the fitted model is simply a binomial logit regression in which the decision to own is taken as the base case. In the case of the informal market, households can choose to own, rent, or occupy, and therefore a multinomial logit model could be fitted. Still, given the scant number of households renting within the informal market, we decided to pool renting and occupying together in a category that accounts for a less permanent status within the informal market, against ownership, which is inherently a more permanent situation of housing informality. The marginal effects of choosing between a somewhat permanent and a less permanent housing solution within the informal market are presented in Table 10. 5.2.1 Ownership in Formal Housing Markets In the case of the formal market, age has a positive impact on the odds of owning the housing solution, suggesting that older heads of household—even after controlling for income—are more prone to owning a housing solution. This reflects the tendency of households in Uruguay of preferring ownership to renting, and for households with older heads, the need to achieve the goal of owning a home seems to get more pronounced. We also find that men are less likely than women to own than to rent, which can be related to women’s increased need of a more permanent housing solution as a safe haven. In fact, the predicted probability of renting increases by 0.024 if the head of household is a man. With respect to the source of income, public employees are more likely to rent compared to wage earners in the private sector (base case), and the income source most increasing the odds of ownership is owning a business. Given that these results hold even after controlling for income, they may reflect differences in taste, e.g., business owners like more to “own” than others, and in the predictability of income: public sector wage earners have more job security since public employees cannot be terminated in Uruguay. As tenure increases, so does the probability of owning a housing solution (at the mean, the ME of more than 10 years of tenure ===== PAGE 36 (score 30, numeric-paren 24, table-words 1, var-words 1) ===== 33 on ownership is 0.046). This may be related to the fact that the longer the tenure, the more likely it is that the household will access credit in the financial market. Households where the head has a partner and those with small children are more likely to own a housing unit in the formal market than to rent. Indeed, being in a partnership increases the probability of owning the housing solution by 0.087, and each additional child under the age of 14 adds 0.01 to the probability. The explanation for this may be that non-traditional families need more stable housing arrangements. Finally, households in the formal housing market who are receiving social assistance are more prone to own, while those receiving remittances are more likely to rent. Table 8. Ownership Choices: Logit and Multinomial Logit Estimation Formal+ Own ME/se Perm Inc. -0.000 (0.000) Age 0.006*** (0.000) Male (d) -0.024** (0.009) Caucasian (d) 0.014 (0.010) Education 0.003 (0.006) Inc. Retiree 0.015 (0.013) Inc. Informal 0.020* (0.010) Inc. Own business 0.044*** (0.011) Inc. Public -0.038** (0.013) Inc. Other 0.110*** (0.016) With Partner 0.087*** (0.011) Divorced/Widower - - # people <14 0.010** (0.004) # people >14 - - Reg. House. Policy (d) - - illegal electricity (d) - - Transf. (d) -0.062*** (0.011) Social assistance (d) 0.033*** (0.008) ===== PAGE 37 (score 12, numeric-paren 6, table-words 2, var-words 1) ===== 34 Table 8., continued Formal+ Own ME/se Tenure< 1Y (d) - - Tenure 1-5 Y(d) -0.025** (0.011) Tenure 5-10 Y(d) -0.003 (0.012) Tenure>10 Y (d) 0.046*** (0.009) Predicted Prob. at mean 0.841 N 13139 Log Likelihood -5,682.583 Pseudo 2 R 0.096 *p<0.1, **p<0.05, ***p<0.001 ME: Marginal Effects. +As per definition, formal housing arrangements only offer two choices (to rent or to buy). Therefore, the estimated model is a simple logit where owning is the base outcome. Another decision within the formal market can also be assessed. Indeed, it is interesting to find out whether households are ownership-biased and hence prefer to own an average or bad quality housing solution that does not necessarily meet their needs rather than to rent a good quality housing solution. Table 9 depicts the results of that decision. At the mean, the choice with the highest predicted probability within the formal market is to own a good quality housing solution, followed by owning a bad or average quality housing unit. To assess if owning a bad quality solution is preferred over renting a good quality one, all other things being equal, we present the results having the latter as the base outcome. As can be concluded from estimation results, both owning and renting bad quality solutions are less probable than renting a good quality one the higher the household’s permanent income. There is no significant difference stemming from income with respect to the odds of owning vs. renting high quality housing units. Older heads of household are more prone to own than to rent a good quality housing unit, confirming the results found earlier. Given the age structure of Uruguay’s population, this confirms the earlier results regarding a certain bias toward ownership. With respect to gender, the results again seem to confirm that households headed by women are more likely to own than to rent given the same quality. The odds of owning a bad quality solution vs. renting a good quality one are not significantly affected by gender or race. ===== PAGE 38 (score 114, numeric-paren 91, table-words 2, var-words 1) ===== 35 Table 9. Ownership Choices: Multinomial Logit Estimation Estimated Coefficients Marginal Effects Own good quality Own bad quality Rent bad quality Own good quality Own bad quality Rent good quality Rent bad quality Perm. Income -0.000 -0.000*** -0.000*** 0.000 0.000 0.000 0.000 (0.000) (0.000) (0.000) Age 0.052*** 0.063*** 0.026*** 0.003 0.003 -0.006 -0.001 (0.003) (0.003) (0.005) male (d) -0.214*** 0.106 0.282** -0.066 0.041 0.014 0.011 (0.079) (0.091) (0.134) race_w (d) 0.202** 0.092 0.155 0.032 -0.012 -0.021 0.000 (0.085) (0.092) (0.132) education_level 0.026 0.366*** 0.355*** -0.047 0.052 -0.013 0.008 (0.052) (0.072) (0.120) Inc. Retiree 0.077 -0.574*** -0.844*** 0.104 -0.092 0.011 -0.022 (0.111) (0.131) (0.203) Inc. Informal 0.152 -0.928*** -1.200*** 0.147 -0.130 0.008 -0.025 (0.099) (0.131) (0.203) Inc. Own business 0.348*** 1.074*** 0.764*** -0.097 0.141 -0.053 0.008 (0.113) (0.147) (0.243) Inc. Public -0.237** -0.179 0.072 -0.034 0.001 0.025 0.008 (0.100) (0.122) (0.182) Inc. Other 1.199*** 0.163 -1.567** 0.210 -0.104 -0.078 -0.028 (0.358) (0.393) (0.797) With Partner 0.716*** 1.506*** 1.194*** -0.037 0.133 -0.108 0.012 (0.089) (0.111) (0.175) # people <14 0.083** 0.242*** 0.221*** -0.016 0.026 -0.014 0.003 (0.035) (0.037) (0.049) Transf. (d) -0.443*** -0.444*** -0.094 -0.048 -0.015 0.054 0.009 (0.076) (0.087) (0.123) Social assistance (d) 0.099 0.801*** 0.365*** -0.097 0.124 -0.031 0.003 (0.078) (0.086) (0.122) Tenure 1-5 Y(d) -0.176* -0.222** -0.069 -0.012 -0.012 0.021 0.003 (0.093) (0.105) (0.140) Tenure 5-10 Y(d) -0.125 -0.019 -0.257 -0.022 0.014 0.012 -0.005 (0.103) (0.117) (0.168) tenure>10 Y (d) 0.360*** 0.263*** -0.130 0.050 -0.004 -0.035 -0.012 (0.087) (0.098) (0.148) Const. -1.880*** -1.118*** -0.117 (0.210) (0.236) (0.344) Predicted Prob at mean 0.646 0.196 0.128 0.030 N 13,139 Log Likelihood -12,598.675 Pseudo R2 0.1233 *p<0.1, **p<0.05, ***p<0.001 Note: Renting a good quality solution is the base outcome. ===== PAGE 39 (score 5, numeric-paren 0, table-words 1, var-words 3) ===== 36 The more educated the head of household, the more likely he or she will prefer to own a bad/average quality housing unit than to rent a good quality one. For business owners, this pattern is also supported by the results in Table 9, given that the odds of owning a bad quality solution significantly increase over the odds of renting a good quality one for this subsample. Again, non-traditional families, consisting of a stable partner and children—increase the likelihood of preferring to own a worse housing solution than to rent a better one. Finally, less tenure increases the odds of owning. This is to be expected, since tenure should be aligned with an increased savings capacity. As our model showed, households with certain characteristics would rather own a bad quality housing unit that rent a property in better conditions. This preference is probably also a byproduct of the low income stability of many households. Income uncertainty discourages households that depend on salaries from entering the rental market, since they view ownership as an extra safeguard in the event of a negative income shock. 5.2.2 Ownership in Informal Housing Markets Within the informal market, we include new variables to understand the choice of engaging in a somewhat permanent housing solution vs. a less permanent one. In fact, according to anecdotal data gathered through interviews of households in the informal housing market, the motivation to enter the informal market seems to be a strong driver of the type of “ownership” chosen. We hence include four dummy variables accounting for the reasons why the household had to leave its previous housing solution. The four reasons included are those in which the household was to some extent forced to leave, namely eviction, restrictions of space, inhabitability of previous solution due to structural problems, and leaving for economic reasons (negative income shock). ===== PAGE 40 (score 15, numeric-paren 8, table-words 2, var-words 1) ===== 37 Table 10. Ownership Choices: Logit Estimation (Marginal effects presented) Permanent Arrangements (illegally own) Perm Inc. 0.000** Left - Evicted§ -0.184 Left - Econ§ -0.261*** Left - Inhabitable§ -0.204** Left - Space§ -0.165* Age 0.000 Male§ -0.039 White§ -0.059* Education -0.084** Inc. Informal§ 0.016 With Partner§ -0.058 Divorced/Widower§ -0.03 # people <14 -0.005 # people >14 0.029** Reg. House. Policy§ 0.067 illegal electricity§ 0.054* Transf. § -0.015 Social assistance§ 0.02 Tenure< 1Y § 0.036 Predicted Prob at mean 0.639 N 1054 Log Likelihood -661.167 Pseudo R2 0.047 Note: Occupying or renting is the base outcome. § Dummy variables *p<0.1, **p<0.05, ***p<0.001 Having been evicted does not significantly affect the type of arrangement chosen, while those households that were indirectly forced into the new housing solution chosen are all less prone to engage in a more permanent illegal housing situation. Being white and having more years of education also significantly lower the odds of a more permanent illegal housing solution. On the contrary, among the households in the informal housing market, income is positively related to more permanent arrangements. The other two variables increasing the likelihood of more permanent illegal housing arrangements are the number of adults in the household and illegal access to electricity. ===== PAGE 41 (score 6, numeric-paren 0, table-words 0, var-words 6) ===== 38 6. Policy Recommendations In the previous sections we analyzed the Uruguayan housing market and the determinants of formality choices. We found that unlike the situation in other Latin American countries, quantitative housing deficit is not a major problem, while qualitative deficit is. As described in Garabato and Ramada-Sarasola (2011), one-fourth of all housing units suffer from at least one form of qualitative deficit. This figure jumps to almost 60 percent when we consider households in the informal market. Also, after a thorough investigation of current public housing programs, we found that even though the number of programs is large, take-up for most programs is very low, hampering their potential impact on housing relief. In light of these findings, this section discusses possible policy options to tackle these issues. Based on the model estimated in Section 5, we will examine how certain policy measures could impact household formality choices, especially for the more vulnerable strata. 6.1 Policy Options Based on the findings by Garabato and Ramada-Sarasola (2011) there seem to be two key issues that need special attention from a policy perspective. The first relates to the sizeable number of people that are found with informal housing arrangements and the poor housing conditions they share (qualitative deficit). The second issue relates to the low mobility and high inertia found in the market, probably driven by the extremely low development of the mortgage and rental markets. Indeed, almost half of the households stay in the same housing unit for more than 15 years and almost half of these remain in the same unit for more than 30 years, irrespective of changes in housing prices, or in household income, as well as of modified household composition. Moreover, this inertia is not restricted to the formal market, since informal housing arrangements also develop into a rather permanent housing solution, with more than 40 percent of households having lived in the same housing unit for more than 10 years. While this inertia can be motivated by unobservables such as idiosyncrasy and cultural elements, there may be mother reasons for it that could be addressed through proper policy measures. ===== PAGE 58 (score 17, numeric-paren 0, table-words 2, var-words 2) ===== 55 Appendix Appendix A. Variables and Descriptive Statistics Table 12. Descriptive Statistics Variable # of Observations Mean Std. Dev. Min Max hage 21,310 53.861 16.780 17 98 gender 21,310 1.297 0.457 1 2 race_b 21,310 0.071 0.257 0 1 race_other 21,310 0.065 0.246 0 1 race_w 21,310 0.864 0.343 0 1 e_primaria 21,310 0.507 0.500 0 1 e_sec 21,310 0.262 0.440 0 1 e_tec 21,310 0.100 0.300 0 1 e_prof 21,310 0.018 0.135 0 1 e_univer 21,310 0.059 0.235 0 1 e_terciario 21,310 0.011 0.103 0 1 civilstatus_2 21,310 0.108 0.310 0 1 civilstatus_3 21,310 0.529 0.499 0 1 civilstatus_4 21,310 0.153 0.360 0 1 civilstatus_5 21,310 0.210 0.407 0 1 n_peopl_house 21,310 3.025 1.699 1 19 n_peopl_income 21,310 1.898 1.030 0 13 asent 21,310 0.055 0.227 0 1 legal 21,310 0.945 0.227 0 1 rural 21,310 0.199 0.399 0 1 transf 21,310 0.135 0.342 0 1 num_jobs 21,310 0.728 0.611 0 11 one_job 21,310 0.583 0.493 0 1 two_jobs 21,310 0.063 0.243 0 1 multiple_job 21,310 0.006 0.076 0 1 no_job 21,310 0.348 0.476 0 1 aporta 21,310 0.436 0.496 0 1 no_aporta 21,310 0.216 0.412 0 1 rent_recibidas 21,310 0.090 0.286 0 1 cambio_trabajo 21,310 0.015 0.120 0 1 acc_fin 21,310 0.045 0.208 0 1 acc_fin_publico 21,310 0.125 0.330 0 1 ayuda_social 21,310 0.036 0.187 0 1 demanda_housingpolicy 21,310 0.024 0.153 0 1 tenure 21,310 7.893 11.270 0 91 edadvivienda_1 21,310 0.071 0.256 0 1 edadvivienda_2 21,310 0.136 0.343 0 1 edadvivienda_3 21,310 0.163 0.369 0 1 edadvivienda_4 21,310 0.131 0.337 0 1 edadvivienda_5 21,310 0.500 0.500 0 1 num_dorm 21,310 1.917 0.981 1 40 num_ambien s 21,310 3.395 1.652 1 80 num_banos 21,310 1.161 0.776 0 21 sanitation 21,310 0.875 0.331 0 1 tipovivienda_1 21,310 0.834 0.372 0 1 tipovivienda_2 21,310 0.058 0.235 0 1 tipovivienda_3 21,310 0.062 0.242 0 1 tipovivienda_4 21,310 0.041 0.198 0 1 tipovivienda_5 21,310 0.004 0.062 0 1 tipo_construccion 21,312 2.496 0.645 1 3 problemas 21,312 0.656 0.475 0 1 problemas_gravess 21,312 0.573 0.495 0 1 calidad_barrio 21,312 1.543 0.498 1 2 ingtotalhogar 21,310 10,566 13,252 0 335,000 owner 21,312 0.677 0.468 0 1 tenant 21,312 0.128 0.334 0 1 occupant 21,312 0.195 0.396 0 1 ===== PAGE 59 (score 7, numeric-paren 0, table-words 1, var-words 5) ===== 56 Table 13. Description of Variables Variable name Variable label Description hage Age of Household Head Age of household head in years gender Gender Gender of Household head race_b Black Dummy variable that takes the value of 1 when HofH = Black race_other Race Other Dummy variable that takes the value of 1 when HofH = Other race_w White Dummy variable that takes the value of 1 when HofH = White e_primaria Primary Education Dummy variable that takes the value of 1 when education of HofH = Primary e_sec Secondary Education Dummy variable that takes the value of 1 when education of HofH = Secondary e_tec Technical Education Dummy variable that takes the value of 1 when education of HofH = Technical e_prof Teacher Dummy variable that takes the value of 1 when education of HofH = Teacher e_univer University or above Dummy variable that takes the value of 1 when education of HofH = University or above e_terciario Tertiary Education Dummy variable that takes the value of 1 when education of HofH = Tertiary non University civilstatus_2 Divorced Dummy variable that takes the value of 1 when HofH is divorced civilstatus_3 Married Dummy variable that takes the value of 1 when HofH is Married civilstatus_4 Widower Dummy variable that takes the value of 1 when HofH is Widower civilstatus_5 Single Dummy variable that takes the value of 1 when HofH is Single n_peopl_house # of people in household # of people living in same household n_peopl_income # of people perceiving income # of people perceiving income in household asent Illegal housing (irregular settlement) Dummy variable that takes the value of 1 when Household is part of an irregular settlement legal Legal Housing Dummy variable that takes the value of 1 when Household is in a legal housing situation rural Rural Areas Dummy variable that takes the value of 1 when Household is located in a rural area transf Household Receives Transfers Dummy variable that takes the value of 1 when Household receives transfers num_jobs # of jobs of head # of jobs of HofH one_job One Job Dummy variable that takes the value of 1 when HofH has one job two_jobs Two Jobs Dummy variable that takes the value of 1 when HofH has two jobs multiple_job + than 2 jobs Dummy variable that takes the value of 1 when HofH has more than 2 jobs no_job No job Dummy variable that takes the value of 1 when HofH is not employed aporta Contribute to SS Dummy variable that takes the value of 1 when HofH is contributing to SS rent_recibidas Rents perceived by household Dummy variable that takes the value of 1 when Household receives Rents ===== PAGE 64 (score 96, numeric-paren 65, table-words 1, var-words 5) ===== 61 Appendix B. Hedonic Price Regressions Table 18. Hedonic Elasticities 1 2 3 4 5 6 7 Coef Elas t. Coef Elas t. Coef Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Not transformed # Bedrooms 1.062* * 0.21 2 1.065* ** 0.18 2 0.604 0.05 9 0.442* 0.10 3 1.252* ** 0.13 9 0.925** 0.11 6 0.165 ** 0.16 9 # Bathroom s 2.147* ** 0.41 2 1.203* ** 0.15 6 0.762* 0.06 1 0.466* * 0.07 7 3.863* ** 0.33 5 0.511** 0.04 8 0.047 0.03 7 WES 8.101 0.78 7 -0.452 - 0.04 2 -2.102 - 0.12 2 3.535* 0.44 - 6.687* * - 0.43 3 14.109* ** 1.01 9 0.889 ** 0.54 6 Severe Problems -2.607 - 0.25 3 -2.36* - 0.21 7 - 10.108 ** - 0.58 7 - 2.567* * - 0.32 -1.823 - 0.11 8 1.258 0.09 1 - 0.775 ** - 0.47 6 Medium Problems - 3.108* ** - 0.30 2 - 1.168* * - 0.10 7 0.441 0.02 6 - 1.229* * - 0.15 3 0.24 0.01 6 -1.05 - 0.07 6 - 0.299 ** - 0.18 4 Small Problems - 2.473* * - 0.24 - 1.313* * - 0.12 1 -0.03 - 0.00 2 -0.459 - 0.05 7 -0.601 - 0.03 9 -0.706 - 0.05 1 - 0.043 - 0.02 6 Years of constructi on 0.009 0.00 4 - 0.569* * - 0.22 6 - 1.029* * - 0.26 2 -0.281 - 0.15 8 - 0.748* * - 0.21 6 -0.475 - 0.15 8 - 0.061 - 0.17 house_D 1.244* 0.12 1 1.234* * 0.11 4 0.666 0.03 9 1.625* ** 0.20 2 0.138 0.00 9 1.79** 0.12 9 0.308 ** 0.18 9 Constructi on Type -1.37 - 0.39 8 4.262* ** 1.16 3 1.366 0.50 1 12.069 ** 2.34 -1.516 - 0.32 6 - 0.071 - 0.13 Neigh. Serv. 0.24 0.22 3 0.796* ** 0.65 1 0.634* * 0.37 0.336* * 0.35 5 0.401* * 0.25 1 0.345 0.20 9 0.074 ** 0.44 type_neig h r Sup. durable goods 1.411* * 0.13 7 1.277* * 0.11 7 2** 0.11 6 0.228 0.02 8 2.287* ** 0.14 8 2.587** * 0.18 7 0.165 * 0.10 1 Domestic Service 2.872* ** 0.27 9 2.3*** 0.21 2 3.64** 0.21 2 0.979 0.12 2 3.024* ** 0.19 6 3.21*** 0.23 2 0.128 0.07 9 m2 avg. Price 0.000 0.14 0.001* ** 0.62 1 0.000 - 0.02 6 0.000 - 0.07 5 0.000* * 0.16 4 0.000** 0.18 7 0.000 ** 0.50 1 Cons 16.087 -0.438 33.501 12.656 -4.209 11.782 5.37 Transformed Per. Inc. 0.046 0.06 2 0.019 0.02 8 -0.029 - 0.04 7 0.045 0.06 6 -0.036 - 0.05 8 0.073* 0.11 8 0.139 * 0.15 3 Age 0.115 0.03 2 0.63** 0.18 1 1.092* ** 0.24 9 0.268 0.09 4 0.857* ** 0.21 1 0.611** 0.16 1 0.194 * 0.15 1 Educ. 0.398 0.05 3 1.217* * 0.15 2 3.1** 0.25 9 0.593 0.09 4 1.582* * 0.14 4 0.757 0.07 5 0.097 0.06 4 Coef P>|z | Coef P>|z | Coef P>|z | Coef P>|z | Coef P>|z | Coef P>|z | Coef P>|z | λ 0.261 0.00 0 0.282 0.00 0 0.346 0.00 0 0.255 0.00 7 0.329 0.00 0 0.323 0.00 0 0.061 0.27 6 N 443 593 283 244 289 312 233 ===== PAGE 65 (score 80, numeric-paren 61, table-words 1, var-words 5) ===== 62 1 2 3 4 5 6 7 Coef Elas t. Coef Elas t. Coef Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. LR 2 χ 238.14 317.33 92.10 99.00 210.67 156.00 130.5 3 Prob> 2 χ 0.000 0.000 0.000 0.000 0.000 0.000 0.000 λ =-1 0.000 0.000 0.000 0.000 0.000 0.000 0.000 λ =0 0.000 0.000 0.000 0.004 0.000 0.000 0.273 λ =1 0.000 0.000 0.000 0.000 0.000 0.000 0.000 *p<0.1, **p<0.05, ***p<0.001 Table 19. Hedonic Elasticities (continued) 8 9 10 11† 12 13 14 Coef. Elas t. Coef. Ela st. Coef. Ela st. Coef. Ela st. Coef. Ela st. Coef. Ela st. Coef. Ela st. Not transformed # Bedrooms 1.483* * 0.1 03 2.048 0.0 68 -0.034 - 0.0 04 0.124* ** 1.174* ** 0.2 42 4.088* ** 0.1 77 0.104 0.0 35 # Bathroom s 2.594* * 0.1 16 1.983* * 0.0 45 5.897* ** 0.3 67 0.011 0.652* * 0.0 79 0.683 0.0 19 0.285* 0.0 55 WES 9.041* * 0.3 47 20.899 ** 0.3 7 3.104* * 0.1 84 0.304* * 1.922* * 0.2 09 12.218 ** 0.2 96 1.134 0.1 94 Severe Problems - 6.485* - 0.2 49 - 12.802 ** - 0.2 27 - 2.678* - 0.1 59 - 0.238* - 1.198* * - 0.1 3 - 8.904* - 0.2 15 - 1.49** - 0.2 55 Medium Problems - 6.609* ** - 0.2 54 - 8.072* * - 0.1 43 -1.369 - 0.0 81 - 0.2*** - 1.045* * - 0.1 14 - 7.321* ** - 0.1 77 -0.58* - 0.0 99 Small Problems - 2.183* - 0.0 84 -5.056 - 0.0 9 -0.172 - 0.0 1 - 0.095* -0.087 - 0.0 09 -3.151 - 0.0 76 - 0.879* * - 0.1 51 Years of construuc tion 0.547 0.0 93 0.11 0.0 08 -0.403 - 0.0 83 0.001 0.236* 0.1 05 -0.474 - 0.0 49 0.013 0.0 09 House 5.034* ** 0.1 93 7.615* * 0.1 35 2.334* * 0.1 38 0.128* * 1.17** 0.1 27 4.78** 0.1 16 -0.081 - 0.0 14 Construct ion Type 3.945* * 0.4 36 10.989 *** 0.5 43 2.323* * 0.3 57 0.421* ** 1.151* * 0.3 38 6.972* * 0.4 73 2.301* ** 1.0 95 Neigh. Serv. 1.445* * 0.4 38 1.916 0.2 15 1.143* ** 0.3 21 0.044* * 0.594* ** 0.4 01 1.533* * 0.2 79 -0.079 - 0.0 83 Type Neigh. 0.161 0.0 2 -1.909 - 0.1 41 1.189* ** - 0.3 32 - 0.766* ** - 0.4 43 - 3.629* * - 0.4 5 0.543 0.5 67 Sup. durable goods 1.301 0.0 5 9.904* * 0.1 76 0.534 0.0 32 0.167* * 1.085* * 0.1 18 4.888* * 0.1 18 0.446 0.0 76 Domestic Service 3.608* 0.1 39 12.844 * 0.2 28 -0.986 - 0.0 58 0.133 1.302 0.1 42 0.856 0.0 21 1.428 0.2 45 m2 avg. Price -0.006 - 1.4 4 0.003 0.3 28 8.95E- 05 - 0.1 18 - 0.1 1 0.002* * 0.2 31 ===== PAGE 66 (score 66, numeric-paren 46, table-words 1, var-words 1) ===== 63 8 9 10 11† 12 13 14 Coef. Elas t. Coef. Ela st. Coef. Ela st. Coef. Ela st. Coef. Ela st. Coef. Ela st. Coef. Ela st. Cons 41.092 -5.871 19.688 4.331 15.107 13.324 3.668 Transformed Perm. Inc. 0.046 0.0 87 0.109* * 0.2 58 0.141* ** 0.2 64 0.099 0.056* 0.0 92 0.063* * 0.1 39 0.168* ** 0.2 49 Age 0.886* * 0.1 73 0.532 0.0 75 -0.008 - 0.0 02 0.092 0.265 0.0 91 1.486* ** 0.2 41 0.345* 0.1 47 Education 2.536 0.1 42 - 5.761* - 0.1 55 -0.393 - 0.0 3 0.05 0.466 0.0 63 1.165 0.0 42 -0.396 - 0.0 82 Coef P>| z| Coef P>| z| Coef P>| z| Coef P>| z| Coef P>| z| Coef P>| z| Coef P>| z| λ 0.405 0.0 00 0.513 0.0 00 0.366 0.0 00 0.287 0.0 00 0.474 0.0 00 0.228 0.0 00 N 295 256 228 211 460 331 248 LR 2 χ 203.85 181.73 162.27 13.31 324.85 199.92 158.4 Prob> 2 χ 0.000 0.000 0.000 0.000 0.000 0.000 0.000 λ =-1 0.000 0.000 0.000 0.000 0.000 0.000 λ =0 0.000 0.000 0.000 0.000 0.000 0.000 λ =1 0.000 0.000 0.000 0.000 0.000 0.000 *p<0.1, **p<0.05, ***p<0.001. + indicates fitted model is log linear. In these cases, elasticities are estimated coefficients and model significance was tested with and F-test and not a 2 χ distribution Table 20. Hedonic Elasticities (continued) 15 16 17 18 19 20 21† Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Not transformed # Bedroo ms 0.775* * 0.12 4 0.27** 0.12 9 0.441* * 0.09 9 0.356 0.05 8 2.182* * 0.13 1 1.905* 0.10 7 0.067* * # Bathroo ms 1.607* * 0.15 2 0.285* * 0.08 0.192 0.02 5 2.993* ** 0.27 1.049 0.03 5 1.787* 0.05 7 0.251* * WES 1.693* * 0.14 9 0.716* * 0.18 5 1.394* * 0.16 8 2.098* * 0.18 7 5.275* * 0.16 9 2.164 0.06 6 0.254* ** Severe Prob- lems - 1.859* * - 0.16 4 - 1.055* ** - 0.27 3 - 2.542* ** - 0.30 6 - 1.967* * - 0.17 5 - 5.948* * - 0.19 1 - 6.32** - 0.19 3 -0.099 Medium Prob- lems - 1.42** - 0.12 5 - 0.574* * - 0.14 8 - 1.501* ** - 0.18 1 - 1.623* * - 0.14 4 -2.487 - 0.08 - 3.704* - 0.11 3 -0.024 Small Prob- lems -0.573 - 0.05 1 -0.31* - 0.08 - 0.954* - 0.11 5 -0.453 - 0.04 -2.556 - 0.08 2 -1.245 - 0.03 8 0.055 Years of Construc t-ion -0.082 - 0.03 -0.036 - 0.03 7 -0.158 - 0.07 2 0.157 0.04 7 -0.573 - 0.06 2 0.721 0.08 8 0.008 House 0.896* 0.07 0.635* 0.16 1.742* 0.21 1.537* 0.13 5.428* 0.17 9.651* 0.29 0.144 ===== PAGE 67 (score 57, numeric-paren 41, table-words 0, var-words 0) ===== 64 9 * 4 ** * 7 * 4 ** 5 15 16 17 18 19 20 21† Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Construc -tion Type 2.585* ** 0.61 4 0.613* ** 0.41 7 0.819* * 0.24 1.131* * 0.23 6 3.207* ** 0.25 2 6.114* ** 0.48 6 0.214* ** Neigh. Serv. 0.377* 0.19 4 0.202* * 0.24 6 0.426* ** 0.26 8 0.652* * 0.24 8 0.529 0.06 8 2.712* ** 0.44 5 0.065* * Type Neigh. - 0.942* * - 0.51 3 - 0.511* * - 0.80 6 -0.509 - 0.44 2 - 1.684* * - 1.08 6 - 7.093* ** - 1.71 4 - 3.835* - 0.83 8 Sup. durable goods 2.035* ** 0.18 0.455* * 0.11 8 1.489* ** 0.17 9 0.703 0.06 3 3.232* * 0.10 4 5.233* * 0.16 0.131* ** Domesti c Service 1.112 0.09 8 0.114 0.02 9 2.271* 0.27 4 -1.524 - 0.13 6 -5.745 - 0.18 4 24.824 ** 0.75 8 - 0.192* * m2 avg. Price 0.001* * 0.37 4 0 0.08 2 0.001* * 0.25 2 -0.001 - 0.30 6 -0.002 - 0.20 5 0.004 0.38 3 Cons 13.408 10.77 15.885 28.418 85.532 23.232 2.473 Transformed Perm. Inc. 0.037 0.06 2 0.119* * 0.16 2 0.063* 0.10 6 0.095* * 0.17 4 0.072* * 0.17 3 0.051* 0.11 4 0.271* ** Age 0.584* * 0.18 1 0.202* 0.10 6 0.264 0.09 7 0.167 0.05 3 0.594* 0.11 8 0.483 0.09 1 0.154* Educatio n 0.083 0.00 9 -0.007 - 0.00 2 0.417 0.06 1 -0.031 - 0.00 3 1.129 0.04 9 0.602 0.02 5 - 0.145* Coef P>|z | Coef P>|z | Coef P>|z | Coef P>|z | Coef P>|z | Coef P>|z | Coef P>|z | λ 0.315 0.00 0 0.178 0.00 1 0.285 0.00 0 0.325 0.00 0 0.466 0.00 0 0.461 0.00 0 N 527 451 487 441 515 288 472 LR 2 χ 283.91 259.36 270.64 225.69 202.11 193.52 16.17 Prob> 2 χ 0.000 0.000 0.000 0.000 0.000 0.000 0.000 λ =-1 0.000 0.000 0.000 0.000 0.000 0.000 λ =0 0.000 0.001 0.000 0.000 0.000 0.000 λ =1 0.000 0.000 0.000 0.000 0.000 0.000 *p<0.1, **p<0.05, ***p<0.001 + indicates fitted model is log linear. In these cases, elasticities are estimated coefficients and model significance was tested with an F-test and not a 2 χ distribution ===== PAGE 68 (score 83, numeric-paren 62, table-words 1, var-words 6) ===== 65 Table 21. Hedonic Elasticities (continued) 22† 24† 26† 28 29 31† 32 Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Not transformed # Bedrooms 0.075* ** 0.022 0.092* ** 0.889* * 0.13 7 0.116* ** 0.12 0.096 ** 0.411* ** 0.16 7 22† 24† 26† 28 29 31† 32 Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. Coef. Elas t. # Bathroom s 0.277* ** 0.272* * 0.052* ** 0.241 0.02 1 0.107* ** 0.06 4 0.083 * 0.365* ** 0.08 5 WES 0.197* ** 0.297* ** 0.22** * 3.474* ** 0.28 4 0.465* ** 0.24 4 0.174 ** 0.991* ** 0.19 9 Severe Problems - 0.195* ** - 0.285* * - 0.245* ** -0.52 - 0.04 2 - 0.371* ** - 0.19 5 - 0.068 - 1.002* ** - 0.20 1 Medium Problems - 0.089* * -0.071 - 0.081* * -1.066 - 0.08 7 -0.076 - 0.04 0.039 - 0.686* ** - 0.13 8 Small Problems -0.013 -0.12 -0.024 0.004 0 -0.009 - 0.00 4 0.025 - 0.352* * - 0.07 1 Years of Const. 0 0.003 0.022* * 0.361 0.11 8 -0.011 - 0.02 4 - 0.012 0.038 0.02 9 House 0.073 -0.011 0.084* * -4.307 - 0.35 2 0.236* * 0.12 4 0.167 * 1.283* ** 0.25 8 Constructi on Type 0.172* ** 0.177* ** 0.17** * 1.557* * 0.27 0.397* ** 0.53 5 0.101 ** 0.604* ** 0.30 7 Neigh. Serv. 0.065* ** 0.016 0.096* ** -0.19 - 0.06 2 0.149* ** 0.44 2 0.075 ** 0.364* ** 0.43 1 Type Neigh. - 0.066* * -0.009 - 0.028* * - 0.217* ** - 0.25 4 - 0.263* ** - 0.12 7 Luxury goods 0.18** * 0.275* ** 0.167* ** 1.677* * 0.13 7 0.292* ** 0.15 3 0.224 ** 0.776* ** 0.15 6 Domestic Service 0.178* * 0.158 0.332* ** 4.326* * 0.35 4 0.547* ** 0.28 7 0.418 ** 1.801* ** 0.36 2 m2 avg. Price Cons 4.181 5.052 3.562 13.42 6.768 3.982 5.742 ===== PAGE 69 (score 19, numeric-paren 14, table-words 0, var-words 0) ===== 66 Coef. Elas t. Coef. Elast. Coef. Elast . Coef. Elast . Coef. Elast . Coef. Elas t. Coef. El as t. Transformed Perm. Inc. 0.108* * 0.06 5 0.182** * 0.079** 0.16 4 0.04 0.04 5 0.154* * 0.12*** 0.167 Age 0.23** * 0.10 1 0.19*** 0.886** * 0.30 2 0.183* * 0.13 5 0.097 0.494** * 0.23 Educatio n 0.084* 0.04 5 0.048 1.1 0.11 1 0.164* * 0.09 1 0.048 0.561** 0.132 Coef P>|z | Coef P>|z | Coef P>|z | Coef P>|z| Coef P>|z| Coef P>|z | Coef P>|z| λ 0.355 0.00 0 0.085 0.00 0 0.210 0.000 N 925 354 1501 413 882 444 1598 LR 2 χ 68.73 17.3 139.16 140.95 714.8 8.6 1082.32 Prob> 2 χ 0.000 0.00 0 0.000 0.000 0.000 0.000 0.000 λ =-1 0.000 0.000 0.000 λ =0 0.000 0.015 0.000 λ =1 0.000 0.000 0.000 *p<0.1, **p<0.05, ***p<0.001 + indicates fitted model is log linear. In these cases, elasticities are estimated coefficients and model significance was tested with and F-test and not a 2 χ distribution ===== PAGE 70 (score 90, numeric-paren 65, table-words 1, var-words 6) ===== 67 Table 22. Hedonic Elasticities (continued) 34† 35 36 37† 38† 39 Coef. Elast. Coef. Elast. Coef. Elast. Coef. Elast. Coef. Elast. Coef. Elast. Not transformed # Bedrooms 0.044 6.852** 0.111 0.219** 0.141 0.074*** 0.058** 0.449 0.084 # Bathrooms 0.339*** 24.946** 0.215 0.514*** 0.193 0.126*** 0.093*** 2.208** 0.245 WES 0.247* 9.928 0.082 0.499** 0.167 0.279*** 0.186** 1.913** 0.196 Severe Problems -0.171 -5.351 - 0.044 - 1.203*** - 0.402 - 0.271*** -0.152* - 2.666** - 0.273 Medium Problems - 0.171** - 20.104** - 0.166 - 0.342** - 0.114 - 0.127*** -0.072 -1.24** - 0.127 Small Problems 0.042 - 17.392** - 0.144 -0.262 - 0.088 - 0.117** -0.045 -0.835 - 0.085 Years of const. 0 4.796** 0.146 0.021 0.024 0.015 0.016 -0.359* - 0.149 House 0.217 1.718 0.014 -0.063 - 0.021 0.249*** -0.035 3.468*** 0.354 Const. Type 0.195** 17.097*** 0.358 0.636*** 0.5 0.104*** 0.168*** 1.559*** 0.388 Neigh. Serv. 0.037 6.963*** 0.248 0.003 0.004 0.039*** 0.064*** 1.074*** 0.549 Type Neigh. - 0.148*** -5.019** -0.1 0.005 - 0.035** - 0.034** -0.285 Sup. durable goods 0.11 12.301** 0.102 0.637*** 0.213 0.136*** 0.143*** 0.877* 0.09 Domestic Service 0.259** -3.014 - 0.025 0.192 0.064 0.14** 0.347*** 2.877** 0.294 m2 avg. Price Cons 4.876 10.882 6.102 4.171 5.408 6.102 Transformed Perm. Inc. 0.235** 0.042** 0.113 0.171*** 0.22 0.171*** 0.084* 0.029 0.048 Age -0.096 1.691** 0.172 0.143 0.086 0.17** 0.062 0.841** 0.29 Education 0.054 9.246* 0.118 0.671** 0.246 0.068 0.268*** 0.978 0.123 Coef P>|z| Coef P>|z| Coef P>|z| Coef P>|z| Coef P>|z| Coef P>|z| λ 0.632 0.000 0.147 0.000 0.303 0.000 N 228 484 719 680 728 376 LR 2 χ 25.15 223.17 323.73 41.25 31.3 224.37 Prob> 2 χ 0.000 0.000 0.000 0.000 0.000 0.000 λ =-1 0.000 0.000 0.000 λ =0 0.000 0.000 0.000 λ =1 0.000 0.000 0.000 *p<0.1, **p<0.05, ***p<0.001 + indicates fitted model is log linear. In these cases, elasticities are estimated coefficients and model significance was tested with and F-test and not a 2 χ distribution