- DRAFT- Housing conditions and income distribution: evidences from São Paulo, Brazil

- DRAFT- Housing conditions and income distribution: evidences from São Paulo, Brazil by Emilio Haddad, PhD Faculdade de Arquitetura e Urbanismo, Universidade de São Paulo, Brazil email: emhaddad@usp.br and João Fernando Pires Meyer Doctoral Candidate Faculdade de Arquitetura e Urbanismo, Universidade de São Paulo, Brazil email: joaomeyer1@yahoo.com.br To be presented at 2007 Urban Research Symposium The World Bank Washington, DC May 14-16, 2007

HOUSING CONDITIONS AND INCOME DISTRIBUTION : EVIDENCES FROM SAO PAULO, BRAZIL Emílio Haddad Professor Doutor Faculdade de Arquitetura e Urbanismo Universidade de São Paulo, Brazil emhaddad@usp.br and João Fernando Pires Meyer Doctoral Candidate Faculdade de Arquitetura e Urbanismo Universidade de São Paulo, Brazil joaomeyer1@yahoo.com.br Summary: The paper presents an empirical study of the characteristics of new residential buildings produced by market in the city of Sao Paulo, Brazil, in the past 20 years. The basic data for the research comes from a monthly survey made by EMBRAESP, a local Real Estate consultancy enterprise. The collected information has been, over the recent years, a major reference in studies about Sao Paulo housing market studies. EMBRAESP survey collects rigorous data of all housing projects, including its location, price, size, number of units, etc. This information has permitted to estimate quantity and prices and related them with family income. Information on housing developer was also provided. It has provided details about the way formal housing market production serves the need of middle and high income families. It also has shown some characteristics of the housing industry, shedding interest with market aspects that seemed to have been more neglected and that would inspire further empirical developments and comparative studies Key Words: housing market, Sao Paulo

HOUSING CONDITIONS AND INCOME DISTRIBUTION: EVIDENCES FROM SAO PAULO, BRAZIL Emílio Haddad, Professor Doutor João Fernando Pires Meyer, Doctoral Candidate Faculdade de Arquitetura e Urbanismo Universidade de São Paulo, Brazil I. INTRODUCTION This idea of developing this paper came out as a response to the Call for Papers for this IV Urban Research Symposium. As a participant of a previous Symposium, held in Brasilia, in 2005, one of the authors, Emilio Haddad, had the opportunity of getting better acquainted with the spectrum of subject matters and approaches that were brought into the discussion and has found opportune the subject proposed. The format of this paper has thus resulted from its origins, and was determined by the following facts: a) The importance of the theme Housing condition has been a major problem in the metropolis in Developing countries. A distinctive feature of this has been the increasing presence of an informal sector, revealing the failure of the formal market. b) The search for new approaches Much of the literature on housing in the large cities of less developed countries has dwelled in the informal sector. This paper focus on the other side of the coin : the formal market, its capabilities and limitations in providing housing. Formal housing markets have been somehow neglected in the literature, following the lesser prominence of the concept of Empowerment that pervaded the literature in the beginning of the 90 s (World Bank, 1993). c) The availability of data Essential for the initiative of proposing this paper for the Urban Symposium was our operational knowledge of a database of new housing been produced for the São Paulo market. Collected and organized by Empresa Brasileira de Estudos do Patrimônio EMBRAESP, a private local consultancy in Real Estate and valuation, this database has documented housing starts in Sao Paulo over almost 30 years.

Systematic information on housing starts and its main characteristics are not common in cities of developing countries and, from the best of our knowledge, the use of such kind of information in housing research has not been made so far. Thus, to introduce and explore this database has become one objective of this study. d) Research content This paper has therefore differences in nature from those that report results of a research or policy application that had been already done. It has been a research work and this paper reports major findings; there are still others to come as the analysis progresses. e) International audience A final concern in writing this paper was with its international audience, which has reflected in the choice of phrasing and of the information presented, that the authors considered necessary in order to provide some revealing background. Yet, the selection aimed at detecting specific characteristics of the formal housing market in São Paulo, which from the best of our knowledge might differ from many other cities. f) Results An election of descriptive data analysis was made and this paper reports them. This information refers to: a) the volume of production; b) sales price c) affordability d) Developers in the market and their perspective. One of the purposes of items a) and b) were also that of provide information that could instigate the audience to do their own comparisons with other cities, and thus building a personal understanding of this market. Item c) is intended to show an interesting view of affordability by plotting in the same graphic a comparative the income distribution in town by income brackets and the distribution of the price of housing starts affordable to each one of these brackets. This figure was made possible by the use of income distribution figures from the Census, and housing prices of the large databank from EMBRAESP. Finally, the item d) has also the purpose of showing some basic characteristics of housing entrepreneurs in Sao Paulo, facilitating comparisons with other cities of the world. Figures to be presented show that housing development in Sao Paulo has been a competitive market with a large number of small producers, and that a process of oligopolization is underway.

After sketching the evolution their production behavior over the past 10 years is shown, and interpreted. Most of housing policy making needs an understand the decision making process of the developer, a central player in housing, and to devise instruments to stimulate them to better respond to socioambiental needs of a given society. This paper explores, in the case of the city of Sao Paulo, Brazil, potentialities and limits of the private sector, as one of the major actors in housing development process, in the provision of housing for low-income families. It is possible to identify the cheapest units offered by the market, its characteristics and location, and trace its evolution of the past two decades, comparing with income-related indicators. And verify the impact of income redistribution on those limits. For this analysis, a view of the market as a whole is informative. II. METHODOLOGY: A NOTE Methodologically, this research somehow does not follow the more traditional scientific approach by which a research question is initially posed and data is then collected to verify to what extent the hypotheses could be verified. In our case, data were already existent. It has been a research situation that has increasingly occurred the mushrooming of digital databases, and one of the growing fields in Statistics is Data Mining a set of concepts and techniques of finding knowledge from pre-existent raw data. III. SOURCE OF INFORMATION III. 1 Housing The basic data for the research comes from a monthly survey made by EMPRESA BRASILEIRA DE ESTUDOS DO PATRIMONIO EMBRAESP 1, a local Real Estate consultancy enterprise. The collected information has been, over the recent years, a major reference in studies about Sao Paulo housing market studies. A project is inserted in this database when the sales are launched, what might occur before, concomitantly or after the beginning of construction. One caveat must be made: it did look adequate for the purpose of the present study based in a large series of data -- to consider a good proxy for what in the econometric studies are considered to be housing start. EMBRAESP survey collects rigorous data of all housing starts, which includes: 1 EMBRAESP home page is www.embraesp.com.br

- name of the project - location (address) - homogeneous market zone - number of units - number of floors (if apartment) - number of elevators - area of the land developed - end of construction date and for each residential unit: - private and total area - number of bedrooms - number of bathrooms - price (in R$, and in U$D equivalent) - form of payment - price per square meter and name and address of the developer name and address of the building contractor name and address of the Real Estate broker name and address of the financial institution (whenever there is mortgage finance) other information available Land price, if not stated, could be indirectly estimated. In those cases, the price refers to the land component after the development had been made, and not the buying price of raw land before the development. III. 2 Income Quality of the housing stock is probably the most conspicuous evidence of income distribution in a given society. Brazil has been known as being one most unequal. As a small sample of the country, São Paulo keeps similar pattern. There are two major sources of information about incomes in the city of São Paulo and its distribution: a) IBGE the Brazilian Census Bureau, and b) The Origin Destination Survey.

Census only collects info on the head of household and for the whole household, which is not very much useful as indicators because in São Paulo, due to cohabitation, the number of families is 20 % higher than the number of households. Because it provides better information on household income, the resulting O/D Surveys, were chosen to be presented. Table presents this information respectively for the years 1997 and 2002. These data refer to the whole Metropolitan Region. Figure 1: Grater São Paulo Metropolitan Region, distribution of population by income 1997 2002 Monthly income per family (Reais of Oct 02) Obs: adjustment for inflation was made using INPC - Índice Nacional de Preços ao Consumidor, do IBGE - Instituto Brasileiro de Geografia e Estatistica. (Source: City of Sao Paulo. O/D survey 2 Incidentally, the figure shows that between 1997 and 2002, income (measured in minimum wages) distribution in Sao Paulo had become more unequal. A book by Davies (2006) documents similar occurrence in different countries of the world, those were under a process of globalization. The housing consequences are also portrayed generating what he denominates: a planet of slums. 2 http://www.metro.sp.gov.br/empresa/pesquisas/afericao_da_pesquisa/afericao_da_pesquisa.shtml)

IV. EMPIRICAL RESULTS IV.1 Volume of Production In order to better situate the participation of the formal market provision of housing, a few background data about Sao Paulo must be given. According to the official Census, the municipality of São Paulo, core of the Greater São Paulo Metropolitan Region, had a population of 10.434.252, in 2000, and of 9.527.426, in 1991; the respective number of house units were 2.985.977, in 2000 and 2.539.953, in 1991. Table 1 and Figure 2 display the size of the formal housing production in São Paulo, averaging 21,562 units/year, during the period of 1977 / 2006. Table 1: New residential starts (1977-2006) Year # of housing projects # of housing units 1977 279 13828 1978 180 12988 1979 210 12781 1980 261 18539 1981 380 32294 1982 425 32554 1983 217 9900 1984 268 11378 1985 333 12186 1986 667 31790 1987 247 16377 1988 308 23455 1989 359 18437 1990 217 14520 1991 157 12780 1992 152 10434 1993 311 21663 1994 376 24868 1995 406 26567 1996 367 30990 1997 337 39426 1998 267 21667 1999 266 26358 2000 370 29666 2001 336 23785

2002 388 22051 2003 413 26547 2004 408 22550 2005 333 24915 Total 9238 625294 Average (1977-2005) 319 21562 Source:EMBRAESP Figure 2: São Paulo city: Residential units starts (1965-2006) 45 40 35 30 31,790 39,426 31,053 1000 units 25 20 15 10 5 0 10,434 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Source:EMBRAESP A first comparison could be now made: it is reported that between the years 1991 and 2000, the number of housing units has grown 446,024 units. On the other hand housing starts in the period totalized 231,639. Therefore, it represented 51.9 % of all housing increment in this period. The figure shows very well the cyclic nature of housing starts, which can be explained by the evolution of the macroeconomy. Yet, changes in housing starts have been used as one first indicator of whether there will be recessions and/or expansion in a given economy. Some explanations could be associated to the behavior of housing starts. As examples:

a) The peak of production in the year of 1986 was a result of the successful implementation of the stabilization policy, named Plano Cruzado, which brought together some positive effects in income distribution. b) The figure also shows a more stable rhythm of housing starts since the year of 2000, which to a large extent reflects an overall economic stability in the period. Table 2.: Total built areas and land used (in thousands of square meters) Year Total built área Total built private área Total used land Average floor/area ratio 2001 3452 1897 905 3,81 2002 4075 2198 1018 4,00 2003 4434 2407 1084 4,09 2004 4293 2362 1088 3,95 2005 4792 2627 1002 4,78 2001-2005 21046 11491 5097 4,13 Source: EMBRAESP This table show that in this period the municipality of Sao Paulo increased its built residential area by 20 million square meters, while, in the same period, and its population growth rate has fallen to 0.58 % a year. Total population is estimated (by SEADE The Sao Paulo State Statistical office) to have increased from 10,426,384 in 2000 to 10,744,060, in 2005. Thus, in a 5 years spam, 21,046,000 square meters of residential space were built while the population went up 317,676 people. It means, an average of 66 square meters per new inhabitant which is more than enough to shelter the population increase in the same period 3. The actual picture has shown, however, a worsening of housing conditions, in the same period, with increase in the number of favela dwellers. The major explanation for this apparent contradiction lies in the income distribution: a few numbers of families do buy very large units. IV.2 Prices Table gives some information of average prices (in U$ D) of the new housing in the market. 3 It has been reported that in PR of China, the standard for popular housing is 9 square meters/person.

Table 3.: Housing starts -- Price per square meters by number of bedrooms (R$/square meter) # of bedrooms 1 2 3 4 2001 R$ 2.866,00 R$ 1.371,00 R$ 1.648,00 R$ 2.909,00 2002 R$ 2.743,00 R$ 1.760,00 R$ 1.982,00 R$ 2.917,00 2003 R$ 3.264,00 R$ 2.153,00 R$ 2.241,00 R$ 3.265,00 2004 R$ 3.597,00 R$ 2.241,00 R$ 2.423,00 R$ 3.814,00 2005 R$ 3.027,00 R$ 2.233,00 R$ 2.598,00 R$ 3.770,00 Source: EMBRAESP Table shows the sale price per sq m of new residential units. One can observe that 2 bedroom units have the smallest unity price, and that this price increases; in the case of 1 bedroom apartments, higher unitary prices are due to: a) the production refers to a specific type of building know apart-hotels that means buildings with provision of home services, and b) the more expensive costs refers to the wet areas (kitchen and bathrooms) that usually have the same The conclusion is that unitary sales price increases with the size of the unity. This perverse combination of numbers this is another expression of the skewness of income distribution in São Paulo. Figure 4 shows the average price of residential units (in U$) by size (number of bedrooms) Figure 4: average price of residential units (in U$) by number of bedrooms

300000 250000 200000 150000 100000 50000 0 1 2 3 4 Source:EMBRAESP The table also provides price information that allow comparisons with other major cities in the world, basically, U$ 57,306 for an one bedroom flat with services; U$ 47,524 for a popular 2 bedroom unity; U$77,735 for a 3 bedroom, and U$ 281,670 for a 4 bedroom. It seems unnecessary to point out that, for the purpose of making international comparisons, these figures might be misleading, as the dollar has had abrupt fluctuations in the given period (1996/2005) (See table below), and also because these numbers indicate average values of the new units 4. Table 5.: Average Exchange rate (1 U$D/R$) 5 Average Exchange rate Year (USD/R$) 1 996 1,01 1 997 1,08 1 998 1,17 1 999 1,83 2 000 1,84 2 001 2,37 2 002 2,92 2 003 3,02 2 004 2,93 4 An alternative approach to the use of average dollar values in international comparisons is the use of the PPP (Purchasing Power Parity) indicator produced by OCDE. See: Akiyama, Yuko (2002) The World Land Survey: an introduction of a methodology for comparing Real Estate values 5 At the moment of the writing of this paper (May 2007) exchange rate is 1 U$D = R$ 2,02.

Source: Banco Central do Brasil. Average estimated by EMBRAESP Finally, data from EMBRAESP allows listing the extremes: the more expensive and largest housing starts at one side, and the cheapest and smallest, on the other side. To illustrate, a cheapest is a unit with 65 square meters, for sale for R$ 52.000 (in real of 2007), approx. U$ 25.000. An a most expensive is a unit with 1200 sq meters, for R$ 8.100.000,00, approx. U$ 4.000.000,00, meaning a difference in selling price of approximately 60 times. IV.3 Issues of affordability The availability of both information on income the major component of demand and supply, coming out of EMBRAESP database, allow us to build an interesting comparison. For the purpose of comparing supply and demand, calculations were made in two stages: a) first an estimation of housing affordability, i.e. calculating the distribution of the prices of housing starts by income brackets; b) then, overlapping the obtained distribution curve, with that of income distribution. Studies of housing affordability are usually made in the basis of its economic dimension, i.e., the relationship between price and income. In a more common -- situation where there is long term financing, the relationship between monthly disposable income and the value of monthly installment payment. An affordability table (Table 6) was built up using loan conditions from Caixa Economica Federal 6 -- a large financial institution from the Brazilian federal government that operates housing credits. A series of Table 6.: Affordability of housing starts Unit starts (3) Family's income brackets (minimum wages) Families (2) 0,00% 0 5 48,60% 5,76% 5 10 24,17% 17,77% 10 15 8,42% 9,20% 15 20 5,63% 8,69% 20 25 2,49% 4,44% 25 30 2,57% 4,30% 30 35 1,76% 5,41% 35 40 1,42% 6 Caixa Economica webpage is www.caixa.org.br

9,57% 40 45 0,63% 34,84% > 45 4,31% 100,00% 100,00% (3) Units starts in the income bracket (Embraesp and CEF loan conditions worked by the authors) (1) Familiar mensal income bracket in minimum salaries (husband & wife) Minimum wage = U$D 160,00 (2) Families in the mensal income bracket - 2000 Census microdata sample - worked by the authors IV.4 Supply and demand Figure 5 displays the resulting comparison between housing demand and what the formal market has supplied in São Paulo. To built this graphic a large number of data were used and for that purpose hypothesis were made concerning index of price adjustment for inflation, correction of Census microdata in order to account for income from all members of the family. Figure 5.: Housing production and affordability - São Paulo City - 2006 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% Famílies (2) Units starts (3) 0% 0 5 5 10 10 15 15 20 20 25 25 30 30 35 35 40 40 45 > 45 Family's income brackets (1) (minimum wages) Source: Elaborated by the authors with data from EMBRAESP, Caixa Economica Federal and Census microdata, from IBGE

The resulting figure show very clearly how much displacement exists between both supply and demand: while most families are in the lower income brackets, formal market production is geared to upper and middle class. Figure 5 shows that the formal market does not produce for families with income smaller than 5 minimum wages, which constitutes almost half of the population; it also shows, on the other extreme that 35 % of housing starts are produced for families with monthly incomes larger than 45 minimum wages, which correspond to less than 5 % of the families. It also indicates that 5 % the formal housing market is affordable to families within the income bracket between 5 and 10 minimum wages. Yet, in this is segment lays a major focus of housing policies, and a close look of the projects will provide a better factual knowledge to be used in policy making 7. IV.5 Housing finance Aside of demography and family income, a third element that determines housing demand is the availability of long term financing, which is able to bring affordability. Yet, as much as the value of monthly payment of mortgages could be brought down, more families enter into the solvable set, following the assumption that it has to be a percentage of disposable income. The implementation of a new Real Estate finance system, in the beginning of the present decade has allowed the development of REITs as well as a system of securitization of receivables, as it happened in the United States, in the 90 s (Miles, Berens, Weiss, 2000), and other countries 8. As it is traditional, these financial resources have been mostly geared towards commercial investment as in Brazil there is virtually no investment in rental housing. Yet, returns have been negatively impacted by policies of residential rent control. Nevertheless, there have been evidences in recent developments that the market is going down to reach middle class demand, which had been a major focus in the beginning of the housing finance system (Sistema Financeiro de Habitacao), created in 1964. This a resulting impact of the synergic effect of decrease in interest rates due to more price stability and more availability of housing finance. IV.6 The developer s perspective 7 The study of those projects is a logical recommendation of the present study. 8 The example of Mexico (See Duhau, forthcoming) has drawn the attention from Brazilian housing authorities and academics.

Housing development in São Paulo has not been a much oligopolized market, as compared to other cities, particularly those in the Developed world. For example, in the year of 2005, the database reports 335 different companies responsible for housing starts in the São Paulo Metropolitan Region; of those, the ten largest are responsible for 22,64 % of the market share. One can notice however a movement towards concentration as, in the year of 2000, there were 360 companies, and the ten largest had a market share of 20,55 %, in the direction to what does happen in other countries. Figure 6, below compares the evolution of quantity and prices of housing starts in the Grater São Paulo Metropolitan Region. Figure 6: Greater São Paulo Metropolitan Region: Residential Unit Starts & Average Total Price (1995-2005) 70000 300 60000 250 Units 50000 40000 30000 20000 Units Total price 200 150 100 Total price (1.000 R$) 10000 50 0 1 995 1 996 1 997 1 998 1 999 2 000 2 001 2 002 2 003 2 004 0 Source:EMBRAESP In this table, it is remarkable the opposite direction of these indicators, showing that high income segment operates as a shelter for investment whenever there is a decrease in demand, i.e., middle income housing is perceived an investment with more risk.

The private sector has consistently produced a small number of very large residential units, averaging more than 800 square meters, and virtually no units affordable to the population in the lowest income brackets that had to find a shelter solution living in substandard housing conditions, in informal settlements. As this outcome should have been consistent with the logic of profit-maximization behavior of housing producers, in order to deepen the study, it is suggested that the use of some knowledge of Real Estate Investment analysis is needed. IV.7 Location aspects of housing Over the years, housing market in São Paulo has been associated with important transformations in the occupation of the city space. Yet, city planning is an element that should not be dissociated from housing studies. As the EMBRAESP database has information on the address of housing starts, it allows the use of spatial analytical capabilities of Geographic Information System s software especially the possibility of integrating of different spatial databases. Haddad (2005) has studied to what extent Real Estate investment decisions do anticipate neighborhood changes, in São Paulo. V. CONCLUSIONS AND NOTES FOR FUTURE RESEARCH This paper was prepared for the Urban Research Symposium 2007, and had as an objective to show a case study of using of housing starts data in market studies. In the case of São Paulo this information has permitted to estimate quantity and prices and related them with family income. Information on housing developer was also provided. It has provided details about the way formal housing market production serves the need of middle and high income families. It also has shown some characteristics of the housing industry, shedding interest with market aspects that seemed to have been more neglected and that would inspire further empirical developments and comparative studies. The study resulted from a first descriptive analysis using information from a databank on housing starts in Sao Paulo. Further analyses are to follow, and in the development of the paper a few suggestions for future research have already been made: a) to focus on the projects affordable to low income families; b) to include in the study the feasibility analysis of housing investment decisions; c) to compare with the present market situation that has seen recent economic and institutional changes VI. BIBLIOGRAPHY

Akiyama, Y. (2002) The World Land Survey: an introduction of a methodology for comparing Real Estate values (available at: http://www.lares.org.br/sl2_akiyama.pdf, last accessed in April 7, 2007) Davies, M (2006) Planet of slums, Verso, London. Duhau, E and Uribe, G. (forthcoming) Los nuevos productores del espacio habitable. Breve historia de una mercancía posible,to be presented at the XXX Encuentro de la Red Nacional de Investigación Urbana, Toluca, Mexico, October Haddad, E (2005) O Mercado imobiliário antecipa as alterações da estrutura dos bairros? Evidências no caso de São Paulo. Paper presented at the V Meeting of LARES Latin American Real Estate Society (available at: http://www.lares.org.br/haddad.pdf, last accessed in May 4, 2007) Metrô SP Aferição da Pesquisa Origem e Destino na Região Metopolitana (available at http://www.metro.sp.gov.br/empresa/pesquisas/afericao_da_pesquisa/afericao_da_pesqui sa.shtml, last accessed in May 4, 2007) Miles, M., Behrens and Weiss (2000) Real Estate Development: Principles and Process. Third Ed., Washington, D. C.: ULI the Urban Land Institute. The World Bank (1993) Housing Enabling Markets to Work, A World Bank Policy Paper, Washington DC: The World Bank, VII. ACKNOWLEGMENT The authors would like to thank Mr. Luis Paulo Pompeia, Executive Director of Empresa Brasileira de Estudos do Patrimônio - EMBRAESP, for his comments and permission to use information from their databank.